Every fund now has an AI paragraph in its annual letter, and every portfolio company has been told to adopt it. Inside the firm, the picture is usually thinner: associates triage inbound teasers by hand, quarterly reporting is a copy-paste exercise across the CRM and a spreadsheet, and LP questions are answered from memory and old decks.
Generic AI courses do not fix this, for a reason specific to funds: the work is confidential and system-bound. A prompt course cannot touch your deal flow, your LP data, or your portfolio reporting — so it teaches tricks, and the real work stays manual.
AI training for private equity, fund operations, and family offices
Specialty Tokens runs AI training as the entry tier of our engagement model. Before the workshop, we connect the AI tools your firm has approved — ChatGPT, Claude, Microsoft Copilot — to the systems where the fund's knowledge actually lives, through the Model Context Protocol (MCP). The exercises are your own work: screening an inbound teaser against your thesis, answering a portfolio question with sources, assembling the factual sections of an IC memo, preparing the quarterly report.
The program runs in two tracks. Partners and principals get the judgment questions: where AI carries weight in the fund, what it costs, what to govern, and what to leave alone. Fund operations, finance, and IR get hands-on time: the people who run reporting, reconciliations, capital calls, and LP communications build automations they keep.
The same format serves the wider table — family offices with consolidated reporting and document-heavy workflows, hedge funds with research and middle-office load, and the fund administrators around them. The training runs where the team already works: for Microsoft houses, agents inside Microsoft Teams and Copilot — see Microsoft Copilot consulting — and for Slack firms, the same pattern in Slack. The private systems every fund runs on — portfolio monitoring, fund administration, proprietary models and databases — are reachable through custom MCP connectors built for your stack, inside your perimeter.
Security is the ground rule, not a slide. Deal flow, LP data, and portfolio numbers stay inside your perimeter; nothing trains public models; reads are free and writes queue behind a named approver, logged. Compliance is welcome in the room.
We have sat on your side of the table. Thomas Schlossmacher worked at a family office (Corecam) and built revenue-operations software for ERP systems; Nik Redl built financial infrastructure and scaled products past a million users. Both have served companies on financial operations where security was a requirement, not a feature. It is the same pairing behind our work with Speedinvest, the pan-European venture firm, and Talentir, a financial-infrastructure company — engagements about making a firm's own data usable by the AI tools it already trusts.
We help investment firms go AI native — from a first workshop, through advisory for partners and CFOs, to agents in production via AI agent development and embedded AI transformation where the mandate is bigger. We are based in Vienna and train firms across DACH, the UK, and the rest of Europe, in English or German. For the finance function itself, see AI workshops for finance teams; for the general format, AI workshops & training.