(圖說：層層疊疊都是關鍵，我不入地獄，誰入地獄。就如這放空吹風的，空風火腿，犧牲自己，造福眾生。信任無法一次到位，需要層層疊疊，一層又一層累積疊上去，第一層，就從按個讚開始吧。圖片來源：Ernest。)
✳️ 無聊的管線工程
企業 AI 部署真正卡住的地方，幾乎都不是模型不夠強。Mistral AI 的 CTO Timothée Lacroix 說得直接：現在的模型能力已經足夠解鎖大量企業價值，但你得先把連接器、資料格式、權限管理這些「無聊的管線工程」全部做好，企業端的 token 消耗量才會真正爆發。他用了一個詞：「plumbing」（管線系統，大家都只看到管線，我覺得有必要強調系統）。是說，他在整個訪談裡講了三次 plumbing。他說，我們還處在建設階段，大多數企業連基本的資料連接都還沒做好，更別提讓 AI agent 在背景大規模執行任務了。（我們看到的現場更悲催，許願跟資料對應不上，難怪大家喜歡直上文字接龍（內建強酸模式）。）
那企業該怎麼開始？先建信任，再談自主。Lacroix 被問到 agent 的自主性時，反問了一個更根本的問題：「比起問 AI agent 多自主，更好的問題是你多信任它。」他們跟航運公司合作自動化貨櫃放行流程時的做法很具體：AI 自動完成所有資料查核和後端驗證，收集散落在不同系統裡的資訊，但每個貨櫃價值極高、不得出差錯，最終決定權留給港口人員。這是信任階梯的第一階：先讓人看到 AI 做了什麼、為什麼這樣做，確認 AI 的判斷可被追蹤、可被驗證，信任才有基礎往上疊。等信任累積到一定程度，才能逐步放手讓 agent 在背景執行更多任務。
我們在 PAFERS 和 Kyklosify 運用土炮工作法以及逆向工作法與客戶一起解題時（aka 求神問卜），邏輯相似。從客戶的具體痛點往回推：需要什麼產品、什麼流程、什麼系統、怎麼讓利害關係人對齊、怎麼簡化理解和消化的成本、如何微調組織文化。不能先選工具再找問題，而要先確認問題的根源及本質，在大桌子或大白板上玩桌遊模擬資料流動，再一起設計解法。我們選擇盡可能與組織決策者直接合作，因為迭代週期中最常卡關的，是決策，而不是技術（是裡子，而不是腦子）。信任無法一次到位，需要層層疊疊，一層又一層累積疊上去。
把 context（上下文、情境）工程化，是下一個關鍵瓶頸。Lacroix 提到他們內部叫 context engine 的概念：agent 在探索企業資料的過程中發現的知識，例如哪些資料表存在、欄位怎麼 join、什麼權限可以存取，這些都應該被儲存下來重複使用，不是每次讓 agent 從零開始重新探索。他說：「這些運算成本應該被攤提。」想像一下：一個 agent 花了五次 API 呼叫加三次 join 才找到某筆資料，這個路徑應該被記住，下次直接取用。現在則是每次都要重新走一遍同樣的探索路徑，這不合理。（更可怕的是又殊途又不同歸。）
這呼應了我一直對自己耳提面命的「Coding is Easy, Context is Hard」。工程瓶頸解開之後，產品思考、與人類溝通、上下文情境的瓶頸才正要開始。程式碼產出速度暴增，但真正的瓶頸不只是寫程式，還包含了上下文的管理和傳遞。誰能把組織的隱性知識系統化、落實化、參數化，誰就掌握了 AI 時代的護城河（暫時吧 ?）（也可能不是護城河，只是先擋一下的空城記，記得派出掃地僧）。我們前幾年默默打造 Kyklosify Business Suite 的時候，身為 AWS 鐵粉，仿照 Bezos 2002 年的 API Mandate，陪著客戶一起將客戶組織的每一段工作流程拆解成可被 API 呼叫的行為動作，確保每筆操作都有機會留下歷史紀錄，這些糧草先行，為後續的 Agent 和 Digital Twins 預先鋪路。這些前期投入確保了流程知識能被系統化保存，被其他 agent 或人類重複使用，不需要每次從頭來過。這些糧草基底讓我們手上的客戶們得以服務 80%+ 回流客戶、得以投資打造創新服務並創下營收年增 +200%（較去年為 3 倍）等實績。
自建基礎設施是務實選擇，不是奢侈（但沒有餘裕之前請先不要輕易嘗試）（務實有很多種選擇）。Mistral 自建資料中心不是歐洲主權的政治宣示，是因為用別人的環境跑數千張 GPU 的訓練任務時穩定性達不到要求。他們的核心價值主張就是 control：軟體堆疊部署後歸客戶所有，模型修改權也歸客戶。整個堆疊是模組化的，客戶可以選擇只用模型、只用平台、或者完整託管，每一層都可以自己決定。Lacroix 甚至說，即使 AGI 出現，銀行也不會讓它控制一切，基礎設施的治理能力必須跟上模型的進步。使用 AWS 超過 18 年，我理解這選擇背後的邏輯：重點不是自建或上雲的二選一，而是在每個環節確保「當我需要改變時，我有選擇權」。
如果你正在組織裡推動 AI 落地，Lacroix 的經驗指向一個明確的起點：盤點組織中重複性高、規則明確的流程，從那裡開始建信任階梯。不要從最酷的案例開始，是從最無聊但最能被驗證的地方開始。
比較無聊的地方，比較能藏得住價值。
✳️ 延伸閱讀
- Coding Is Just Typing，然後呢？從 Explicit Programming 到 Implicit Programming
- 從 Vibe Coding 到 Agentic Coding：把話講清楚才是新戰場
- 2025 年度回顧：慢下來，才能更快
- AWS Summit Hong Kong 2025 Dev Lounge: 重塑程式設計: AI 如何轉變企業的程式開發方式
- Amazon Bedrock AgentCore 正式上線 (GA)：為 AI Agents 量產環境打造企業級基礎設施
✳️ 知識圖譜
(更多關於知識圖譜…)
✳️ 逐字稿
企業需求與控制
- I think the expectation is that demand and amount of tokens generated for the enterprise will completely jump once you are not bound anymore by humans asking questions or reading them.
我認為一旦不再受限於必須由人類來提問或閱讀，企業端的需求以及生成的 Token 數量將會迎來爆發式增長。 - As soon as you have enough trust to have agents running in the background, you’re not really limited by the number of tokens.
一旦你有足夠的信任讓 agent 在背景執行，你就不再受限於 token 數量了。 - The term we use is control.
我們使用的詞是「控制」。 - The software stack once deployed is in the hands of our customers.
軟體堆疊一旦部署完成，就掌握在我們客戶的手中。 - They own the model changes that we make.
他們擁有我們所做的模型變更。 - And I think it’s really important as a customer to consider that your expertise and what makes your company valuable stays yours.
我認為身為客戶，很重要的是你的專業知識和讓你公司有價值的東西，都會留在你手上。
Podcast 開場介紹
- » Hi, I’m Matt Turk.
» 嗨，我是 Matt Turk。 - Welcome back to the Mad Podcast.
歡迎回到 Mad Podcast。 - Today we have a special episode with Timote Lacro, the CTO and co-founder of Mistrol, the company that proved that you could build frontier models with a fraction of the compute of the US giants.
今天我們有一集特別節目，來賓是 Mistral 的 CTO 暨共同創辦人 Timothee Lacroix，這家公司證明了你可以用美國巨頭們算力的一小部分來打造前沿模型。 - But recently, Mistrol has quietly evolved into a much more ambitious full stack industrial power, building not just the models, but the platform, the deployment stack, and their own massive supercomputing clusters.
但最近，Mistral 已經悄悄演變成一個更有野心的全方位工業力量，不只打造模型，還有平台、部署堆疊，以及他們自己的大型超級運算叢集。 - We covered a lot of ground in this one, the engineering behind Mistral 3, what sovereign AI actually means in practice, and Tim’s contrarian view on why trust matters more than autonomy for agents.
我們在這集涵蓋了很多內容：Mistral 3 背後的工程、主權 AI 在實務上到底意味著什麼，以及 Tim 對於為什麼信任比 agent 的自主性更重要的反直覺觀點。 - If you’re tired of the AI hype, Tim is refreshingly nononsense.
如果你對 AI 的炒作感到厭倦，Tim 的務實風格令人耳目一新。 - Please enjoy this great conversation with Timote LRA.
請享受這場與 Timothee Lacroix 的精彩對話。 - » Hey Timote, welcome.
» 嘿 Timothee，歡迎。 - » Hey.
» 嘿。
轉向全方位解決方案
- » So, as I was prepping for this, I was struck by how much has been going on at Mistrol over the last few months.
» 在我準備這次訪談的時候，我對 Mistral 這幾個月的動作之多感到驚訝。 - I think most people probably know Mistrol as a provider of open-source models.
我想大多數人可能認識 Mistral 是作為開源模型的提供者。 - It seems that you guys evolved from an AI lab to more of a full stack solution focused on enterprise and sovereign customers.
看起來你們已經從一個 AI 實驗室演變成更偏向全方位解決方案，專注在企業和主權客戶上。 - So just to set it up, in the last year you guys raised a€ 1.7 billion euros series C led by ASML at an 11.7 billion post money valuation.
先做個背景介紹，過去一年你們完成了由 ASML 領投的 17 億歐元 C 輪募資，投後估值達 117 億。 - you launch a bunch of models which we’re going to talk about is the big vision behind all of this that enterprises and sovereign states are going to need their own AI infrastructure and MR is going to be the provider.
你們推出了一系列模型，我們待會會聊到。這一切背後的大願景是企業和主權國家將需要自己的 AI 基礎設施，而 Mistral 將會是提供者。 - » Uh so the big vision has been evolving and as you stated we started uh as a company that built models uh because with Arthur and Guom this was what we knew how to do at the start.
» 大願景一直在演進，如你所說，我們一開始是一家打造模型的公司，因為跟 Arthur 和 Guillaume 在一起，這是我們一開始知道怎麼做的事。 - The premise on which we built Misual AI was immediately solving for enterprise needs uh and we started with open weights model.
我們建立 Mistral AI 的前提是直接解決企業需求，我們從開放權重模型開始。 - After this uh and working with enterprise we realized uh the need for basically the rest of the stack.
在這之後，與企業合作的過程中，我們意識到基本上需要堆疊的其餘部分。 - So we built uh the serving platform because infrastructure was needed.
所以我們建立了服務平台，因為基礎設施是必要的。 - Um and then all of the tooling around it uh was also something that we saw was missing.
然後圍繞它的所有工具，也是我們發現缺少的東西。 - more than the tooling, it also requires a lot of work and expertise still to get deep into uh an enterprise workflows and really help that transformation.
