[{"data":1,"prerenderedAt":629},["ShallowReactive",2],{"blog-ko-dont-let-your-vdrs-ai-tank-your-deal":3,"blog-related-ko-dont-let-your-vdrs-ai-tank-your-deal":201},{"id":4,"title":5,"author":6,"body":7,"coAuthors":188,"date":189,"description":17,"detail":190,"extension":191,"featured":190,"image":192,"imageSize":188,"meta":193,"navigation":194,"path":195,"railTitle":188,"seo":196,"stem":197,"summary":188,"tags":198,"unlisted":190,"updatedAt":188,"video":188,"__hash__":200},"blog\u002Fblog\u002Fen\u002Fdont-let-your-vdrs-ai-tank-your-deal.md","Don’t let the VDR’s AI tank your deal","Pierre-Louis",{"type":8,"value":9,"toc":178},"minimark",[10,18,21,25,28,31,34,37,42,45,48,51,54,57,60,64,67,70,73,77,80,83,86,91,94,97,101,104,107,110,114,117,124,128,131,134,137,140,143,146,150,153,156,159,162,164],[11,12,14],"h4",{"id":13},"vdr-vendors-are-racing-to-add-ai-features-most-are-getting-it-wrong-putting-deals-at-risk",[15,16,17],"em",{},"VDR vendors are racing to add AI features. Most are getting it wrong, putting deals at risk.",[19,20],"hr",{},[22,23,24],"p",{},"A mid-market M&A team uploads 3,000 documents to their virtual data room (VDR). They activate the vendor's flagship AI redaction tool: \"bulk-redact thousands of files in seconds\", the marketing promised. Minutes later, the system reports completion. Yet buried on page 37 of an employment contract, a social security number remains plainly visible. The AI found punctuation marks to redact instead.",[22,26,27],{},"This is the reality of what happens in legacy VDRs today. It even appears in the marketing demo from the world's largest VDR provider... Power users know it well.",[22,29,30],{},"VDR providers are racing to bolt generative AI onto their platforms, pitching \"due diligence agents\" and \"intelligent redaction\" to M&A teams who need reliability above all else.",[22,32,33],{},"The technology isn't ready. The implementations are naive. And the consequences fall on deal teams who discover, sometimes after buyers have already accessed files, that their AI-powered features failed.",[22,35,36],{},"The pitch sounds compelling: let algorithms handle repetitive work, answer buyer questions instantly, flag sensitive information automatically. The reality resembles autonomous driving circa 2010: impressive demos that work on a circuit, but frequent accidents in real life, and no one really willing to remove their hands from the wheel, or to take responsibility for crashes.",[38,39,41],"h3",{"id":40},"the-demo-breaks-in-production","The demo breaks in production",[22,43,44],{},"Consider two examples that illustrate the pattern: document redaction and Q&A automation.",[22,46,47],{},"Automated redaction: Some platforms employ large language models to scan documents for personally identifiable information, promising to eliminate manual review. In practice, these systems struggle with many steps, starting with document parsing: PDFs with complex layouts confound them, scanned images defeat them entirely, tables scramble their logic. A model might correctly identify social security numbers in some documents but miss others, etc.",[22,49,50],{},"Worse: there are no safeguards. When these systems err, they do so confidently. No warning flags, no uncertainty scores. Just a green checkmark and a latent liability.",[22,52,53],{},"The Q&A implementations present different risks. Some vendors now offer chatbots that answer buyer questions by querying uploaded documents through retrieval-augmented generation. The appeal is obvious: accelerate diligence, reduce email volleys, free up deal team time.",[22,55,56],{},"But imagine the scenario: a strategic buyer asks about warranty claims history. The chatbot, trained on incomplete document chunks with a context window that can't span the full file set, responds with a figure. The figure is wrong, off by 40%, or citing a superseded version, or hallucinating entirely. The buyer relies on it. The deal reprices or, worse, litigation follows post-close.",[22,58,59],{},"No M&A professional would let a junior analyst answer buyer questions without review. Yet they're being asked to trust a system that cannot explain its reasoning, cannot flag its uncertainty, and cannot be held accountable.",[38,61,63],{"id":62},"the-control-problem","The control problem",[22,65,66],{},"The fundamental issue isn't that AI performs poorly, it's that autonomous AI removes human oversight at precisely the moments when oversight matters most.",[22,68,69],{},"Sell-side teams don't need software that makes decisions. They need tools that preserve their ability to make decisions while reducing mechanical friction.",[22,71,72],{},"Not