We don’t pitch the future. We hand you the things we already built in it.
LFG Consulting is a small studio with one founder, conducting a battery of frontier models in a truly orchestrated fashion. Three layers of one stack, each one built before it was offered.
Fig. 1 — A site is a seed we project. The same instrument reads both ways: grown from a conversation, or measured from the site you already have.
Three layers of one stack.
Everything we ship is built on the same instruments we point at ourselves. Agentic Kernel is the platform. Stamhus and the Math Solver are products built with it. The applied work for partners is built on all of it. When one layer gets better, the layers above it do too.
Agentic Kernel
A platform for LLM-driven orchestration and code generation. Specialised agents plan, build and check their own work against a shared knowledge fabric — the same fabric every agent reads, so the checking is done against what the whole system knows, not what one prompt remembered.
Fig. 2 — Orchestration, not a chain of prompts. The loop closes on the fabric.
Stamhus
Websites coherent by construction. A site is a seed we project. It can be grown from a conversation, or measured from the site you already have — either way, every page derives from the seed, so the whole site holds together by construction rather than by upkeep.
Fig. 3 — Two ways in, one seed, one derived structure.
The Math Solver
A tutor on the phone that walks a student through their own exercise, step by step: it checks, hints, and waits. The product that made the Kernel necessary, and is now built with it.
Fig. 4 — The student’s own exercise, worked on the phone.
Fig. 5 — What we build, connected to what you already run.
Applied work
Bespoke software, AI automation and orchestration, finding what is worth building, and your existing systems connected. Built for partners on the same instruments we point at ourselves — which is the reason a lab this small can take them on.
One founder, a battery of frontier models.
The work is a conversation that doesn’t stop. On one side, the founder: direction, architecture, and the call on what is good enough to ship. On the other, the models: drafting, refactoring, exploring ten approaches before lunch. Neither side does the other’s job.
Fig. 6 — Every line runs both ways. The founder holds the score; the bank plays it.
Founder’s side
- Direction
- Architecture
- The call on what is good enough to ship
Models’ side
- Drafting
- Refactoring
- Exploring ten approaches before lunch
The years behind it.
The work, and the decades behind it.
What I have built, what I build now, and what it runs on.
Start with a conversation.
Write and say what you are building, or what has to change this year. Say what you already have and what you already know. That is enough to start.
Or the long way round: /contact