不僅是工具，要深入企業工作流程並真正幫助轉型，仍然需要大量的工作和專業知識。 - And so we built that uh FDE function and more recently uh with MSR compute uh we’re going a bit lower uh in the stack as well.
所以我們建立了 FDE 功能，最近透過 Mistral Compute，我們也在堆疊的更底層發展。 - So we’ve done all of this uh because it was required for enterprise success uh while still continuing uh on our models journey.
我們做了這一切，是因為企業成功需要它，同時我們也持續在模型的旅程上前進。 - All of this stack uh being modular is really important to us as it gives full control to u enterprise and our clients as to which part of the stack they decide to uh own and control which is maybe more involved or that they decide to have serverless or basically this modularity that we like.
這整個堆疊的模組化對我們非常重要，因為它讓企業和客戶可以完全掌控他們決定擁有和控制堆疊的哪個部分，可能是更深度參與的部分，或是他們選擇用 serverless 的部分，基本上就是我們喜歡的這種模組化。 - » All right.
» 好的。
自建資料中心
- So let’s take some of those modular components uh in in order.
那我們就按順序來看看這些模組化組件吧。 - Let’s start with mistral compute.
先從 Mistral Compute 開始。 - So that was a big announcement uh I guess in June of 2025 putting a big partnership with Nvidia to um help with this effort.
那是一個大公告，我記得是 2025 年 6 月，與 Nvidia 建立了一個大型合作來推動這件事。 - Uh what’s the current status?
目前的狀態是什麼？ - Is that live yet?
已經上線了嗎？ - Are you building it?
還在建造中嗎？ - You know, how does one go about building data centers or or leveraging data centers in Europe?
你知道的，在歐洲要怎麼著手建造或利用資料中心？ - » Maybe first to go into the reasons uh why we decided to start building our own data centers.
» 也許先來談談我們決定開始建造自己資料中心的原因。 - uh we tried uh a lot of different partners over the years and we realized that our use uh of the AI compute for large scale training was not necessarily well understood by a lot of providers and our uh need for stability especially like when you run inference on a few GPUs or when you run small scale trainings on a hundreds of GPUs margin for error is a lot larger than when you run trainings on uh thousands of GPUs at the same time.
這些年我們嘗試了很多不同的合作夥伴，發現我們對大規模訓練的 AI 算力使用，並不一定被很多供應商充分理解。我們對穩定性的需求特別高，因為當你在幾顆 GPU 上跑推論或在幾百顆 GPU 上跑小規模訓練時，容錯空間比你同時在數千顆 GPU 上跑訓練要大得多。 - And so to address this need for stability, we saw a way for us to basically build our own data centers and maintain it with our understanding of what quality looks like.
為了滿足這種穩定性的需求，我們看到了一個方法，就是自己建造資料中心，並用我們對品質標準的理解來維護它。 - And so that was why we uh launched MRL compute.
這就是我們推出 Mistral Compute 的原因。 - And when we decided to do it, we also realized well maybe others will benefit from it.
當我們決定這麼做的時候，也意識到其他人可能也會從中受益。 - We launched into uh a bigger uh basically development than what was previously intended.
我們啟動了一個比原先計畫更大規模的開發。 - And so this was announced in June as you said since then the building of the facility has progressed quite well.
正如你所說，這在六月宣布了，從那之後設施的建設進展得相當順利。 - It’s in the south of Paris and we are right now running through the stabilization uh stabilization of the first trench.
它在巴黎南部，我們目前正在進行第一批次的穩定化作業。 - Uh so it’s uh quite a large data center so delivery doesn’t happen in one day.
這是一個相當大的資料中心，所以交付不是一天內能完成的。 - And the first part of this data center is something that we are working on as we speak.
資料中心的第一部分正是我們現在正在進行的工作。 - We have a few jobs running and we’re fine-tuning uh basically all of the last uh things uh to run at speed and with the right stability.
我們有幾個任務在跑，正在對最後的細節進行微調，讓它能以正確的速度和穩定性運作。
算力即服務
- » Okay, great.
» 好的，很好。 - And uh did I understand correctly, it’s going to be for your customers and your own needs uh around training, but also you’ll be providing it as a service to others uh in in Europe and beyond.
我的理解正確嗎？這將用於你們的客戶和自身的訓練需求，同時也會作為服務提供給歐洲及其他地區的人。 - » Yeah, exactly.
» 對，完全正確。 - So we will use part of that capacity for ourselves as one of our training clusters but we will also provide a managed Kubernetes and managed serum stack on top.
我們會把部分算力用作自己的訓練叢集，但也會在上面提供託管的 Kubernetes 和託管的服務堆疊。 - » Okay.
» 好的。
建造資料中心的經驗教訓
- Any uh lessons learned so far?
到目前為止有什麼經驗教訓嗎？ - I mean as you said you guys come from a very deep background in in AI and AI research.
我的意思是，如你所說，你們有非常深厚的 AI 和 AI 研究背景。 - It’s a whole different thing to build a whole like data center facility.
但建造一個資料中心設施是完全不同的事情。 - How have you gone about it and uh what are some things that that surprised you and any lessons so far?
你們是怎麼進行的？有什麼事情讓你意外？到目前為止有什麼心得？ - As most uh new experiences as a founder, I relied on the knowledge of others.
就像創辦人的大多數新經歷一樣，我依賴他人的知識。 - Uh and so I was uh lucky to have a very a few seasoned uh HPC experts uh and and a lot of uh cloud software experts as well to build that solution.
我很幸運有幾位經驗豐富的 HPC 專家以及許多雲端軟體專家一起來打造這個解決方案。 - For me personally, and it’s one of the things I love about uh my position at Mistral is that I get to uh discover so many new things uh and so many new problems I hadn’t thought possible.
就我個人來說，這也是我喜歡在 Mistral 這個位置的原因之一，我能發現這麼多新事物和以前想都沒想過的新問題。 - having to learn to like all of the different parts of building a data center, all of the different trades that you have to coordinate, uh all of the potential um synchronization uh between all of the different trades.
必須學習建造資料中心的各個不同部分、需要協調的各種不同工種，以及所有工種之間潛在的同步問題。 - I mean, it’s a huge building.
我是說，這是一棟巨大的建築。 - It involves hundreds of people working on it.
有數百人參與其中。 - You have this then when you uh stand up the thing, uh you have to question what works.
然後當你把東西架起來時，你必須質疑什麼是可行的。 - You have to filter through the blades that are faulty.
你得篩選出有故障的刀鋒伺服器。 - It’s just an entire new area of work where I get to see um experts in their field go through things and try to explain to me what their daily work is.
這是一個全新的工作領域，我能看到各領域的專家處理事情，並試著向我解釋他們的日常工作。 - It’s always fascinating to see um an expert in his field like do something that you don’t know how to do.
看到一個領域的專家做你不知道怎麼做的事情，總是很有意思。 - I think the logistics of it uh and the timelines are also quite different from what I’m usually um dealing with in software and research.
我覺得物流規劃和時程也跟我通常在軟體和研究中處理的很不一樣。 - for new capacity to uh be built, you have to plan around uh having energy available, you have to plan for the uh space to be available and on time.
要建造新的算力，你必須規劃能源供應，必須規劃場地的可用性和時程。 - And so it’s a lot more long-term planning than a few software features.
所以這比開發幾個軟體功能需要更多的長期規劃。
歐洲的能源考量
- » How do you guys go about power since you mentioned energy?
» 既然你提到了能源，你們是怎麼處理電力問題的？ - » In what we’ve been doing in Europe so far hasn’t been a huge blocker, although uh there is constraint.
» 在我們目前在歐洲做的事情中，這還不是一個很大的阻礙，雖然確實有限制。 - Uh I think the grid in various parts of Europe is not uh necessarily easily extensible.
我認為歐洲各地的電網不一定容易擴展。 - I know it’s uh an issue in in France.
我知道這在法國是一個問題。 - A lot of the sites are uh contended.
很多場地都有競爭。 - Um so we we’ll see how it all develops.
所以我們看看事情如何發展。 - We are lucky in Europe to have uh very uh clean and affordable energy uh either with uh green energy in the Nordics and nuclear in France.
我們在歐洲很幸運擁有非常乾淨且負擔得起的能源，北歐有綠色能源，法國有核能。 - So it’s it’s been relatively okay for us today.
所以對我們來說目前還算可以。 - as you describe this uh what comes to mind is the gigantic amounts of money that are being invested in the US around uh data centers.
你這樣描述的時候，讓人想到美國在資料中心方面投入的巨額資金。
與大型科技公司競爭
- How do you guys uh go about that from a financing standpoint and and perhaps even more taking a step back if you think about the race between the big AI labs globally whether that’s you know the opensis and anthropic of the world and and XAI uh it seems that all of them are affiliated with a gigantic pocket of money somewhere obviously there’s Gemini and Google to add to the list and and Meta I’m just curious like how where do you guys stand on on that you have a bunch of partnerships with
從融資的角度來看你們是怎麼應對的？更進一步想，全球大型 AI 實驗室之間的競賽，不管是 OpenAI、Anthropic 還是 xAI，似乎他們都跟某個巨大的資金庫有關聯，當然還有 Gemini 和 Google，還有 Meta。我很好奇你們在這方面處於什麼位置，你們有一些合作夥伴關係， - um SAP and Nvidia but there is you don’t have one of those gigantic companies on your cap table.
包括 SAP 和 Nvidia，但你們的股東名單上沒有那些巨型公司。 - So how do you how do you think about uh competing in that general context?
所以你怎麼看在這個大環境下的競爭？ - So with those uh companies, so the hypers scalers, it’s um there are two parts to the game and we’ve played the partnership uh part quite well with them and we’re integrated within Google’s Verex uh Amazon Bedrock and uh Asia Studio and that is uh the choice that we’ve made in term of having access to uh gigantic pockets of monies.
跟那些公司，就是那些超大規模雲端廠商，這場遊戲有兩個部分，我們跟他們的合作關係經營得相當好，我們已經整合進了 Google 的 Vertex、Amazon Bedrock 和 Azure AI Studio，這是我們在取得巨額資金方面所做的選擇。 - We’ve been focused on efficiency from the start.
我們從一開始就專注於效率。 - Uh and I think we’ve done quite well at building uh models that are uh competitive with the uh investments that we’ve uh put in.
我認為我們在用投入的資源打造具有競爭力的模型方面做得很不錯。 - For us, it’s important to build uh the company as efficiently as we can.