a system that bulk-redacts files, but one that identifies every instance of a pattern and lets humans verify before applying changes. Not a sloppy chatbot inside the data room, but infrastructure that lets deal teams in control, and potentially use their preferred AI platforms while maintaining custody and control of documents.",[38,74,76],{"id":75},"what-smart-controls-should-actually-look-like","What smart controls should actually look like",[22,78,79],{},"The alternative approach doesn't try to make AI autonomous. It makes AI context-aware.",[22,81,82],{},"Entropia, a VDR built by former Google engineers, is implementing this philosophy. Rather than adding AI that make autonomous decisions, the platform focuses on three principles: grounding AI suggestions in user behaviour, scaling individual decisions across bounded contexts, and replaying those decisions as circumstances change.",[22,84,85],{},"Such a context-aware system observes what you're doing, learns from the patterns in your specific environment, and scales your decisions without making new ones.\nCritically, every suggestion requires human validation before execution.",[87,88,90],"h5",{"id":89},"ground-suggestions-in-what-users-are-actually-doing","Ground suggestions in what users are actually doing",[22,92,93],{},"When a deal team renames a file, the system doesn't just accept the change. It analyzes the document type, examines how similar files have been named, and suggests standardized nomenclature for the next document. The AI isn't deciding on a naming convention, it's inferring the convention the team is already using and offering to apply it consistently.",[22,95,96],{},"The context matters. An agreement uploaded to the \"Contracts\u002FSuppliers\" folder gets different naming suggestions than the same agreement uploaded to \"Contracts\u002FCustomers.\" The AI reads the environment, not just the file.",[87,98,100],{"id":99},"scale-individual-actions-across-bounded-contexts","Scale individual actions across bounded contexts",[22,102,103],{},"An analyst redacts a social security number in an employment agreement. That single action (hiding this specific nine-digit string) becomes a bounded context the system can replicate. The AI searches every document in the data room for that exact string and flags each instance. The analyst reviews the results, then applies the redaction across all confirmed matches.",[22,105,106],{},"One decision, executed fifty times. The human made the judgment call about what to redact. The AI handled the pattern matching and execution.",[22,108,109],{},"This extends to recurring tasks. A team uploads financial statements monthly throughout a deal. They redact certain fields in the first upload (specific salary bands, particular contract values). When new statements arrive, the system identifies analogous fields based on position, formatting, and previous redaction patterns, then flags them for review. The team confirms or adjusts, then applies.",[87,111,113],{"id":112},"replay-decisions-when-circumstances-change","Replay decisions when circumstances change",[22,115,116],{},"Documents don't arrive all at once. A data room grows throughout diligence as teams locate additional materials or buyers request specific files. Traditional approaches require reviewing each new upload from scratch: checking for sensitive information, verifying naming conventions, confirming version control.",[22,118,119,120,123],{},"Context-aware systems treat previous decisions as reusable templates. If ten contracts have already been reviewed and redacted following a specific pattern, the eleventh contract gets automatically scanned for similar elements. Not automatically redacted, but automatically ",[15,121,122],{},"flagged for the same review process",".",[38,125,127],{"id":126},"let-users-work-on-the-data-room-from-their-preferred-ai-platforms-while-staying-in-control","Let users work on the data room from their preferred AI platforms while staying in control",[22,129,130],{},"M&A teams already have powerful AI tools they trust. Some use general-purpose platforms (Google Gemini, Mistral, ChatGPT, Claude) for analysis or drafting. Others rely on specialized platforms like Harvey, Hebbia, or Legora for diligence.",[22,132,133],{},"These platforms outperform the generic chatbots VDR providers bolt onto their products. They have hundreds (or thousands) of engineers refining models and orchestrating them within sophisticated products. VDR providers, by contrast, are just adding chatbots as naïve interface features, with poor back-end orchestration or infrastructure, and often with minimal quality