Philippe Reynier

Co-founder & CEO @ Cigno. The AI-native core consulting system. Ex BCG · Ex UBS.

[Philippe Reynier @PhilippeCigno]

@brian_armstrong @coinbase Great example. Frontier models as generalists with specialist open-weight models on narrow tasks, working side by side, feels like where a lot of company value will be built.

[Philippe Reynier @PhilippeCigno]

@ClementDelangue Great call. Open environments let the whole community build on each other's work, and the pace at which open source picks things up is impressive.

[Philippe Reynier @PhilippeCigno]

@GoWithNoah Nice take. Deep distribution and embedded workflows across the wider legal market are a real advantage, and a lot of the value in legal AI will come from the firms outside the Am Law 100.

[Philippe Reynier @PhilippeCigno]

@DavidLinthicum Well put. Model-agnostic design is what makes that choice real, because teams can swap models and vendors without rebuilding the workflows around them.

[Philippe Reynier @PhilippeCigno]

@OpenMedFuture @openEHR Great point. Models moving onto personal devices makes open weights even more important, since people keep control of their own data and tools.

[Philippe Reynier @PhilippeCigno]

@ryanmckeen Love this. Putting your own draft next to the machine's turns AI into a mirror for your judgment, which is a great way for associates to build real skill.

[Philippe Reynier @PhilippeCigno]

@mardehaym Great list. Starting from the workflow people already complain about, and agreeing early on what a correct answer looks like, makes everything after that much easier to measure.

[Philippe Reynier @PhilippeCigno]

Some firms have built a fully sovereign AI setup. Others run 2: one for regulated engagements, one for the rest.

Either way, the AI setup you choose decides which clients you can serve, and which ones you cannot.

[Philippe Reynier @PhilippeCigno]

During a recent briefing on our State of AI in Consulting report, the managing partner of a consulting firm told me sovereign AI makes no sense. 🧵

[Philippe Reynier @PhilippeCigno]

She asked what I thought. My answer: it does not really matter. A client who invested hundreds of millions in a sovereign setup will not give up on sovereignty the moment they hire a consulting firm.

[Philippe Reynier @PhilippeCigno]

@skupor Specialized tasks on proprietary data are where open weights really shine. It rewards the teams who know their domain best and keeps the choice of model open.

[Philippe Reynier @PhilippeCigno]

@jerryjliu0 Defining the right evals becomes the real craft. That makes domain expertise the scarce input and favors vertical specialization over generic models.

[Philippe Reynier @PhilippeCigno]

@keithsalins_ @JoshuaKushner @sequoia @ycombinator Great resource. The services firms pairing deep domain expertise with model-agnostic AI delivery look like the most interesting ones to watch.

[Philippe Reynier @PhilippeCigno]

@jkubicki Testing across models inside real workflows is the best way to find out. Partners, associates and engineers building together means the firm keeps the know-how whichever model leads next quarter.

[Philippe Reynier @PhilippeCigno]

@rohanpaul_ai Routing narrow jobs to the cheapest model that passes is where a model-agnostic layer really earns its keep. The orchestration above the models looks like the durable part.

[Philippe Reynier @PhilippeCigno]

@litigationai Great point on where the capital is going. The good news is the layer above the models keeps getting cheaper to build, so small firms that shape workflows around their own practice can move faster than any vendor tier.

[Philippe Reynier @PhilippeCigno]

@jaypmo Experienced judgment plus better tooling is the part that compounds. The model is the commodity, the practice built around it is where the value sits.

[Philippe Reynier @PhilippeCigno]

@PamirEhsas Congrats to the whole team. Pairing real legal expertise with AI-native delivery is such a strong model, and it shows what happens when the people who know the work design how it gets done.