對我們來說，盡可能高效地建設公司是很重要的。 - uh and I deeply believe that with the capabilities that we have today in the models there is so much to be unlocked uh in enterprise that um I I don’t think my main focus uh today would be into going into the gigawatts of power we still need to build uh so much with our clients and unlock so much values with theap capacities that we have » all right so let’s go into uh the enterprise reality of all of this um so if I’m an enterprise or if I’m a sovereign and I want to deploy a MR
我深信以我們今天模型的能力，在企業端還有太多價值可以釋放，我覺得我目前的重點不會放在追求千兆瓦的電力，我們還需要跟客戶一起建設很多東西，用現有的算力釋放很多價值 » 好的，那我們來看看這一切在企業端的實際情況，如果我是一家企業或主權國家，我想要部署一個 Mistral
企業部署與客製化
- open-source model what is it that I do these days with everything that you that you’ve built the way we work with um enterprise I mean as you mentioned like we have a few of our models that are open source and Apache and all of our clients are welcome to use them uh as they need what we have seen in terms of success is that given the current stack It still requires um a lot of expertise uh to manage to come to um actual value and um and things that go to production.
開源模型，以你們建立的一切，這些天我該怎麼做？我們與企業合作的方式，正如你提到的，我們有幾個模型是開源的，以 Apache 授權，我們所有的客戶都可以按需使用。但我們看到的成功經驗是，以現有的堆疊，仍然需要大量的專業知識才能達到實際的價值和上線到生產環境。 - Basically the way we interact is that we usually stand up our um Misual AI studio which is our platform and we can deploy uh all of our stack on the client’s choice uh of deployment methods.
基本上我們互動的方式是，我們通常會架設 Mistral AI Studio，這是我們的平台，我們可以根據客戶選擇的部署方式來部署我們的整個堆疊。 - So it can be on prem uh it can be on their VPC it can be on uh in several places.
可以是地端部署，可以在他們的 VPC 上，也可以在好幾個不同的地方。 - The reason we do this is that it lets uh clients build where their data is uh and without having to shuffle things around which as I’ve learned as a CTO is something that you don’t want to do ever because it asks it raises a lot of questions uh and it’s uh quite a stressful thing to do.
我們這樣做的原因是讓客戶可以在資料所在的地方建構，不需要搬移資料。作為 CTO 我學到的是，這是你永遠不想做的事，因為它會引發很多問題，而且壓力很大。 - So once this is deployed uh we then uh work with the business units to understand where their pain points are.
一旦部署完成，我們就會與業務部門合作，了解他們的痛點在哪裡。 - Sometimes it’s knowledge management and I think it’s the most well-known uh use case from the output from the outside of the enterprise world but it’s also around um automating core workflows for the enterprise.
有時候是知識管理，我認為這是從企業世界外部來看最知名的使用場景，但也涉及到為企業自動化核心工作流程。 - Um it’s you know some tooling that you wouldn’t expect where one thing that we’ve done is around code modernization uh where you you turn a bunch of Excel sheets into an actual like Python app.
有些工具是你意想不到的，我們做的其中一件事是程式碼現代化，把一堆 Excel 表格轉換成真正的 Python 應用程式。 - Uh and if you have many many of those sheets then potentially you want to use AI for this.
如果你有很多很多這樣的表格，那你可能就會想用 AI 來做這件事。 - So once the infrastructure is built then we basically look for what’s the most valuable to the customer and we start acrewing value uh inside a stack of AI assets that then accelerates all of the other developments with that customer » and is part of the idea is that you do actual model work at the customer and for the customers in particular fine-tuning.
一旦基礎設施建好，我們基本上就在尋找對客戶最有價值的東西，開始在 AI 資產堆疊中累積價值，進而加速該客戶的所有其他開發 » 這個想法的一部分是你們在客戶端為客戶做實際的模型工作，特別是 fine-tuning。 - » Yes, we we customize in various ways.
» 是的，我們用各種方式進行客製化。 - Uh so we have done continued pre-training and this is most useful when you want to uh change the capabilities of a model uh more deeply.
我們做過持續預訓練（continued pre-training），這在你想更深層地改變模型能力時最有用。 - So we’ve done this to sometimes change uh the mix of languages in a model to get something that’s a lot better at thou east Asian uh languages for example or you could have require this if your internal data uh which doesn’t happen on the public web is something that’s so new uh that you need a large amount of to of tokens uh to get a model that understands it and becomes fluent with it.
我們有時會這樣做來改變模型中的語言組合，得到在東亞語言方面好很多的模型。或者如果你的內部資料不在公開網路上、而且非常新，你需要大量 token 才能讓模型理解並流暢地使用它。 - So we do uh these kinds of continued pre-training fine-tuning.
所以我們做這類持續預訓練和 fine-tuning。 - We also um like and this is more for an efficiency reason.
我們也會做，這更多是出於效率的原因。 - When you get to smaller models, you have to make trade-offs.
當你用更小的模型時，你必須做取捨。 - Uh the models won’t be as good in their knowledge of the world.
模型在世界知識方面不會那麼好。 - And so when you lose uh a lot of things, you have to focus on what you really care about.
所以當你失去很多東西時，你必須聚焦在你真正關心的事情上。 - And so this is typically important if you want really uh fast, really cheap uh models that will be really good at a specific task.
所以如果你想要非常快、非常便宜的模型，在特定任務上表現很好，這通常是很重要的。 - It’s also useful if you want models that run on the edge uh that get very very tiny.
如果你想要能在邊緣執行、變得非常非常小的模型，這也很有用。 - Uh and so for all of these fine-tuning is a tool of choice.
所以對所有這些需求來說，fine-tuning 是首選工具。 - Another uh reason to do fine-tuning.
另一個做 fine-tuning 的理由。 - It can be to adapt to uh data that’s not necessarily massive but that’s also not available on the web.
可以用來適應不一定很龐大但也不在網路上的資料。 - So typically in coding uh what happens is that you will have massive code bases sometimes acrewed over decades uh that the model will need to be able to uh work with in terms of uh having like vibe uh deployed on it typically and so being able to come in not move the code base and uh learn an actual coding agent for that codebase is really powerful as well.
通常在寫程式方面，你會有累積了數十年的龐大程式碼庫，模型需要能在上面工作，像是部署 vibe coding 工具，所以能直接進入而不搬動程式碼庫，為那個程式碼庫訓練一個真正的 coding agent，也是非常強大的。 - » And who does the all of this you have evolved towards an FDA model.
» 這些都是誰來做的？你們已經演進到一個 FDE 模式了。 - So we have indeed a large uh FD section.
我們確實有一個很大的 FDE 部門。 - It’s it’s a mix of software and uh FDEEs and we split our FDES into what we called um AI engineers and applied scientists.
這是軟體和 FDE 的混合，我們把 FDE 分成所謂的 AI 工程師和應用科學家。 - Um and so uh applied scientists will tend to use the tools that we’ve just uh uh talked about.
應用科學家傾向於使用我們剛才談到的那些工具。 - So fine-tuning, continued pre-training and the likes where AI engineers will focus more on adaptation to the enterprise environment and figuring out what workflows to automate and all of this.
像是 fine-tuning、持續預訓練等等，而 AI 工程師則更專注於適配企業環境，以及找出要自動化哪些工作流程。 - They work with the customers to make sure uh that the use cases are indeed providing values and going to production.
他們與客戶合作，確保使用案例確實在提供價值並進入生產環境。 - But it’s also a fantastic way for us to understand what matters in an enterprise context and be faster at building the right platform.
但這對我們來說也是一個絕佳的方式，來了解在企業環境中什麼最重要，以及更快地建立正確的平台。 - And uh again those customers are the kind of customer for whom customization and privacy is essential.
而且這些客戶正是那種客製化和隱私至關重要的客戶。 - Uh how do you how do you position again open of the world that are going very hard at the enterprise?
面對其他同樣猛攻企業市場的公司，你們怎麼定位自己？ - Is that data sovereignty?
是靠資料主權嗎？ - Is that customization?
還是靠客製化？
定義企業控制
- » The term we use is control.
» 我們用的詞是控制。 - The value that we see is both in our expertise and the software stack that we provide.
我們看到的價值在於我們的專業知識和我們提供的軟體堆疊。 - The software stack once deployed uh is in the hands of our customers and they can change it, they can add to it.
軟體堆疊一旦部署，就在我們客戶的手中，他們可以修改它、可以擴充它。 - They own model changes that we make and I think it’s really important as a customer to And so in working with us and building uh because it takes effort uh to build an AI advantage uh today and so having this effort built into uh something that you own is I think a choice that makes sense.
他們擁有我們所做的模型變更，我認為作為客戶這真的很重要。跟我們合作和建構，因為今天建立 AI 優勢需要努力，把這些努力建構在你擁有的東西上，我認為是一個合理的選擇。
Agent 作為基礎構建模組
- » Let’s talk about uh agents uh obviously part of the overall effort at Mistrol.
» 讓我們來聊聊 agent，這顯然是 Mistral 整體工作的一部分。 - How does that work?
那是怎麼運作的？ - Uh how do you uh build an agent and uh what key use cases have you seen so far?
你怎麼建構一個 agent？到目前為止你看到哪些關鍵使用場景？ - Personally, I think I’ve moved uh from agents to uh workflows, which is I guess an abstraction uh on top.
就我個人來說，我已經從 agent 轉向了 workflow，我想這是上面的一層抽象。 - Um so agents are I think the building blocks uh where you have a given expected input, a set of tools and you are trying to reach a uh set of uh you have a goal that you want to reach.
Agent 是基礎構建模組，你有一個預期的輸入、一組工具，然後你有一個想要達成的目標。 - The set of inputs uh that we’ve enabled are um images, text uh and audio.
我們啟用的輸入集包括影像、文字和語音。 - When you build an agent, to me it’s really important that you build it on a focused uh task with a data set that you understand and that you can iterate on and that you can improve.
建構 agent 時，對我來說很重要的是你要把它建在一個聚焦的任務上，用你理解的資料集，你可以迭代和改進。 - What we see in enterprise is rarely things that are solved with agents because that’s not necessarily where you would expect uh an FDE to be most useful.
我們在企業中看到的，很少是單獨用 agent 就能解決的事情，因為那不一定是你期望 FDE 最有用的地方。 - Those ideally would be built uh on our platform by the customers directly.
理想情況下，那些會由客戶直接在我們的平台上建構。 - Where there is more values value is in uh more complex workflows where you will have several uh agents interact through a workflow to automate something slightly more complex.
更大的價值在於更複雜的 workflow，讓多個 agent 透過 workflow 互動，自動化更複雜的事情。 - And so that’s what we’ve been focusing on.
所以這就是我們一直在聚焦的方向。
自動化複雜工作流程：CMA CGM 案例
- What would be an example?
可以舉個例子嗎？ - » An example is something that we’ve built uh with the shipping company CMACGM where we’ve automated the uh container release process.
» 一個例子是我們跟航運公司 CMA CGM 合作建構的，我們自動化了貨櫃放行流程。 - Uh and so it’s um a use case where I I don’t know how familiar you are with shipping.
這是一個使用場景，我不知道你對航運有多熟悉。 - I wasn’t at first.
我一開始也不熟。 - Uh but a container reaches a port and you have to uh harbor uh probably in English.