guardrails.",[22,135,136],{},"The relevant question becomes whether VDR providers will accommodate the tools teams have chosen, or force them to use inferior alternatives embedded in the platform.",[22,138,139],{},"Today, the only alternative is either to use inferior AI features in data rooms (a poor option which often requires extra payment to the VDR provider), or taking all documents out of the data room to upload them on other AI platforms (which creates copies outside the VDR's security perimeter).",[22,141,142],{},"There is a third option. Entropia is providing a secure connection between the VDR and the AI platforms, via a protocol named MCP (for \"Model Context Protocol\") *.",[22,144,145],{},"An MCP server enables users to work with whichever AI platform they prefer, but those platforms query the data room through controlled APIs rather than receiving file copies. A lawyer using Claude to analyze sale agreements doesn't download fifty contracts. Claude queries the VDR for relevant clauses, receives text snippets within defined parameters, and generates analysis. The documents never leave secure storage. Every query gets logged inside the VDR, providing insights to the seller. Access permissions remain consistent across human and AI users.",[38,147,149],{"id":148},"what-happens-next","What happens next?",[22,151,152],{},"The AI gold rush in enterprise software follows a predictable pattern: vendors add language models because competitors are doing it and investors expect it, not because customers need them. Features proliferate faster than quality control. Early adopters become beta testers with real stakes.",[22,154,155],{},"VDRs are particularly unsuited to this dynamic. M&A deals involve sensitive information, tight timelines, and legal exposure. They reward reliability over innovation theater. A chatbot that's right, say, 75% of the time sounds impressive until you consider that the 25% of errors could tank a deal or trigger litigation.",[22,157,158],{},"Perhaps the more interesting question is whether M&A teams will demand better, or simply adapt to the new risks. The evidence suggests they're choosing a third path: using AI tools they trust and managed to get right. And those tools don't come from VDR providers.",[22,160,161],{},"Control, it turns out, means knowing when to build and when to connect.",[19,163],{},[165,166,167,173],"blockquote",{},[22,168,169],{},[170,171,172],"strong",{},"* MCP Servers Explained",[22,174,175],{},[15,176,177],{},"An MCP server acts as a secure intermediary between AI platforms and sensitive data. Rather than uploading files to ChatGPT or Claude directly, which creates copies outside your control, the MCP server exposes predefined APIs that let these tools query your data room for specific context. Users choose their preferred AI platform. Those platforms access only what the user has permission to see, and their actions get logged in the data room. Think of it as a read-only interface for language models, with granular access controls and full audit trails. Teams leverage powerful AI analysis without surrendering document custody.",{"title":179,"searchDepth":180,"depth":180,"links":181},"",2,[182,184,185,186,187],{"id":40,"depth":183,"text":41},3,{"id":62,"depth":183,"text":63},{"id":75,"depth":183,"text":76},{"id":126,"depth":183,"text":127},{"id":148,"depth":183,"text":149},null,"2025-12-19",false,"md","\u002Fblog\u002Fscreenshot-2025-12-19-at-10.39.01.png",{},true,"\u002Fblog\u002Fen\u002Fdont-let-your-vdrs-ai-tank-your-deal",{"title":5,"description":17},"blog\u002Fen\u002Fdont-let-your-vdrs-ai-tank-your-deal",[199],"insights","CKAW5gASnZgAPjfprkMeCBQu56MTQDWLY7Dgae4wb50",[202,297,572],{"id":203,"title":204,"author":205,"body":206,"coAuthors":282,"date":284,"description":285,"detail":190,"extension":191,"featured":194,"image":286,"imageSize":188,"meta":287,"navigation":194,"path":288,"railTitle":289,"seo":290,"stem":291,"summary":292,"tags":293,"unlisted":190,"updatedAt":188,"video":188,"__hash__":296},"blog\u002Fblog\u002Fen\u002Fentropia-jimini-partnership.md","Entropia connects the data room to Jimini AI","Pierre-Louis Corteel",{"type":8,"value":207,"toc":276},[208,219,224,227,231,234,238,246,250,253,261,269],[22,209,210,211,218],{},"We are pleased to partner with ",[212,213,217],"a",{"href":214,"rel":215},"https:\u002F\u002Fwww.jimini.ai",[216],"nofollow","Jimini AI",", the sovereign legal AI built for law firms and in-house legal teams. Legal professionals can now search, read, analyse or act on their Entropia data rooms directly from Jimini AI.",[220,221,223],"h2",{"id":222},"a-new-way-to-run-legal-due-diligence","A new way to run