Meeting Notes — 2026-04-16

Participants: Max, Maha, Reiner (joined later) Duration: ~58 min Video: YouTube

Context

External signal from AI entrepreneurs/investors received around this call:

• Zurich pilot demonstrates a viable product — AI accompanying live conversations and intelligently intervening in real time.
• "Live Companion" concept (AI prompting participants to elevate dialogue quality) called out as a massive value-add.

Strategic direction

• Compliance product dropped. Commodity/scale territory owned by cloud giants; wrong fit for a tailored, niche team.
• Product reframed: AI working on actual conversations (record, rehearse, improve) — not roleplay in "cloudy nowhere-land" like generic coaching bots.
• Zurich = low-hanging fruit, not the whole market. General case: "conversations in defined systems" (organisations, team meetings) — predictable, repeatable structures. Zurich is special because NPS is already measured and conversations are already treated as productive capacity.
• Matches external investor framing: Zurich pilot = already-viable product.

Demo 1 — AI Call Assistant (live companion)

Max shared a working prototype running on the live call.

What it does today

• Real-time transcript with speaker diarisation (distinguished Maha and Reiner on a single mic).
• Tracks intentions per speaker.
• Tracks agenda items, ticks them off as covered.
• Suggests next moves to the host.

Validation so far

• Used on a real work meeting — valuable as a catch-up aid when attention drifts.
• Tested on a YouTube mock-interview video.

Open issues

• Recording etiquette / compliance (Chatham-House-style "Speaker 0/1" mitigation possible).
• UI rough.
• Notion has shipped a similar feature — competitive pressure rising.

Expansion ideas (Maha)

• Per-participant agent ("Reiner, you've been silent 5 min — bring a creative thought").
• "Dojo" for practising better conversations: teams legitimately record meetings, each member privately replays and improves.
• Chairperson/secretary-specific functions (decisions taken vs. decisions still vague).
• Differentiation required: Zoom / Teams will enter this space — "art of dialogue" slant is the moat.

Maps directly to the investor-flagged "Live Companion."

Demo 2 — Agent-based modelling

Feed text → system extracts actors → builds persona graph → simulates their conversation → produces report → lets you chat with any agent or poll the group.

Cases tested

• MiCA regulation (49 actors: central banks, journalists, owners, regulators, incl. Polymarket-style private vs. public stance modelling).
• Personal scenario: 20 people from Max's network reacting to "pivot from ecological to fintech?" — including a family member replying about spring weather (plausible real-life noise).
• Earlier Iran-US test blew up (~1000 actors — too expensive to simulate).

Framing: Max is reading Farmer's Making Sense of Chaos — bottom-up modelling of complex systems as the underlying lens.

Business applications (Maha)

• Convector — EU chemical-directive lobbying. Model the directive process, map coalition options, strengthen the proposal.
• Convector 360 assessment — use cross-evaluations from 25 leaders to project "who they'd be together in 5 years"; reframe proposal as an ongoing leadership modelling base.
• Phoenix Consultancy — leaders preparing for difficult conversations; presentation opportunity this evening.
• Unified leader tool = individual trajectories + strategic simulations.

Open risk (Max): any predictive claim needs backtesting on known past events before it's credible.

Other threads

• Meetup: blocked on lining up 3 speakers; date follows.
• Business connector (former founder of an early entrepreneurship network) — introduced via a mutual contact. Meeting Monday; possible meetup partner, building a new AI-included business club.
• Banker (blockchain in banking) — Maha offering warm intro Friday.
• Trade Republic signal: call centres shifting from chatbots → trained humans. More "Zurich-shaped" prospects in financial services.
• Family-dynamics conversations flagged as a longer-tail segment.

Tension to resolve

Investor enthusiasm points at the live companion (Demo 1). Near-term deal flow points at modelling (Demo 2, via Convector and Phoenix Consultancy). These are not the same product. Decision needed on where to double down — external pull vs. near-term revenue.

Action items