但一個貨櫃到達港口，你必須在港口處理它。 - some decision has to be made that this uh container is ready for release to the uh next person on on the line to handle this container and so there are lots of uh checks uh that need to be um run and data to be accessed in the back end uh before that decision is made.
必須做出決定這個貨櫃是否準備好放行給流水線上的下一個人來處理，所以在做出決定之前，需要執行很多檢查和存取後端資料。 - So as you can imagine, some of those containers are extremely valuable and you can’t really afford a mistake.
你可以想像，有些貨櫃非常有價值，你真的承擔不起任何錯誤。 - And so what we’ve done in this case is an application that’s integrated into um how these uh harbor worker work and it automates a lot of the manual work that they did to check the data and they make the final decision uh given all of the evidence.
所以我們在這個案例中做的是一個整合進港口工人工作方式的應用程式，它自動化了他們過去手動檢查資料的大量工作，然後他們根據所有的證據做出最終決定。 - » Okay, this is super interesting.
» 好的，這真的很有趣。
Agent 的信任與治理
- Obviously the the key question about agents these days especially when they are combined into workflows is the question of uh autonomy.
顯然，現在關於 agent 的關鍵問題，特別是當它們組合成 workflow 時，就是自主性的問題。 - How do you guys think about it?
你們怎麼看這件事？ - How autonomous are those agents uh in in your deployments?
在你們的部署中，這些 agent 有多自主？ - » I don’t know if it’s the way I think about it.
» 我不知道這是不是我思考這件事的方式。 - To me the better question usually is how much you trust uh the agents and there are a few dimensions uh around this.
對我來說，更好的問題通常是你有多信任 agent，這有幾個面向。 - What worries me when building those kind of workflows is that typically if you want the value to acrue and if you want to build faster and faster the more workflows that you build, what you will want to do is uh reuse assets and make them reusable by others.
在建構這類 workflow 時讓我擔心的是，如果你想要價值累積，而且建構越多 workflow 就建得越快，你會想要重用資產並讓其他人也能重用。 - Uh as soon as you do this with agents, you then start to ask the question, well this agent has access to some data that is privileged uh but maybe this other agent uh is publishing it to something that’s public.
當你用 agent 這樣做的時候，你就會開始問，這個 agent 有權限存取某些特權資料，但也許另一個 agent 正把它發布到公開的地方。 - You might have governance concerns where uh some agent is acting on something very critical and you don’t know necessarily that the data that it got uh has been approved or something like this.
你可能會有治理方面的擔憂，某個 agent 正在處理非常關鍵的事情，你不一定知道它拿到的資料是否已經被核准。 - It’s really a new way to develop where uh the parts of your workflows have to be trusted.
這真的是一種新的開發方式，你的 workflow 的每個部分都必須是可信賴的。 - Each of them to be trusted requires uh quite a lot of tooling uh and quite a lot of observability uh to get confidence and to basically enable this at scale in an enterprise.
要讓每個部分都可信賴，需要相當多的工具和相當多的可觀測性，才能建立信心，並在企業中大規模啟用。 - So the question that you’re asking about autonomy to me this is something that I see happening when I vibe code.
所以你問的自主性問題，對我來說這是我在 vibe code 的時候看到會發生的事。 - Sure like longer running tasks and making and improving on this is going to be critical and we’re uh working on it daily.
當然，像是更長時間執行的任務，改善這方面會是關鍵的，我們每天都在做。 - But today, the problems that we’re solving on the software side of things are really about how you trust what you’ve built and how you improve it.
但今天，我們在軟體方面解決的問題，真正是關於你如何信任你建構的東西，以及如何改進它。 - Uh, and how you allow an entire company to build on it with confidence.
以及你如何讓整個公司都能有信心地在上面建構。
Studio 元件與版本控制
- » Maybe describe some of the things that you guys have built in studio around governance as you mentioned and trackability and uh registry all the things.
» 也許可以描述一下你們在 Studio 中圍繞你提到的治理、可追蹤性和 registry 所建構的一些東西。 - What what are the key components of an a modern agent suite?
一個現代 agent 套件的關鍵元件是什麼？ - So workflows as I mentioned is something that uh we’ve worked a lot on uh with our customers and it’s not GA yet.
正如我提到的，Workflow 是我們跟客戶大量合作的東西，目前還沒有 GA。 - Uh so look out for this uh sometimes in the future but it’s also one of the benefits uh of working with enterprise we can um have a lot of design partners and once we’re confident uh with the solution uh we we make it G. So a workflow solution is critical.
所以未來請關注，但這也是跟企業合作的好處之一，我們可以有很多設計合作夥伴，一旦我們對解決方案有信心，我們就讓它 GA。所以 workflow 解決方案是關鍵的。 - Workflows are built on various uh model capabilities.
Workflow 建構在各種模型能力之上。 - So u vision, audio and text and reasoning.
包括視覺、語音、文字和推理。 - It is important to uh have a registry of uh connectors and MCPS.
有一個連接器和 MCP 的 registry 是很重要的。 - Uh and so for this we have uh our connections.
為此我們有自己的 connections。 - The observability is an area where we’re still working on.
可觀測性是我們仍在努力的領域。 - Um it’s important for me to be able to iterate and really define uh precisely what an agent does and control each of its goal uh and see how it’s progressing um being able to maintain evaluations and uh build build on them.
對我來說，能夠迭代並精確定義 agent 做什麼、控制它的每個目標、看它如何進展、能夠維護評估並在此基礎上建構，這很重要。 - What is um difficult in this entire sea of complexity is that you also have to maintain proper versioning and tagging and think about how you’re going to deploy and improve uh upon what you’ve built.
在這一整片複雜性之海中困難的是，你還必須維護適當的版本控制和標籤，並思考你要如何部署和改進你所建構的東西。 - So let’s say you’ve built a kickass workflow based on a lot of agents and models that Mrol has released in the past.
假設你基於 Mistral 過去發布的很多 agent 和模型，建構了一個很棒的 workflow。 - Then a few months pass and there are new sets of models that are out.
幾個月過去了，有新的模型推出了。 - Maybe you can simplify that workflow.
也許你可以簡化那個 workflow。 - Maybe the next uh mist 4 is good enough that you can factor out a few agents.
也許下一代 Mistral 4 已經夠好，你可以移除幾個 agent。 - Basically, what you need to be able to do is create a new agent, run it on the same set of inputs and outputs and control that you haven’t broken anything and then deploy it in the wild.
基本上你需要能做的是建立一個新的 agent，在相同的輸入和輸出上執行它，確認你沒有破壞任何東西，然後部署到正式環境。
Context Graph 與企業上下文
- All of this software uh suite basically which has been built for software development over years I feel it isn’t there yet uh in the AI world and that’s what we’re building » as I’m sure you’ve seen there was uh for the last few weeks in startup and venture circles there’s been this whole idea of the context graph as an infrastructure that made the rounds.
這整套基本上是為軟體開發建構多年的軟體套件，我覺得在 AI 世界中還沒到位，而這就是我們在建構的 » 我相信你看到了，過去幾週在新創和創投圈，context graph 作為基礎設施的整個概念一直在流傳。 - Is that something that you think about or a layer that would basically uh enable one to know how the agents made a decision and how those decision relate to one another?
這是你們在思考的東西嗎？或是一個能讓人知道 agent 如何做出決策以及這些決策之間如何關聯的層？ - » I’ve seen this indeed and I think there are two uh levels to that discussion.
» 我確實看到了，我認為這個討論有兩個層次。 - the part that you mentioned at the end where uh it’s interesting to know how an agent came to so in that discussion when when we talk about understanding how an agent came to a decision or an action the game is really to understand how a human uh agent really made this decision.
你最後提到的部分，了解 agent 如何得出結論是很有趣的，在那個討論中，當我們談到理解 agent 如何做出決策或行動時，真正的關鍵是理解人類代理如何做出這個決策。 - It’s understanding how an enterprise does what it does and it’s certainly interesting.
要理解一個企業如何做它所做的事情，這確實很有趣。 - uh what keeps me up at night and what I really want to solve first is just the basic idea of gathering a workable enterprise context.
讓我夜不能寐、我真正想先解決的，只是收集可用的企業上下文這個基本概念。 - Right now uh with uh any model uh and with a lot of effort you will be able to get some connections to tools and you will ask a questions and your agent will do a bunch of things.
現在用任何模型，加上很多努力，你可以建立一些工具連接，你問一個問題，你的 agent 就會做一堆事情。 - it will realize that oh by doing five API calls and three joins I can probably get uh what Timothy asked immediately what should happen is that um all of that uh discovery and all of that intelligence should be stored somewhere to be reused.
它會意識到，做五次 API 呼叫和三次 join 就能拿到結果。但應該發生的是，所有這些發現和智慧都應該被儲存起來以便重用。 - It’s not really how things happen.
但現實不是這樣運作的。 - It’s just basic knowledge uh about what the infrastructure of the company is.
這只是關於公司基礎設施是什麼的基本知識。 - So knowing where the tables are, what they contain, how they’re joined.
像是知道表格在哪裡、包含什麼、如何 join。 - So all of this um is compute that should be amortized basically and to me it’s really the entire game with the context engine as we call it internally is to um be in a setup where over time knowledge of the company and the context that’s available to the agent uh acrru and is maintained.
所有這些都是應該被攤提的算力，對我來說，我們內部稱之為 context engine 的整個重點，就是建立一個設置，讓公司的知識和提供給 agent 的上下文隨著時間累積並被維護。 - The second order thing of oh how was that decision reached?
第二層的問題，就是那個決策是怎麼做出的？ - Sure.
當然。 - Uh it’s going to be super interesting and it’s important, but right now I feel we’re not even in a place where it’s easy for an enterprise to have any worker uh in it be able to build an agent that has access to the right context.
這會非常有趣而且重要，但現在我覺得我們甚至還沒有到一個企業中的任何員工都能輕鬆建構有正確上下文的 agent 的階段。 - For this to happen, you have huge uh data privacy concern.
要做到這一點，你有巨大的資料隱私顧慮。 - If you want this to be efficient, you need to give access to uh the agent system to the entire uh data of your enterprise.
如果你想要高效，你需要讓 agent 系統存取整個企業的資料。 - And there is going to be arbbacks everywhere and you need to make this safe.
到處都會有護欄，你需要確保安全。
企業部署的現實狀況
- » Speaking of which, what what’s current reality of enterprise deployments of of generative AI from your perspective?