legal due diligence",[22,225,226],{},"Law firms, in-house legal teams and notaries working on M&A transactions spend days in the data room. Until now, to analyse a data room document in their legal AI, they had to download it from the data room first. With the Entropia connector, Jimini AI users can search and read the data room alongside the legal sources they already use in Jimini AI.",[220,228,230],{"id":229},"the-deal-documents-and-the-law-in-one-answer","The deal documents and the law, in one answer",[22,232,233],{},"In a single request, users can query the Entropia data room and legal sources together. They analyse contract clauses and check each one against the applicable law, with citations to the page in the data room and to the legal source. With the Analysis Grid, they apply the same questions to hundreds of data room contracts at once, each cell linked to the passage it came from.",[220,235,237],{"id":236},"entropia-tools-available-from-jimini-ai","Entropia tools, available from Jimini AI",[22,239,240,241,245],{},"The ",[212,242,244],{"href":243},"\u002Fresources\u002Fentropia-mcp-server","Entropia MCP server"," enables Jimini AI's Agent to browse the data room folders, run full-text searches, read documents and spreadsheets, and retrieve the questions and answers exchanged in Q&A threads. A clause to find, a seller's answer in the Q&A, a figure in a spreadsheet: legal teams simply ask Jimini AI.",[220,247,249],{"id":248},"under-the-rights-you-already-hold","Under the rights you already hold",[22,251,252],{},"Every request is initiated by a legal professional, scoped to their access rights in Entropia and logged under their name. Data room administrators decide which groups can use AI. Where the sell side allows it, the bidder's advisers work within their own permissions. Nothing reaches a counterparty from Jimini AI: a person reviews each drafted answer and sends it from Entropia. Both companies keep customer data under European control.",[165,254,255],{},[22,256,257,260],{},[15,258,259],{},"\"Lawyers on a deal need the deal documents and the law side by side. Now they can work on both from the AI platform they have chosen,\""," said Pierre-Louis Corteel, CEO of Entropia.",[165,262,263],{},[22,264,265,268],{},[15,266,267],{},"\"Our ambition is to make Jimini AI the place where legal work gets done, seamlessly connected to the tools legal teams rely on. In M&A, that means bringing the data room directly into their workflow. With Entropia, lawyers can now manage the entire due diligence process within Jimini AI, from reviewing deal documents to drafting the final report. And as two European companies, we share a commitment to keeping sensitive deal data under European control,\""," said Raphaël Arroche, CEO of Jimini AI.",[22,270,271,272,123],{},"Discover Jimini AI at ",[212,273,275],{"href":214,"rel":274},[216],"jimini.ai",{"title":179,"searchDepth":180,"depth":180,"links":277},[278,279,280,281],{"id":222,"depth":180,"text":223},{"id":229,"depth":180,"text":230},{"id":236,"depth":180,"text":237},{"id":248,"depth":180,"text":249},[283],"Raphaël Arroche","2026-11-04","Legal teams can now search, read, analyse and act on their Entropia data rooms directly from Jimini AI, the sovereign legal AI.","\u002Fblog\u002Fjimini-partnership-cover.png",{},"\u002Fblog\u002Fen\u002Fentropia-jimini-partnership","Jimini AI partnership",{"title":204,"description":285},"blog\u002Fen\u002Fentropia-jimini-partnership","Legal teams can now search, read, analyse and act on their Entropia data rooms directly from Jimini AI.",[294,199,295],"solution","changelog","ZMbapJnn_5FTkBkQ6AGv1JLABeKfPZzMIMrcPHUS1cU",{"id":298,"title":299,"author":205,"body":300,"coAuthors":559,"date":562,"description":563,"detail":190,"extension":191,"featured":190,"image":564,"imageSize":188,"meta":565,"navigation":194,"path":566,"railTitle":188,"seo":567,"stem":568,"summary":569,"tags":570,"unlisted":190,"updatedAt":188,"video":188,"__hash__":571},"blog\u002Fblog\u002Fen\u002Fma-tech-circle-paris-september-2026.md","AI in M&A: lessons from Clipperton and Edmond de Rothschild",{"type":8,"value":301,"toc":545},[302,305,308,312,316,320,326,329,335,338,341,344,348,351,354,358,363,366,369,372,376,381,384,387,391,394,434,438,443,446,449,452,459,463,468,471,474,477,481,486,489,492,495,498,502,507,510,513,517,520,523,527,530,538],[22,303,304],{},"The M&A Tech Circle, hosted by Entropia, brings M&A professionals together every quarter to explore how technology is transforming the profession, through real cases and first-hand accounts.",[22,306,307],{},"Here is our recap of the Paris edition, held on 23 September 2026 at