say hi_
path: ~/portfolio ↑ up to ~

products & services · outbound shelf

The shelf points out.

Four products, all working names, all early. Each will move to its own name and home when it launches - nothing gets merged into this site, so nothing gets diluted. Here is the point of view behind each one.

the products · honest labels, working names

topic-modelling

private beta

A service and toolset for finding structure in large piles of text. You hand it a corpus; it detects which documents are closely related, classifies them automatically, and gives you discovery tools to browse what it found. Built for data that keeps arriving - it works in real time, so a growing archive stays organised instead of rotting.

In private beta ahead of public launch. The one product where you can already leave your email to be in early.

Text-2-Action AI

in development

A lightweight, deterministic agentic stack. Agents detect intent from natural language - written, spoken, visual - and from the data around them, then execute actions. Deterministic means repeatable: the same input does the same thing, which is the difference between a tool and a toy. Agents are set up and trained around one problem domain and learn the real world as they work.

Where it fits: customer support, ticket handling, and the field-worker flows next door.

Field-AI-ssistant

in development

Natural AI communication for field workers in logistics, maintenance and construction - inside the messaging apps they already use. Chat with AI in a room: easy. A group of people: easy. WhatsApp, Telegram, Discord: easy. Images, threaded conversations, group chat: easy. Fully private, outside big tech: also easy.

Fleet-AI

in development

An agentic stack at scale, under the control of your company - not big tech. An accelerator that installs a full digital-product creation and software-lifecycle setup on one principle: you own your business, products and services yourself. Guardrails for an AI-powered organisation, auditability trails, no vendor lock-in, pluggable components, open or commercial models swapped at will. Runs as a service or integrated on-prem, in your cloud space.

The belief underneath: organisations that outsource their AI stack also outsource their judgement. I build the alternative.

the advisory work · what I take on

Two shapes of work

hands-on technology leadership

CTO or technology-director roles where I take an engineering organisation into AI-native delivery, or build a new AI-driven product with the team that will keep it. Decisions in the open, juniors who grow, standards that stick - and I stay in the code. That is the point.

interim and assignment work

Time-boxed engagements that set up the platform, the team and the way of working behind future AI products - then hand over. New technology gets honest verdicts: pilot it, ship it, or kill it fast. No demo-ware survives here.

One stance, said plainly: I keep the IP of concepts and ideas. Clients license it, or use the hosted service. That is also why there are no client logos on this site - what I build stays mine to build on, and yours to run.

One of these is your problem?

One line is enough to start. State the shape of the mess, I will tell you whether I am the right person for it.

say hi