» 說到這個，從你的角度來看，目前企業部署生成式 AI 的現實狀況是什麼？ - just listening to like some of the concern like since like we very early » to me we are still in the building phase and I think it’s kind of the frustrating thing for enterprise is that when you come to um a chat assistant you feel that it’s it’s magic and it’s all going to work but as most things that have value in life there is still work to be done uh to get to them and so most of the enterprise value of AI will happen once you’ve gone through that first building phase of just setting up
聽到一些擔憂，因為我們還很早期 » 對我來說我們仍在建構階段，我覺得這對企業來說有點令人沮喪，因為你用聊天助手的時候覺得它很神奇、一切都能運作，但跟生活中大多數有價值的事情一樣，還有很多工作要做才能達到目標。所以大部分 AI 的企業價值會在你完成第一個建構階段之後才實現，就是先設置好 - all of the machinery.
所有的基礎架構。 - You’ve got to set up all of the connections.
你必須設置好所有的連接。 - You’ve got to make all of that data available.
你必須讓所有資料都可以被存取。 - And the reality is even despite a lot of work recently to make u data more available in enterprise, it’s still not easily available in the format and at the scale that we need uh for the true ROI of AI to to happen.
現實是，即使最近做了很多工作讓企業資料更容易取得，但以我們需要的格式和規模來說，要實現 AI 的真正 ROI，資料仍然不容易取得。 - And so when we come in uh there is still that phase of work that is uh just work uh to connect everything and then be able to build on it.
所以當我們進場時，仍然有一個工作階段，就是把所有東西連接起來，然後才能在上面建構。 - » So do you think we are years away from generi actually being deployed in the enterprise?
» 所以你認為生成式 AI 真正在企業中部署還要好幾年嗎？ - not years uh I think years singular uh it’s uh also to be fair to us we’ve started working I mean the company started two years ago and so most of our uh » it’s a good reminder right it’s a good reminder that like you guys have have done all of this and the company was started in yeah June 23 right if I recall » yeah and so for most of our clients uh we we started working with them recently the tooling uh for everyone is still in its infancy And so I hope that the tooling will stabil
不是好幾年，我覺得是一年。而且公平地說，公司兩年前才創立，所以我們大部分的 » 這是個好提醒，對吧，你們做了這一切，而公司是在 2023 年 6 月創立的 » 是的，對大多數客戶來說，我們最近才開始合作，每個人的工具都還在起步階段。所以我希望工具會穩定 - stabilize uh and I hope that we will have true value.
下來，我希望我們能看到真正的價值。 - True value to me is really okay we’ve gone through that first phase of building connections and now employees of that enterprise are able to use everything that we’ve built.
對我來說真正的價值是，我們已經完成了建立連接的第一階段，現在那家企業的員工能夠使用我們建構的一切。 - Right now I think we’re in a phase where we build siloed things uh because we’re scared of uh data going through walls and everything.
現在我覺得我們還在建構孤立事物的階段，因為我們害怕資料穿越界線之類的問題。 - And so to me, the real success is when you’re confident enough to give all of that control back to the company’s employees at large and they start really building on it.
所以對我來說，真正的成功是當你有足夠的信心把所有控制權交還給公司的全體員工，他們開始真正在上面建構。
未來企業需求成長
- » You’re talking about MRO in particular about the industry in general, right?
» 你說的是特別指 Mistral 還是整個產業，對吧？ - Is that do I understand this correctly?
我理解得對嗎？ - Uh because obviously that’s that’s the big question, right?
因為這顯然是大問題，對吧？ - we we all collectively building this whole thing and data centers and models and pouring uh billions and I think it’s pretty clear that from a personal use case or uh from maybe some discrete like coding use cases like the the demand is very clear uh but the big question is whether demand is going to materialize at the same level as the extraordinary level of supply we’re building » yeah around this I think the expectation is that demand and basically amount of tokens generated uh for the
我們集體建構這一切，資料中心和模型，投入了數十億，我覺得從個人使用案例或一些獨立的程式碼使用案例來看，需求是很明確的，但大問題是需求是否會達到我們正在建設的超大規模供給的同等水準 » 是的，我認為期望是企業端的需求和生成的 token 數量 - enterprise will uh completely jump once you are not bound anymore by humans asking questions or reading them.
一旦不再受限於人類提問或閱讀，將會大幅跳升。 - As soon as you have enough trust uh to have agents running in the background, as soon as you’ve set them to run a bunch of ETLs, as you’ve got them running lots of workloads, uh and you’ve got them consolidating data and knowledge across your entire company, then you’re not really um limited by the number of tokens that humans can create or read.
一旦你有足夠的信任讓 agent 在背景執行，讓它們執行一堆 ETL，讓它們處理大量工作負載，讓它們整合整個公司的資料和知識，那你就不再受限於人類能產生或閱讀的 token 數量。 - And so we I think everyone in the industry expect the demand to jump at that point.
所以我認為業界所有人都預期需求在那個時間點會大幅跳升。 - And the reality is for this to happen, you just need a lot of boring software and control and things like this.
而現實是，要讓這件事發生，你只需要大量無聊的軟體、控制和這類東西。 - » It’s amazing how much uh all of this is engineering, right?
» 令人驚訝的是，這一切有多少是工程問題，對吧？
工程 vs. 模型效能
- Versus just sheer performance of uh of models.
相對於模型純粹的效能。 - » Yeah, it’s a lot of plumbing and the goal is to make all of this plumbing easy and easier and to make it faster.
» 是的，有很多管線工程，目標是讓所有這些管線更容易、更快。 - » All right.
» 好的。 - And you said we’re about a year away.
你說我們大概還有一年。 - I » I’m not the most optimistic person.
» 我不是最樂觀的人。 - It might be faster.
也許會更快。 - Uh who knows?
誰知道呢？
重磅使用案例與 ROI 驅動力
- And we we talked about use cases a bit already, but let’s just put that one to to bed because it’s such an important question.
我們已經聊了一些使用案例，但讓我們把這個問題徹底談清楚，因為這是一個非常重要的問題。 - What do you think are the kind of the banger uh use cases in the enterprise?
你認為企業中有哪些重磅使用案例？ - Let’s assume like all agents work uh in in a in a workflow kind of way that you describe uh based on either your uh industry watch or or more specifically talking to your customers.
假設所有 agent 都以你描述的 workflow 方式運作，根據你對產業的觀察或更具體地跟客戶的交談。 - What is it that is going to generate a amazing ROI beyond coding which is pretty established at this at this stage?
除了已經相當成熟的程式開發之外，什麼會產生驚人的 ROI？ - » Yeah, there are several dimensions to this.
» 是的，這有好幾個面向。 - Coding is an obvious one and um to me to get the full um ROI of coding you need customization.
程式開發是一個明顯的方向，對我來說要獲得程式開發的完整 ROI，你需要客製化。 - Uh because a lot of ROI is unlocked uh on like sprawling code bases that are completely impossible to know for uh for something that’s been trained on the web.
因為很多 ROI 是在那些龐大的程式碼庫上釋放的，那些程式碼庫對於在網路上訓練的東西來說完全不可能了解。 - uh if you’ve got uh an enterprise that’s been building its own like domain specific languages for years, you’ll need some customization for an agent to come in and be competent uh in that respect.
如果你有一家企業多年來一直在建構自己的領域特定語言，你就需要一些客製化，才能讓 agent 進來並在這方面勝任。 - Um so coding is definitely a big one.
所以程式開發絕對是一個重要方向。 - Um if everything uh comes true as I hope I think there is still a huge jump in how we accelerate knowledge worker um and I believe the magical experience of uh you go to your chat assistant it’s connected to your system and you can ask it anything uh about the enterprise just hasn’t realized yet and it’s really obvious uh when you see the kind of queries that people are making expecting them to just work.
如果一切如我所願成真，我認為在加速知識工作者方面還會有巨大的飛躍。我相信那個神奇的體驗，就是你打開聊天助手、它連接到你的系統、你可以問它任何關於企業的問題，這個體驗還沒有實現。當你看到人們提出的那種期待它們直接運作的查詢時，這真的很明顯。 - And to me who’s building the system, it it feels like magic.
對我這個建構系統的人來說，這感覺像魔法。 - Like if you need to somehow send an email to three people and coordinate a meeting and also like gather data from some BI system, it’s just something that requires um a lot more plumbing and capabilities that we have today.
比如你需要發送電子郵件給三個人、協調一個會議、同時從某個 BI 系統收集資料，這需要比我們今天擁有的更多的管線和能力。 - Um so that’s going to be a huge lift.
所以這會是一個巨大的提升。 - And I think the last one which is maybe closer to my heart is really when we start to customize models to uh a kind of data that is particular to an industry.
我認為最後一個，也許更貼近我內心的，就是當我們開始為某個產業特有的資料客製化模型的時候。 - So typically if we uh work in oil and gas they will have systemic data that we can help uh understand and make sense of.
比如如果我們在石油和天然氣行業工作，他們會有系統性的資料，我們可以幫助理解和解讀。 - If we work with um computer assisted designs, uh they might have uh full databases of specific data formats that are not widely understood by the most general models yet.
如果我們跟電腦輔助設計合作，他們可能有完整的特定資料格式資料庫，這些格式還沒有被最通用的模型廣泛理解。 - And if we manage to build a system where in a light touch way from us or in in my dream world, we don’t really have to intervene.
如果我們能建構一個系統，我們只需要輕度介入，或在我的理想世界裡，我們根本不需要介入。 - It’s all uh self-s served for the customers.
全部由客戶自助完成。 - they can consolidate that data and then build themselves a model that really understands what their actual uh private IP is made of and make sense of this.
他們可以整合那些資料，然後自己建構一個真正理解他們私有智慧財產組成的模型，並加以解讀。 - Uh then I’ll be super happy and I think there is huge value to unlock there.
那我會非常開心，我認為那裡有巨大的價值可以釋放。 - » Great.
» 好的。
邊緣部署的理由
- Where does the edge uh fit in all of this?
邊緣運算在這一切中扮演什麼角色？ - » There are a few reasons to go edge.
» 有幾個理由要走邊緣部署。 - Uh first there are some regions where it’s more convenient to um be able to work without internet and there are also a lot of capabilities that don’t necessarily require uh a huge model.
首先，有些地區在沒有網路的情況下工作會更方便，而且也有很多功能不一定需要很大的模型。 - So if you just need something that goes uh voice to action on any device uh today with uh typically the voxal models that we develop this is doable.
所以如果你只需要在任何裝置上做到語音轉動作，以我們開發的 Voxal 模型來說，這是做得到的。 - Again, an area where the more uh focused your use case is, the smaller you can make the model through fine-tuning or um through just distillation in a in an even smaller architecture.
同樣地，你的使用場景越聚焦，就越能透過 fine-tuning 或蒸餾把模型縮到更小的架構。 - I think voice to uh action is going to be a big use case.
我認為語音轉動作會是一個很大的使用場景。 - I think it will simplify a lot uh the current stacks uh for these types of things.
我覺得這會大幅簡化目前這類應用的技術堆疊。 - There is also some privacy things uh where you could imagine uh all of the context consolidation stays on your personal device and for most things uh you can deal with a small model uh that answers a lot of your questions and then you potentially can gate uh what goes out to uh another like cloud-based models.