Entropia's offices.",[309,310],"blog-gallery",{":images":311},"[{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Froom-1.jpg\",\"alt\":\"The M&A Tech Circle room (1\u002F4)\",\"width\":2419,\"height\":3226},{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Froom-2.jpg\",\"alt\":\"The M&A Tech Circle room (2\u002F4)\",\"width\":4898,\"height\":3265},{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Froom-3.jpg\",\"alt\":\"The M&A Tech Circle room (3\u002F4)\",\"width\":3265,\"height\":4898},{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Froom-4.jpg\",\"alt\":\"The M&A Tech Circle room (4\u002F4)\",\"width\":1348,\"height\":899}]",[313,314],"section-label",{"label":315},"Announcements",[220,317,319],{"id":318},"entropias-partnerships-with-claude-anthropic-rogo-and-model-ml","Entropia's partnerships with Claude (Anthropic), Rogo and Model ML",[22,321,322,323],{},"The evening opened with a look back at the partnerships Entropia recently signed with the leading AI platforms used by investment banks. Entropia was the first data room in the world with a native connector in Claude, Rogo and Model ML. M&A teams can connect Entropia to most AI platforms on the market: every one that supports MCP (Model Context Protocol). ",[212,324,325],{"href":243},"Learn more",[22,327,328],{},"It was also a chance to spell out Entropia's differentiation strategy. Some data room market leaders sell their own AI layer as a paid add-on and deliberately limit what their MCP server can do, when they have one at all. Yet their clients have already chosen their AI platform, built their skills on it and connected their sources. And even after paying for the add-on, they still lack essential tools, such as reading the full text of the documents in the data room.",[22,330,331,332],{},"Entropia sided with its clients: from their own AI platform, their teams work on the data room with far more complete tools, full document reading included, at no extra cost. ",[212,333,325],{"href":334},"\u002Fblog\u002Fmost-capable-data-room-mcp",[22,336,337],{},"Those capabilities make all the difference for teams that work with AI every day.",[22,339,340],{},"That is the case for the M&A teams at Edmond de Rothschild, who rolled out Rogo, and at Clipperton, who work with Claude. Manon Elbaz and Olivier Combaudou then took the stage to talk through both rollouts.",[313,342],{"label":343},"Panel",[220,345,347],{"id":346},"edmond-de-rothschild-x-clipperton","Edmond de Rothschild x Clipperton",[22,349,350],{},"Manon Elbaz, Chief of Staff at Edmond de Rothschild, and Olivier Combaudou, Partner at Clipperton, both lead the AI rollout at their bank. They came to share how it is going.",[309,352],{":images":353},"[{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Fpanel-1.jpg\",\"alt\":\"The panel with Manon Elbaz and Olivier Combaudou (1\u002F5)\",\"width\":557,\"height\":836},{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Fpanel-2.jpg\",\"alt\":\"The panel with Manon Elbaz and Olivier Combaudou (2\u002F5)\",\"width\":557,\"height\":836},{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Fpanel-3.jpg\",\"alt\":\"The panel with Manon Elbaz and Olivier Combaudou (3\u002F5)\",\"width\":1254,\"height\":836},{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Fpanel-4.jpg\",\"alt\":\"The panel with Manon Elbaz and Olivier Combaudou (4\u002F5)\",\"width\":618,\"height\":836},{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Fpanel-5.jpg\",\"alt\":\"The panel with Manon Elbaz and Olivier Combaudou (5\u002F5)\",\"width\":624,\"height\":936}]",[38,355,357],{"id":356},"how-to-choose-an-ai-platform","How to choose an AI platform",[22,359,360],{},[15,361,362],{},"Choices that are anything but final.",[22,364,365],{},"Edmond de Rothschild uses Rogo, a platform specialised in financial analysis. Manon Elbaz detailed the criteria that tipped the balance. AI models move very fast, and Rogo lets teams switch from one to another without changing tools.",[22,367,368],{},"Clipperton built its own environment, with Claude as its main platform, and handles orchestration, connections and skills itself.",[22,370,371],{},"Both stressed that these choices are anything but final and may well change, because the bulk of the work does not lie in picking the platform. After all, if every bank works with the same tool, how do you stand out? For Clipperton, the first step was to write down its processes and its knowledge base.",[38,373,375],{"id":374},"rolling-out-ai-in-an-ma-team-what-takes-time","Rolling out AI in an M&A team: what takes time",[22,377,378],{},[15,379,380],{},"Structuring knowledge, making methods explicit, training