Gavin Slater

Independent AI consultant. Thirty years in risk, data and finance; now building AI agents.

ESHER, SURREY

30+

Years in risk, data and finance

5

Client builds this year

25

Personal builds in the last year ›

1

Personal AI Assistant, in daily use ›

What I'm building · 2026

A selection of projects showcasing client work and personal projects shaped by my experience and first-principles approach.

Shared investment platform for six family offices

In Production

Client work

Automated ingestion of investment opportunities, routing each document to the right one, and an AI assistant pipeline for research, sharing and collaboration on deals. Used daily by the team and its partners.

Family offices

Six

Hosted

Google Cloud

Next.js · FastAPI · GCP

Brand consultancy agent

In Build

Client work

AI agents that free creative professionals to focus on creative work. Built from the founders' interviews, so it answers in the agency's voice.

Built from

Founder interviews

Built with

Claude · MCP

Claude · MCP

Personal AI Assistant

In Production

Personal AI

The AI infrastructure I run my own life on: a vault for memory, hooks that capture every session, a nightly reflection and a dashboard.

Daily notes

216

Services

About twenty

Claude Code · FastAPIRead more ›

AI-assisted private markets syndication platform

In Production

Client work

Helps corporate finance firms manage their relationships, with intelligent collaboration to build sell-side syndicates with other firms.

Built with

TypeScript · Gemini

Hosted

Google Cloud

TypeScript · Gemini · GCP

AI training course

Completed

Teaching

A weekly course taking non-technical users from playing with chatbots to building their own personal AI assistant to manage their daily lives. All sessions are free on YouTube.

Format

Weekly, 90 minutes

Published

YouTube

Claude · MCP · YouTubeWatch on YouTube ›

The Legible Organisation

In Development

Data

Organisations are built around the limits of human attention; AI removes the limit. After four golden-source attempts at four banks, a working system built by agents, where the record of what the organisation does is the thing that runs.

Change sets

20

Views in production

144

Agents · SQL · MCP

How I work

People first, then the model.

i.

Interview

I talk to the people whose judgement the agent has to carry, and read the documents they actually use.

ii.

Build small

One skill that saves real time, in front of real users within weeks rather than a programme.

iii.

Teach

The team learns to change it themselves, so it doesn't depend on me.

Taking on new engagements.

If you need someone who understands both the business and the bits.