還有隱私方面的考量，你可以想像所有的上下文整合都留在個人裝置上，大部分事情用小模型就能回答，然後你可以控制哪些資料要送到雲端模型。 - I myself take the train a lot.
我自己經常搭火車。 - Uh I like having coding assistance.
我喜歡有程式碼輔助工具。
國防產業應用
- uh having uh DevTool run on my laptop while I code on the train is uh comfortable despite the bad Wi-Fi » and uh presumably there are some uh defense uh use cases as well.
在火車上讓 Devstral 跑在我的筆電上寫程式很舒適，即使 Wi-Fi 很差 » 而且應該也有一些國防方面的使用場景吧。 - So you you guys do quite a bit of defense work as I understand it with France with with Germany.
據我了解，你們在法國和德國做了不少國防方面的工作。 - I think you you mentioned some partnership with Helsing is AI on drones and that kind of stuff.
我記得你提過跟 Helsing 合作，做無人機上的 AI 之類的東西。 - Is that a reality?
那是真的嗎？ - » A reality it’s uh it’s something that we work on.
» 是真的，這是我們正在做的事情。 - Yes, we have a robotics division that works with these uh partners.
是的，我們有一個機器人部門在跟這些合作夥伴合作。 - Having a very um well- definfined use cases uh makes us able to really take the model down to u lighter uh types of sizes.
有非常明確定義的使用場景，讓我們能真正把模型縮小到更輕量的尺寸。 - Um and it’s of course uh use cases where control is super critical uh and you need to be um yeah able to really validate the solution.
而且這當然是控制非常關鍵的使用場景，你必須能真正驗證解決方案。 - » All right, let’s switch to the model part of the discussion.
» 好的，讓我們切換到討論模型的部分。
Mistral 3 與 MoE 架構
- In December, you guys uh released Mistrol 3, which was a big release still with thee architecture, which is at the core of what you guys have been um doing.
去年 12 月，你們發布了 Mistral 3，這是一個很大的版本，仍然採用 MoE 架構，這也是你們一直在做的核心。 - You mentioned efficiency uh earlier in the conversation.
你在對話中稍早提到了效率。 - maybe walk us through the general thinking and and approach like in a highly competitive world uh of uh AI models both in terms of closed source but also very much open source and all the Chinese labs.
可以帶我們了解一下整體思路和做法嗎？在 AI 模型高度競爭的世界裡，無論是閉源還是開源，還有所有中國的實驗室。 - What is it that you guys are trying to do and how do you position?
你們到底想做什麼？怎麼定位自己？ - Yeah.
是的。 - So we’ve released Mistral Large 3 which is uh an MOE.
我們發布了 Mistral Large 3，它是一個 MoE 模型。 - MOEs are uh really nice systems to train uh because of the lower uh amount of flops which uh makes us able to push performances um a lot more uh during training.
MoE 是非常好的訓練系統，因為較低的 flops 讓我們在訓練期間能更大幅度地提升效能。 - They are not necessarily the best formats for uh on-prem deployment because as of today uh if you want to get uh the best efficiency out of uh a mixture of experts model you require a lot of volume uh because you’re looking at deployments across dozens of GPUs usually um and to justify that amount of GPUs uh you need to have the right throughput.
它們不一定是地端部署的最佳格式，因為到目前為止，如果你想從 MoE 模型獲得最佳效率，你需要很大的流量，因為通常需要跨數十顆 GPU 部署，而要合理化那麼多 GPU，你需要有足夠的吞吐量。 - We are training uh large moes to get the best performance um with the most efficiency during training.
我們正在訓練大型 MoE 模型，以在訓練期間用最高效率達到最佳效能。 - We are also continuing to train u dense models at other scales because depending on the environments uh in which our clients want to deploy this might be the more uh costefficient solution.
我們也持續在其他規模訓練 dense 模型，因為取決於客戶想要部署的環境，這可能是更具成本效益的解決方案。 - I think both architectures are still valuable.
我認為兩種架構仍然都有價值。 - um on edge as well.
在邊緣運算上也是。 - Uh sometimes you just don’t have the RAM capacity uh to deploy something like a sparse mixture of experts and so going dense is helpful there as well.
有時候你就是沒有足夠的 RAM 容量來部署稀疏 MoE 模型，所以用 dense 模型在那裡也很有幫助。 - But yeah, definitely for training uh mixture of experts and their lower flops are very interesting.
但是，在訓練方面，MoE 和它較低的 flops 確實非常有趣。
模型開發的終極目標
- » What is the ultimate goal of the model effort?
» 模型工作的終極目標是什麼？ - I mean clearly you guys are a frontier AI lab but um are you trying to create the the best models and and solve AGI or are you trying to be the best open-source model compared to the Chinese labs or you know whatever open source eventually comes out of the uh US what is it that you’re trying to do » we’re trying to get the best uh models that we can and the model that’s most useful for uh the use cases that we cover uh in enterprise.
你們顯然是一個前沿 AI 實驗室，但你們是在試圖打造最好的模型來解決 AGI，還是想成為跟中國實驗室相比最好的開源模型？你們到底想做什麼？ » 我們在努力打造我們能做到最好的模型，以及對我們在企業端涵蓋的使用場景最有用的模型。 - And so typically with the rise of uh agentic uh behavior, one thing that’s very important is how you deal with uh various contexts, how you deal with various um documents uh being added to the input.
隨著 agentic 行為的興起，一件非常重要的事情是你如何處理各種上下文，如何處理各種被加到輸入中的文件。 - And so having the capabilities to do architecture iterations really trying new things in terms of model training is critical.
因此擁有做架構迭代的能力，真正在模型訓練方面嘗試新事物，是至關重要的。 - Um so we’re pushing the boundaries of what the current models can do with uh the compute capacity that we have but we’re also trying to focus on the things that are is most annoying uh in our deployments today.
我們在用現有的算力推動當前模型能做到的極限，但我們也在努力聚焦在當前部署中最惱人的問題上。 - And so one of the consideration that has been solved with a few harness uh tricks is the context of uh those agentic systems.
其中一個已經用一些巧妙方法解決的考量，就是那些 agentic 系統的上下文問題。 - So it’s visible typically in vibe coding but it’s um definitely uh applicable to a lot of other use cases where through all of the tool calls you’ll have to uh consolidate uh and summarize the context to be able to fit everything and uh have the model focus on the right parts.
這在 vibe coding 中最明顯，但絕對適用於很多其他使用場景，在所有的 tool call 過程中，你必須整合和摘要上下文，才能把所有東西塞進去，讓模型聚焦在正確的部分。 - To me this is just an artifact of the current architectures.
對我來說，這只是當前架構的一個產物。
Context Window 限制與解方
- uh we’re trying to fit uh things in a linear context windows where essentially the questions that we’re asking aren’t really necessarily all linear.
我們在試圖把東西塞進線性的 context window 裡，但本質上我們問的問題不一定都是線性的。 - Um and so we rely today on the file system for this and that I think that was the big change in u and realization through vibe coding is that agents are good enough at uh manipulating file systems that they can use this as a replacement for uh their context window.
所以我們現在依賴檔案系統來處理這個問題，我認為這是透過 vibe coding 帶來的重大改變和認知，就是 agent 已經足夠擅長操作檔案系統，可以用它來替代 context window。 - Basically uh they can select parts of what they want to read.
基本上它們可以選擇想讀的部分。 - they can select parts of the tool results uh and this minimizes uh the context length requirements.
它們可以選擇 tool 結果的部分內容，這樣就能把 context 長度需求降到最低。 - This is the state today.
這是今天的現況。 - I think we can do much better and I think there is a lot of uh improvements to be done on those types of uh questions.
我認為我們可以做得更好，在這類問題上還有很多改進空間。 - » Do your agents run on sandboxes?
» 你們的 agent 是在 sandbox 裡執行的嗎？
Agent 沙箱與隔離
- » It depends on the types of agents.
» 這取決於 agent 的類型。 - Uh but the answer would be yes.
但答案是肯定的。 - If it’s uh if it’s coding agents, usually uh we have uh sandboxes that will let the agent iterate uh and run.
如果是程式碼 agent，通常我們有 sandbox 讓 agent 可以迭代和執行。 - I think the depth of the uh isolation will depend on the use case.
我認為隔離的深度取決於使用場景。 - Uh typically if the file system is just representing textual context and you’re not expecting the agent to do much action on it, then you don’t really need a full sandbox.
通常如果檔案系統只是用來表示文字上下文，你不期望 agent 在上面做太多動作，那你就不需要完整的 sandbox。 - Uh you just need some representation of that context as a file system and it can be any sort of abstraction.
你只需要把上下文用檔案系統來表示，它可以是任何形式的抽象。 - But if you are I don’t know typically running asynchronous code development then yes you need a sandbox.
但如果你在做非同步的程式開發，那是的，你需要一個 sandbox。 - » Great.
» 好的。
Mistral 4 的限制因素
- What is the current constraint that um you guys are facing to make uh MR 4 when it eventually comes out do much better than ML 3.
你們目前面臨的限制是什麼，要讓 Mistral 4 最終推出時比 Mistral 3 好很多？
未來模型開發：算力、資料與合成資料
- Is that a question of MR compute or is that a question of of data and uh in particular are you guys doing anything around synthetic data that you can talk about?
這是算力的問題還是資料的問題？特別是你們有在做合成資料相關的事情可以分享嗎？ - Definitely compute and uh the current deployment that we have will help uh as it’s going to be giving us a lot more grace blackwell capacity than we had in the past.
絕對是算力，我們目前的部署會有幫助，因為它會給我們比過去多很多的 Grace Blackwell 容量。 - And so that’s uh something that we’re very excited about.
這是我們非常興奮的事情。 - And when you add uh compute, you also have to add data.
當你增加算力時，你也必須增加資料。 - And so we’ve been hard at work uh making sure that our uh data mixtures are uh as high quality as ever and growing in size.
所以我們一直在努力確保我們的資料混合品質一如既往地高，並且規模持續成長。 - But as you mentioned, one of the ways to do this is through synthetic synthetic data.
但如你所說，其中一個方法是透過合成資料。 - In terms of um where we use synthetic data the most.
在我們最常使用合成資料的地方。 - I think a lot of the interesting work that’s happening is for the post-training part where we can um build environments uh that look similar to uh an enterprise and then uh try to uh synthetically create queries that are hard and that will require multiple hops.
我認為很多有趣的工作都發生在 post-training 階段，我們可以建立類似企業環境的模擬，然後嘗試合成產生困難的、需要多步跳轉的查詢。 - And so all of this work um is in addition to the coding work, the reasoning work is really what makes the final model able to perform uh in the various uh environments that we work in.