the teams.",[22,382,383],{},"At Clipperton, the time went into the conditions of use more than into the technology. The biggest workstream is still the knowledge base: the partners wrote down methods, guidelines and processes that had never been formalised. That pushed them to think more explicitly about what makes them distinctive. The team also drafted a charter, and one rule sums up its spirit: \"The AI did it\" is never an excuse.",[22,385,386],{},"At Edmond de Rothschild, Manon Elbaz described a pilot phase in which the team is learning fast. The time goes into training, so people learn to use the tool and to challenge it, into defining the framework of use, into tracking usage and into working with internal \"champions\". Her goal is to measure how each person uses it, so that AI stays in the service of judgement. Usage varied widely at first. Today the whole team uses it, and real power users have emerged among the analysts. The challenge now is to make usage consistent.",[38,388,390],{"id":389},"the-ai-use-cases-that-work-in-investment-banking","The AI use cases that work in investment banking",[22,392,393],{},"On use cases, Manon Elbaz and Olivier Combaudou agreed on the essentials. Their teams already use AI every day across many use cases:",[395,396,397,404,410,416,422,428],"ul",{},[398,399,400,403],"li",{},[170,401,402],{},"Sourcing and buyer identification",": quickly finding prospects, potential buyers and sector data.",[398,405,406,409],{},[170,407,408],{},"Sparring",": challenging an approach, exploring scenarios, finding angles nobody had thought of.",[398,411,412,415],{},[170,413,414],{},"Knowledge management",": building knowledge bases on specific topics. One example at Clipperton is comparables: building lists of multiples to use as benchmarks.",[398,417,418,421],{},[170,419,420],{},"First drafts",": bankers still write the final version of an investment thesis themselves, but save significant time on the first one.",[398,423,424,427],{},[170,425,426],{},"Checking",": spotting typos and inconsistencies between the sections of a document.",[398,429,430,433],{},[170,431,432],{},"Q&A",": preparing answers to buyers' questions. Olivier Combaudou walked through Clipperton's setup, the tools connected to Claude and the in-house skills, then the Q&A skill applied to a recent mandate. The whole team uses it.",[38,435,437],{"id":436},"what-banks-dont-want-to-automate","What banks don't want to automate",[22,439,440],{},[15,441,442],{},"You can ask AI to write an IM, but you shouldn't.",[22,444,445],{},"Both speakers draw the same line. Negotiations and direct exchanges with clients stay entirely human: the relationship, the nuance and reading what goes unsaid cannot be delegated.",[22,447,448],{},"As for automating key deliverables such as the IM, the tests have not been conclusive. Bankers get generic text, while the IM is one of the documents where a bank most wants to stand out. The variety of sources and formats also makes the AI's job harder.",[22,450,451],{},"Olivier Combaudou pointed to the difficulties that come with the nature of Clipperton's mandates, mostly with tech companies, where you price a vision more than an EBITDA multiple.",[22,453,454,455,123],{},"Automating deliverables is probably more within reach for mature companies in standardised markets, such as the accounting-firm build-up Aesus Advisory presented at ",[212,456,458],{"href":457},"\u002Fblog\u002Fma-tech-circle-paris","the previous edition of the M&A Tech Circle",[38,460,462],{"id":461},"how-to-measure-the-roi-of-ai-in-ma","How to measure the ROI of AI in M&A",[22,464,465],{},[15,466,467],{},"Ask a better question.",[22,469,470],{},"The ROI of AI is still hard to put a number on. You can get a rough idea by dividing the annual cost by the number of pitches and mandates. You can try to estimate the time saved per person and per grade, but a complete costed simulation is not realistic at this stage. The cost of AI is far from neutral and keeps rising, current prices probably don't reflect the real cost of these tools, and nobody knows what they will be in six or twelve months.",[22,472,473],{},"Above all, it isn't really the right question to ask for now. Nobody measured the ROI of the internet or of laptops overnight, but everyone saw the way people work change profoundly.",[22,475,476],{},"M&A processes will evolve and teams will work differently. Both speakers agreed that the best bankers will pull further ahead thanks to AI. The point is to raise the quality of deliverables, not so much productivity. The ROI will become clear over time.",[38,478,480],{"id":479},"will-ai-reduce-junior-hiring-in-investment-banking","Will