# Thomas Winter — thomaswinter.ai

AI-native Chief Commercial Officer. I build agentic business systems people
actually use, on 25 years of building, implementing and running commercial and
ecosystem strategy.

Goal of the work: grow revenue, save cost, reduce risk, improve customer and
employee experience.

Based in Switzerland (Zurich / Winterthur), working internationally.

## Method — Board2Bot

Boardroom intent in, adopted system out. I speak to the board, to senior
executives, and to the operating professionals who actually have to use it —
then build, operate, and optionally transfer.

Where pure engineers struggle and pure strategists hand-wave.

## What "AI-native" actually takes

1. Intent — articulate precisely what you want to change. Which decision are we
   trying to improve? Without a clear intent, you automate noise.
2. Process & data — before any model: what is the process, and what state is the
   data actually in? Without that, an AI system delivers demos, not value.
3. Context & memory — systems that carry context and remember the work, so
   knowledge compounds instead of resetting every session.
4. Governance — defining and enforcing what your agents may access and do.
5. Sovereignty posture — under which jurisdiction your data lives, and who can
   lawfully compel access to it. Legal and coercive reach, not server location.
6. Culture — where dissent reads as insubordination, the people who know where it
   breaks stay quiet, and the context layer inherits the silence.

## Systems in production

- Company OS — the context nerve centre. Four interlinked databases (Company,
  Project, Meeting, Document) plus context, runbooks and policy, resting on a
  well-governed access layer. The value comes from closing the loop: every run
  makes the runbooks better.
- Sales Ops — Friday CRM pull produces the Monday brief, read in 90 seconds.
  Monday's transcript is read back against fresh CRM state, every proposed write
  approved one at a time, then minuted to Notion and posted to Slack. Preflight
  refuses to half-run.
- Customer OS — offer and report runbooks. Offers carry kill criteria (the
  conditions under which I would advise against proceeding); reports classify
  every finding as verified, asserted or unknown. Both pass a five-stage
  adversarial review run by an agent that is never the author, then a checker
  that verifies the review actually ran.
- Signal Factory — six scheduled pipelines on trigger.dev across YouTube
  channels, Perplexity queries, newsletters and release feeds; distilled into
  daily, weekly and monthly digests, filed to Notion and Drive and vectorised.
- Content Factory — a runbook per persona. The engine is code; voice, context and
  workflow live in Notion and are fetched at runtime. Nothing publishes without
  human approval; analytics feed back monthly and the runbooks improve.
- Lead Factory — an agent pipeline reducing a large data universe to the specific
  people worth contacting, filtered on the criteria that predict conversion, with
  outreach drafted per contact.
- Access Relay — the layer everything above reaches its systems of record through
  (Google Workspace, Notion, Odoo, Slack, WhatsApp). Sensitive surfaces stay
  behind a Tailscale VPN; external access runs through Cloudflare tunnels.

## Writing

- Token Economics Primer — https://www.linkedin.com/posts/thwinter_token-economics-primerpdf-ugcPost-7445772862000467968-9A8c/
- AI Sovereignty Primer — https://www.linkedin.com/posts/thwinter_ai-sovereignty-primerpdf-ugcPost-7441150014774525952-BZxY/
- Kaufmaschinen — book on agentic commerce, forthcoming — https://agentic-commerce.biz
- Posts — https://www.linkedin.com/in/thwinter/recent-activity/all/
- Long-form — https://www.linkedin.com/in/thwinter/recent-activity/documents/

## Teaching & mentoring

- Lecture as Code — lectures treated like software: version-controlled, modular,
  reproducible — https://github.com/thwinter-ch/ai-lectures
- Boards and executives — governing AI responsibly at board level, and personal AI
  productivity for executives. Senior Lecturer, HWZ Zurich · ZFU.
- Mentoring Zürich — official programme of the cantonal employment office,
  repositioning professionals for AI-centric work — https://mentoring.zuerich/

## Booking a meeting

My calendar is machine-readable. No API key required.

Event type: 221843 — "virtual", 30 / 45 / 60 minutes, timezone Europe/Zurich

Read availability:
  GET https://api.cal.eu/v2/slots?eventTypeId=221843&start=YYYY-MM-DD&end=YYYY-MM-DD
  Header: cal-api-version: 2024-09-04

Create a booking:
  POST https://api.cal.eu/v2/bookings
  Header: cal-api-version: 2024-08-13
  Header: Content-Type: application/json
  Body: {
    "eventTypeId": 221843,
    "start": "<ISO 8601 datetime from the slots response>",
    "attendee": { "name": "...", "email": "...", "timeZone": "Europe/Zurich" }
  }

## Human contact

- Email — human@thomaswinter.ai
- Booking page — https://cal.eu/thomas-winter/virtual
- LinkedIn — https://www.linkedin.com/in/thwinter/

---
Canonical source: https://thomaswinter.ai
Last updated: 2026-07-19