所有這些工作加上程式碼工作、推理工作，才是讓最終模型能在我們工作的各種環境中表現良好的關鍵。 - So before it was about uh acrewing world knowledge and the uh web helps a lot with this.
以前是關於累積世界知識，網路在這方面幫助很大。 - Now it’s more and more about acquiring knowhow.
現在越來越多是關於獲取實作知識。 - Uh and for this uh it’s really about um trying to find what our uh customers are trying to do, trying to replicate it inside of our training environment and uh let the the model run basically.
為此，真正要做的是找出我們的客戶在試圖做什麼，在我們的訓練環境中複製它，然後讓模型去跑。
Pre-training vs. Post-training 與強化學習
- » You mentioned post-training and that’s one of the key topics of the last 12 months in particular this evolution of um LMS uh into systems with both pre-training and post- training and a lot of reinforcement learning.
» 你提到了 post-training，這是過去 12 個月的關鍵話題之一，特別是 LLM 演變成同時具備 pre-training 和 post-training 以及大量強化學習的系統。 - Where do you guys uh fall in that spectrum?
你們在這個光譜上落在哪裡？ - Are you uh pushing a lot of uh reinforcement learning?
你們在大力推進強化學習嗎？ - Do you believe that pre-training has still room to grow?
你認為 pre-training 還有成長空間嗎？ - How do you think about it?
你怎麼看這件事？ - » Yeah, everything still has room to grow.
» 是的，一切都還有成長空間。 - What I’m interested in as the CTO is really how you make uh all of the steps of the pipeline uh work well together and how everyone can uh develop most efficiently.
身為 CTO，我真正感興趣的是你如何讓管線的所有步驟良好地協同運作，以及每個人如何能最有效率地開發。 - Um, typically what happens in uh post training is that you will have a team that’s working on uh improving code.
通常在 post-training 階段，你會有一個團隊在改進程式碼。 - You will have another team that’s improving um different uh enterprise uh behaviors.
另一個團隊在改進不同的企業行為。 - You will have another team that’s uh improving on uh instruction following.
還有一個團隊在改進指令遵循。 - Uh and so all of this uh at some point has to come together because customers aren’t happy if you require them to deploy five different uh models to get their job done.
所有這些在某個時間點必須整合在一起，因為如果你要求客戶部署五個不同的模型才能完成工作，他們不會開心。 - There is really an internal engine and capability around making all of these work stream come together uh in the way that you expect that is super interesting to build and so but yeah uh internally we’re building and improving all of the parts of the stack.
真的有一個內部引擎和能力，讓所有這些工作流以你期望的方式整合在一起，建構這個非常有趣。是的，內部我們正在建構和改進堆疊的所有部分。 - I think the post training is very rich because it also touches all of the new use cases of LLMs and I think it’s been very exciting to see just all of the the new use case that pop up every day.
我覺得 post-training 非常豐富，因為它也觸及 LLM 的所有新使用場景，每天看到冒出來的新使用場景真的很令人興奮。 - Anytime someone on Twitter finds a new exciting things that they’ve done, then suddenly, you know, you’ve got to make it this proof of concept into potentially a base capability on which your model will perform well.
任何時候有人在 Twitter 上發現了新的有趣東西，突然之間你就得把這個概念驗證變成模型能良好表現的基礎能力。 - And that’s uh potentially an entire stream of work and you’ve got to do this efficiently and prioritize.
這可能是一整個工作流，你必須高效地做並排定優先順序。 - Well, » where doesing fall in all of this?
» 推理在這一切中處於什麼位置？
推理與工具使用整合
- Uh you guys launched a reasoning model called Magistro a few months ago.
你們幾個月前推出了一個叫 Magistral 的推理模型。 - Is that is that a big priority?
那是一個重要的優先事項嗎？ - So reasoning is a a big priority.
推理確實是一個重要的優先事項。 - And the interesting thing about reasoning was really how you can train models with reinforcement learning.
推理有趣的地方在於你如何用強化學習來訓練模型。 - And so it was first shown through reasoning, uh, because the system would learn to create better reasoning traces to, uh, get to better results.
這最先是透過推理展示出來的，因為系統會學習建立更好的推理軌跡來得到更好的結果。 - But the system is the same whether you create reasoning traces, or whether you iterate on the tools that you call, or mix them both.
但不管你是建立推理軌跡、還是迭代你呼叫的工具、或兩者混合，系統都是相同的。 - And so I think more and more the way to train, uh, all of this, uh, is going to come together.
所以我認為訓練這一切的方式越來越會整合在一起。 - And sometimes you’ll have reasoning traces, sometimes they’ll be long, sometimes they’ll be short, sometimes there won’t be any because it’s not necessary.
有時候你會有推理軌跡，有時候很長，有時候很短，有時候不需要就沒有。 - And there’s no real difference between creating a new thinking trace or calling the right tool.
建立一個新的思考軌跡和呼叫正確的工具之間沒有真正的區別。 - It’s, it’s all the same to me because what you’re optimizing at the end is what is the best, uh, output for the model to create before it gets a results, to, uh, to me.
對我來說都是一樣的，因為你最終優化的是模型在得到結果之前，能產生的最佳輸出。 - Great.
好的。
Devstral 與 Vibe CLI：Agentic Coding 與企業智慧
- Let’s talk about, uh, Devstrol 2 and the Vibe CLI.
讓我們來聊聊 Devstral 2 和 Vibe CLI。 - So walk us through those products and what they do and, uh, why people should use them.
帶我們了解這些產品、它們做什麼，以及為什麼人們應該使用它們。 - Sure.
好的。 - Um, so DevTool is our, uh, agent tech coding model.
Devstral 是我們的 agentic coding 模型。 - And so it’s something that you typically vibe code with.
所以它是你通常用來 vibe code 的東西。 - And you are more than welcome to vibe code with it through our CLI aptly named Vibe.
歡迎你透過我們恰如其名的 CLI 工具 Vibe 來 vibe code。 - Value of vibe coding and why we focus on it.
Vibe coding 的價值以及我們為什麼聚焦在這上面。 - Coding is a huge use case in enterprise, um, and especially a lot of our clients have, uh, yeah, large code databases where it’s helpful for us, uh, to take our system and customize it to their codebase to let, um, our agent run.
程式開發在企業中是一個巨大的使用場景，特別是我們很多客戶有大型的程式碼庫，我們把系統客製化到他們的程式碼庫上讓 agent 執行，這很有幫助。 - Now, the Devstrol and agentic coding is not only about, uh, vibe coding.
Devstral 和 agentic coding 不僅僅是 vibe coding。 - The same system when you run it, uh, asynchronously can be used to review PRs.
同一個系統非同步執行時可以用來審查 PR。 - Uh, it can be used to check code for specific conditions.
可以用來檢查程式碼是否符合特定條件。 - It can be used to modernize code.
可以用來現代化程式碼。 - So its applications even in coding are, uh, quite wide, as I alluded to as well.
所以它即使在程式開發方面的應用也相當廣泛，正如我之前提到的。 - Um, having a system that, uh, is good at handling a file system is more generally very interesting.
擁有一個擅長處理檔案系統的系統，在更廣泛的意義上也非常有趣。 - Uh, even if you’re not using it to code, you can use it to reason, uh, about enterprise knowledge.
即使你不是用它來寫程式，你也可以用它來推理企業知識。 - You can use it to connect to enterprise systems, and it’s, to me, it’s the basis of really the enterprise intelligence that we’re starting to build.
你可以用它連接到企業系統，對我來說，這是我們正在開始建構的企業智慧的基礎。 - And so the big news is, yeah, the, that those systems are, uh, going GA.
所以大新聞是，這些系統即將 GA。 - Uh, we’ve got, uh, an offer where chat users, so Luca, our assistant, um, will also, uh, get the ability to use Vibe and the associated models, and we’re trying to, uh, basically make that usage as wide as possible.
我們推出了一個方案，聊天使用者，也就是我們的助手 Le Chat，也將能使用 Vibe 和相關模型，我們正在努力讓使用範圍盡可能廣泛。
OCR3 與文件處理
- Another thing that you, uh, released reasonably recently, I believe, is, uh, OCR3.
你們最近發布的另一個東西，我相信是 OCR3。 - What does that do?
那是做什麼的？ - That enables you to just like, uh, scan any, uh, any form, any document.
它讓你能掃描任何表單、任何文件。 - Yeah, OCR is a huge use case in enterprise.
是的，OCR 在企業中是一個巨大的使用場景。 - Uh, a lot of our customers have, I mean, the typical example is KYC where someone will submit a form and you need to input that information in a structured way in your systems or you need to reason about it.
我們很多客戶都有這個需求，典型的例子是 KYC，有人提交一份表單，你需要以結構化的方式將資訊輸入系統，或者你需要對它進行推理。 - And so OCR, interestingly, is, uh, it’s not the types of systems that I would have expected, uh, LLM to really, uh, make large strides on.
有趣的是，OCR 不是我原本預期 LLM 會大幅進步的系統類型。 - The visual reasoning and the visual understanding has gotten so good that it’s, it’s just an easier way to process things.
視覺推理和視覺理解已經變得非常好，這只是一種更容易處理事情的方式。 - Uh, in my mind, you have any sort of input, um, and you can get the the data that you care about.
在我看來，你有任何形式的輸入，你都能取得你關心的資料。 - As I mentioned, when you build agents, you have, uh, a different type of inputs for the task that you’re trying to solve.
正如我提到的，當你建構 agent 時，對於你要解決的任務，你有不同類型的輸入。 - Documents and visual information are just a very, very frequent kind of kind of input.
文件和視覺資訊就是非常非常頻繁的一種輸入。 - Uh, sometimes it’s a lot cheaper, uh, to use a small OCR model to just get the text that you care about and then potentially post-process it or deal with it with another system than to run it through a large, uh, multimodal model that will, you, basically do the same thing but at a higher cost.
有時候用一個小的 OCR 模型來取得你要的文字，然後再後處理或用另一個系統處理，會比用一個大型多模態模型做基本上相同的事但成本更高，要便宜得多。
多模態：影像、語音與影片
- Yeah, you mentioned multimodal.
是的，你提到了多模態。 - To to which extent is Mistrol multimodal, or to which extent is that, um, voice is is video, something that you guys either do or think about, or is that just not a big enterprise use case?
Mistral 在多模態方面做到什麼程度？語音、影片，這些是你們在做的還是在思考的事情，還是說這不是一個重要的企業使用場景？ - So to answer on the first part of the question on whether, uh, we build multimodal models.
先回答問題的第一部分，關於我們是否建構多模態模型。 - Yes.
是的。 - uh it’s always a balance between exploring in a direction, getting good capabilities and getting the first model out there and then integrating it uh into the trunk like the main model that we use for everything else.
這總是在一個方向上探索、獲得好的能力、推出第一個模型，然後將其整合到主幹模型中，就是我們用於其他一切的主要模型之間取得平衡。 - And so those will always happen at separate times but for uh audio um we have uh voxil as I mentioned and all of our um main models uh understand images and can reason about them.