AI reduce junior hiring in investment banking?",[22,482,483],{},[15,484,485],{},"No, but it changes what is expected of them.",[22,487,488],{},"The short answer: nothing points that way today. Neither bank has changed its hiring.",[22,490,491],{},"Edmond de Rothschild is still observing. For Manon Elbaz, even as AI lifts every level, interns must keep mastering the fundamentals on real cases: building a business plan, a DCF, a valuation, calculating a WACC. A junior who has never built a DCF by hand cannot challenge the AI, just as a manager must know how to do what their team does.",[22,493,494],{},"Clipperton is not cutting its hiring at all, but it is looking for slightly different profiles, and its expectations of juniors have risen markedly. Olivier Combaudou doesn't want \"prompt operators\" but analysts who can articulate a line of thought. He values new skills: in the past, a tedious task done well was enough, and the senior brought the thinking afterwards. Once that task can be automated, the bank expects the junior to deliver a first layer of added value.",[22,496,497],{},"The question is how to use AI to augment every member of the bank, not how to automate the interns' work. Comparing an analyst's productivity with a model's only holds for certain tasks. And a bank that hired fewer juniors would deprive itself of its future associates, managers and partners.",[38,499,501],{"id":500},"the-investment-bankers-job-in-ten-years","The investment banker's job in ten years",[22,503,504],{},[15,505,506],{},"The bank's added value doesn't change.",[22,508,509],{},"What is becoming a commodity is the exploration and analysis of large amounts of information: the first layer of market analysis, summaries of public data, data structuring. Differentiation will therefore come down to judgement, vision, the ability to ask the right questions and see what others miss, and knowledge of markets, sectors and players.",[22,511,512],{},"The job stays competitive: a client comes looking for the best deal and the right investor. An investment bank's deliverables are a means to that end, not the end itself. If part of the work becomes a commodity, the bank's added value doesn't change. Bankers will be able to spend more time advising, and the bank will be able to take more risks. Execution speeds up, but deciding what to do with it remains an in-house craft. And AI only works on part of a deal's raw material: the data, provided it is well organised!",[220,514,516],{"id":515},"cocktails","Cocktails",[22,518,519],{},"The conversations went on late into the evening, sparking many introductions and plenty of new projects.",[309,521],{":images":522},"[{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Fcocktail-1.jpg\",\"alt\":\"Cocktails after the panel (1\u002F6)\",\"width\":1105,\"height\":737},{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Fcocktail-2.jpg\",\"alt\":\"Cocktails after the panel (2\u002F6)\",\"width\":557,\"height\":836},{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Fcocktail-3.jpg\",\"alt\":\"Cocktails after the panel (3\u002F6)\",\"width\":557,\"height\":836},{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Fcocktail-4.jpg\",\"alt\":\"Cocktails after the panel (4\u002F6)\",\"width\":557,\"height\":836},{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Fcocktail-5.jpg\",\"alt\":\"Cocktails after the panel (5\u002F6)\",\"width\":624,\"height\":936},{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Fcocktail-6.jpg\",\"alt\":\"Cocktails after the panel (6\u002F6)\",\"width\":1348,\"height\":899}]",[220,524,526],{"id":525},"in-the-room","In the room",[22,528,529],{},"A big thank you to everyone who joined us:",[531,532,535],"callout",{"color":533,"icon":534},"neutral","i-lucide-users",[22,536,537],{},"2A Capital · ABN Amro · Adviso Partners · Amala Partners · Arcéane · Ardian · Atout Capital · BNP Paribas · Bucéphale Finance · CA CIB · Cambon Partners · Clipperton · D&A Corporate Finance · Edmond de Rothschild · Eight Advisory · EY M&A Advisory · EY Parthenon · Forvis Mazars · Fournel Advisory · Groupama · in Extenso · Indosuez CF · Inflection Point Technology Partners · Investec · Klutch Capital · Largillière Finance · Leuwen Euroland · M&A Advisor · Mansartis · Maupiti Partners · Mazarine Partners · MBA Capital · Priscus Finance · Raisers Partners · Rochefort & Associés · Rydge Conseil · Sanofi · Spayne Lindsay · Tevali Partners · TTS Group · Training You · Trialis · Windome Banking",[539,540],"matc-credits",{":entropiaHeight":541,":partners":542,"entropiaHref":543,"label":544},"19","[{\"src\":\"\u002Fblog\u002Fmatc-202609\u002Fclipperton-logo-dark.png\",\"alt\":\"Clipperton\",\"href\":\"https:\u002F\u002Fwww.clipperton.net\u002F\",\"height\":14},{\"src\":\"\u002Fcustomers\u002Fedmond-de-rothschild.png\",\"alt\":\"Edmond de Rothschild\",\"href\":\"https:\u002F\u002Fwww.edmond-de-rothschild.com\u002F\",\"height\":56}]","\u002F","Organized by",{"title":179,"searchDepth":180,"depth":180,"links":546},[547,548,557,558],{"id":318,"depth":180,"text":319},{"id":346,"depth":180,"text":347,"children":549},[550,551,552,553,554,555,556],{"id":356,"depth":183,"text":357},{"id":374,"depth":183,"text":375},{"id":389,"depth":183,"text":390},{"id":436,"depth":183,"text":437},{"id":461,"depth":183,"text":462},{"id":479,"depth":183,"text":480},{"id":500,"depth":183,"text":501},{"id":515,"depth":180,"text":516},{"id":525,"depth":180,"text":526},[560,561],"Manon Elbaz","Olivier Combaudou","2026-09-23","Clipperton and Edmond de Rothschild on their AI rollouts: their setups, what works, what takes time. ROI, the impact on staffing, and where the job is heading.","\u002Fblog\u002Fma-tech-circle-paris-202609.png",{},"\u002Fblog\u002Fen\u002Fma-tech-circle-paris-september-2026",{"title":299,"description":563},"blog\u002Fen\u002Fma-tech-circle-paris-september-2026","Clipperton and Edmond de Rothschild on their AI rollouts in M&A.",[199],"zt0P3BZtR3s6mzIU4VCukGs4nwFjha6syVSRNrrog3U",{"id":573,"title":574,"author":6,"body":575,"coAuthors":616,"date":618,"description":619,"detail":190,"extension":191,"featured":194,"image":620,"imageSize":188,"meta":621,"navigation":194,"path":622,"railTitle":623,"seo":624,"stem":625,"summary":626,"tags":627,"unlisted":190,"updatedAt":188,"video":188,"__hash__":628},"blog\u002Fblog\u002Fen\u002Fentropia-model-ml-partnership.md","Entropia's Partnership with Model ML",{"type":8,"value":576,"toc":614},[577,587,590,597,600,603,611],[22,578,579,580,586],{},"We are pleased to announce a partnership with ",[212,581,585],{"href":582,"rel":583,"target":584},"https:\u002F\u002Fmodelml.com",[216],"_blank","Model ML",", the agentic operating system for financial services, used by some of the world's largest banks, asset managers and advisory firms.",[22,588,589],{},"Model ML builds custom and off-the-shelf agents which create, review and verify complex, document-heavy workflows for global investment banks, private equity houses, consulting firms and other major financial institutions.",[22,591,592,593,596],{},"Entropia was designed to support M&A professionals with AI, and that design extends beyond our own walls. A user connected from Model ML gets ",[212,594,595],{"href":243},"the same tools"," as Entropia's own AI, at the same depth.",[22,598,599],{},"Bankers working in Model ML can reach their Entropia data rooms without leaving it. From Model ML, a banker can search the data room, read documents, presentations and spreadsheets page by page, browse the folder structure, follow Q&A threads, check who has access to what, and act: create folders, tag documents, set permissions, draft Q&A answers that a human reviews and sends.",[22,601,602],{},"Each request is initiated by the user, scoped to the access rights they already hold, and logged in the data room audit trail under their name. Nothing reaches a Model ML user that they could not open in Entropia directly, and nothing reaches a counterparty from Model ML: a drafted answer is reviewed and sent by a human in Entropia, and the access of an active group in the data room cannot be changed from outside it.",[165,604,605],{},[22,606,607,610],{},[15,608,609],{},"\"Our partnership with Entropia is a meaningful addition to the Model ML platform for our M&A users. Access to Entropia's agentic data rooms directly in the Model ML platform means that our users can harness the full power of Entropia's agent-native platform without disrupting their existing workflows,\""," said Chaz Englander, CEO at Model ML.",[22,612,613],{},"The integration is available to joint customers with active Entropia data rooms, starting today.",{"title":179,"searchDepth":180,"depth":180,"links":615},[],[617],"Chaz Englander","2026-09-10","Bankers working in Model ML can now search, read and act on their live Entropia data rooms without leaving the platform, under their own access rights.","\u002Fblog\u002Fmodel-ml-partnership-cover.png",{},"\u002Fblog\u002Fen\u002Fentropia-model-ml-partnership","Model ML partnership",{"title":574,"description":619},"blog\u002Fen\u002Fentropia-model-ml-partnership","Bankers working in Model ML can now reach their live Entropia data rooms without leaving the platform.",[294,199,295],"RrYTYx7Q0M0_Gs5IBSEzgqXbk-fckPrjcrUqIT3yuA8",1794484432271]