所以這些總是在不同時間發生，但在語音方面我們有我提到的 Voxal，我們所有的主要模型都能理解影像並進行推理。 - for videos.
至於影片。 - It’s a subject that we tackle through the lens of robotics uh first and so we’re doing our first explorations on that topic.
這是我們先透過機器人的視角來處理的主題，我們正在進行首次探索。 - » Okay.
» 好的。 - Well, again the the velocity uh has been super interesting to uh to to watch.
再次地，這個速度看起來真的很有趣。 - I um again appreciate you your reminding us that you guys have been doing this for only uh a couple of years.
我再次感謝你提醒我們你們做這些只有幾年的時間。 - So um just uh very impressive all together.
總而言之非常令人印象深刻。
工程效率與團隊建設
- maybe take taking a step back and thinking all of this in terms of uh engineering and lessons for for for builders.
也許退一步從工程和對建構者的教訓來思考這一切。 - So as we alluded to a couple of times through the conversation like you you you guys are doing a lot with uh comparatively it’s always it’s very relative in the world of AI less uh resources.
正如我們在對話中提到幾次的，你們用相對較少的資源做了很多事情，當然在 AI 世界中這都是相對的。 - How have you uh been able to do this from an efficiency standpoint?
從效率的角度來看，你們是怎麼做到的？ - We focused on the parts that we knew would provide the most uh impact uh and we focused on basically what we could afford at different times.
我們聚焦在我們知道能產生最大影響的部分，基本上聚焦在不同時期我們負擔得起的事情。 - So when we started and we had uh enough resources to train uh a few models and uh then we focused on getting the data perfect because we knew um this was potentially not the most exciting part of the work but it was absolutely critical and any improvement uh on the data quality would 10x the uh improvements that we would get by really um improving on the model architecture or things like this.
所以我們剛開始有足夠資源訓練幾個模型時，我們就聚焦在把資料做到完美，因為我們知道這可能不是工作中最令人興奮的部分，但絕對是關鍵的，任何資料品質的改善都會帶來十倍於改進模型架構之類事情的提升。 - And so I think it’s focusing the right effort uh depending on the scale and the um yeah depending on the scale of the company » and from a team uh building perspective how have you gone about it the the three of you the three co-founders have a deep background in in AI um are you these day focused mostly on building like an FDA team or are you still uh building this large kind of like research lab effort and how do you uh think about the right ratio?
所以我認為重點是根據公司的規模來聚焦正確的努力 » 從團隊建設的角度來看，你們是怎麼做的？你們三位共同創辦人在 AI 方面有深厚的背景，你們現在主要是在建立 FDE 團隊，還是仍在打造大型研究實驗室？你們怎麼看正確的比例？ - » We are growing uh all of our teams both uh research uh FDES uh product engineering uh infrastructure for compute and all of the teams have their own uh challenges in how you build and what order you uh recruit people in.
» 我們正在擴大所有團隊，研究、FDE、產品工程、運算基礎設施，每個團隊在建構方式和招聘順序上都有自己的挑戰。 - It’s been important to me um at the start to I mean to me and uh and GM and Arthur we both like the three of us were uh good AI practitioners so we knew how to train models and we knew how to code and so we started with people like us to get to the models trained the fastest um but that doesn’t work as you scale uh you it is critical to build the right uh infrastructure uh for research And so this takes different skill sets.
對我來說從一開始就很重要的是，我跟 Guillaume 和 Arthur，我們三個人都是優秀的 AI 實踐者，我們知道如何訓練模型、如何寫程式，所以我們一開始找了跟我們一樣的人來最快地完成模型訓練。但當你擴大規模時這行不通，建立正確的研究基礎設施至關重要，這需要不同的技能組合。 - Uh and it’s something that we’ve been uh building over the years as well.
這也是我們這些年來一直在建構的。 - Uh and it’s fascinating as someone who used to do uh research in a at a smaller scale to see the kind of systems that are involved and the the gains uh that you can have at scale.
作為一個曾經在較小規模做研究的人，看到涉及的系統種類以及在規模化時能獲得的收益，真的很迷人。 - Uh in terms of engineering, it’s kind of the same story really.
在工程方面，其實是同樣的故事。 - uh where you start with um a team that’s broad in its knowledge and self-sufficient and can iterate fast and then more and more you bring in experts or people that are that have seen larger scale and will tell you like well this won’t work in six months and so we should fix that now.
你從一個知識面廣、自給自足、能快速迭代的團隊開始，然後越來越多地引進專家或見過更大規模的人，他們會告訴你這個六個月後行不通，所以我們現在就應該修好它。 - So, it’s been super interesting growing the company and seeing all of the uh successive things that break at each scale and overcoming them through either changing the system, changing the organization or building new things.
所以，成長公司並看到在每個規模下會壞掉的東西，然後透過改變系統、改變組織或建構新東西來克服它們，真的非常有趣。 - » How have you navigated the whole Europe to US and rest of the world dimension of this?
» 你們是如何處理歐洲到美國和世界其他地方這個面向的？
全球營運與公司理念
- I you’re the very much the the pride of France, the pride of uh Europe as well equally.
你們是法國的驕傲，也同樣是歐洲的驕傲。 - This is a global race.
這是一場全球競賽。 - How have you uh made it work?
你們是怎麼讓它運作的？ - » So, we work um on all three continents.
» 我們在三大洲都有營運。 - We have offices uh in PaloAlto.
我們在 Palo Alto 有辦公室。 - We have offices in Singapore as well.
在新加坡也有辦公室。 - Most of our employees work uh from Paris.
我們大部分員工在巴黎工作。 - It’s a good representation of uh what we’re trying to build, which is a solution that’s uh independent and that people control.
這很好地代表了我們試圖建構的東西，就是一個獨立的、人們能掌控的解決方案。 - and in and this target uh it doesn’t really matter uh where we’re from or who we’re building for.
在這個目標下，我們來自哪裡或我們為誰建構並不重要。 - Uh we provide the tools uh and the customer the end customer then owns uh everything that’s built on it.
我們提供工具，最終客戶擁有在上面建構的一切。 - And so I I think it it hasn’t really been something that I’ve spent much thought on.
所以我覺得這不是我花太多時間思考的事情。
未來展望：ROI 與民主化
- » So uh what what should we um expect from uh Mistl over the next uh couple of years?
» 那我們未來幾年應該對 Mistral 有什麼期待？ - Over the next couple of years, I would say uh diminishing doubts on the ROI of AI uh ideally so faster uh time to success uh larger and larger uh use cases being built and really democratization uh of building tools with AI in enterprise.
未來幾年，我會說對 AI ROI 的懷疑會逐漸減少，理想情況下成功的時間會更短，建構的使用場景越來越大，並且真正在企業中民主化地用 AI 建構工具。 - I think this is really what I target for our customers.
我認為這真的是我為客戶設定的目標。 - uh it should be easy uh and most people should be able to accelerate themselves through the use of AI.
這應該是容易的，大多數人應該能透過 AI 的使用來加速自己。 - I think we’ve seen this happen quite uh impressively for coding and it should be something that happens uh a lot more widely.
我認為我們已經在程式開發方面看到這件事令人印象深刻地發生了，這應該會更廣泛地發生。 - » I was uh struck throughout this uh conversation by how pragmatic uh you you are and and focused on precise goals around enterprise success.
» 在整場對話中，我對你有多務實、多聚焦在企業成功的精確目標上印象深刻。
AGI 與企業控制
- What do you make uh of the whole, you know, rush to AGI conversation and people being AGI pill in San Francisco and other places?
你怎麼看那個，你知道的，急著追求 AGI 的討論，還有舊金山和其他地方的人都在嗑 AGI 藥丸？ - Is that is that something that you see happening or does that to some extent not matter from your perspective?
那是你看到正在發生的事情，還是從你的角度來看在某種程度上不重要？ - » I mean it it matters because the the better your systems are, the more uh impressive things you’ll be able to do and it it’ll become easier and easier.
» 我的意思是，它很重要，因為你的系統越好，你能做的事情就越令人印象深刻，而且會變得越來越容易。 - requirements I see for control and governance in enterprise make me think that even if I had uh some AGIS model on my uh servers right now if I were to go uh into a large bank and say here is a thing please let it control everything for you they wouldn’t be happy to let it do it and so I think building the infrastructure uh properly is uh quite key to following the progress of these models and really being able to quickly unleash all of their capabilities.
但我在企業中看到的控制和治理需求讓我覺得，即使我現在伺服器上有某種 AGI 模型，如果我走進一家大銀行說，這裡有一個東西，請讓它控制你的一切，他們不會願意讓它這麼做。所以我認為正確地建構基礎設施是跟上這些模型進步並真正能快速釋放所有能力的關鍵。 - So to me it’s it’s two directions that are necessary.
所以對我來說，這是兩個必要的方向。 - You need to improve the capabilities of the model and it’s super exciting to do so but the journey of uh making it trivial and uh easy for everyone to unleash those models on your enterprise workflows uh without really wondering what’s going to happen is is equally important.
你需要提升模型的能力，這樣做非常令人興奮，但讓每個人都能輕鬆地在企業工作流程上釋放這些模型而不用擔心會發生什麼事的旅程，同樣重要。 - And honestly super uh super fun as well to develop.
而且說實話，開發起來也非常有趣。
結語與致謝
- There are lots of super interesting questions.
有很多非常有趣的問題。 - » Wonderful.
» 太好了。 - Well, Timote, thank you so much for uh doing this uh deep dive on Mistrol with us.
Timothee，非常感謝你跟我們一起做了這次關於 Mistral 的深度對談。 - It’s been fascinating.
這真的很精彩。 - Congratulations on everything that you’ve built again in this very short period of time.
恭喜你們在這麼短的時間內建構的一切。 - Uh and excited for what’s uh coming next.
期待接下來的發展。 - So, thank you for spending time with us.
謝謝你花時間跟我們在一起。
Podcast 結尾與聽眾互動
- » Thanks.
» 謝謝。 - It was a pleasure.
很開心。 - » Hi, it’s Matt Kirk again.
» 嗨，我又是 Matt Turk。 - Thanks for listening to this episode of the Mad Podcast.
感謝收聽這一集的 Mad Podcast。 - If you enjoyed it, we’d be very grateful if you would consider subscribing if you haven’t already, or leaving a positive review or comment on whichever platform you’re watching this or listening to this episode from.
如果你喜歡的話，如果你還沒有訂閱，我們會非常感謝你考慮訂閱，或在任何你觀看或收聽這集節目的平台上留下正面評論或評價。 - This really helps us build a podcast and get great guests.
這真的幫助我們建構 podcast 並邀請到優秀的來賓。 - Thanks, and see you at the next episode.
謝謝，下一集見。