AI doesn't fix how a company works. It amplifies it.
We build founder-led teams a single source of truth their people and their AI both run on.
We run on it
Our own company runs on the exact thing we build, updated every day. We don't sell what we wouldn't use.
It does real work
We've put our own AI tools to work on routine tasks: pick up a job, do it, hand it back to check. Working now.
Built in the open
The wins, the dead-ends, the decisions. Built where you can watch, not behind a curtain.
The knowledge that runs your company isn't written down. It's in people.
How the business actually works lives in your head and a few senior people, and it got there because you were busy building, not documenting. That's normal. It's also the wall. New hires take months to absorb what the experienced people just know. Everything important routes through the few who hold it.
Then you add AI on top, and it runs blind, with no memory and no context, starting from zero every time. So it hands back confident, generic answers that miss how you actually operate.
The problem was never the AI. It's that nobody ever gave it anything real to work from.
Onboarding
months, not days
Key knowledge
in a few heads
AI's answers
generic, every time
One memory your team and your AI both run on
What's true now
The current state of how the business actually runs. One trusted answer, not ten stale docs disagreeing with each other.
Why it changed
Every decision, the dead-ends, the paths you rejected and the reason you killed them. The thinking survives, not just the outcome.
The evidence underneath
The calls, the notes, the raw detail the rest is built from. Every answer traces back to where it came from.
operating-memory/
├─ artifact what's true now
├─ ledger why it changed
└─ source the evidence
One source. Your team and your agents both read it.
Context before agents. Every time.
Everyone's racing to add AI agents. Almost nobody's doing the unglamorous thing that makes them work: building the context they run on. The boring layer everyone skips is the whole game. That's what we build.
We build your operating memory, hands-on, in two steps.
- Retainer rateoverview says £3,500 · now £4,800HIGH
- Primary contactoverview says Tom · now PriyaHIGH
- Plumbing IFC exportblocks Estimating Assistant 22/25MED
Drift surfaced against the decision log.
Get your biggest bottleneck off your desk
We build the operating memory for your single biggest bottleneck, the area where everything routes through you, plus an assistant your team can ask. About 2 to 3 hours of your time, and you feel it the first week: the questions that used to land on your desk get answered without you.
If it's not worth what you paid, full refund, and you keep everything we built.
From $1.5k · credited in full to the build
Expand it across the company
We extend it across more of the company, role by role, with the assistant answering real, company-specific questions. We agree the exact questions it has to answer before we start.
If it can't answer them, we keep working until it can.
- 04-20 retainer £3,500 → £4,800 (scope grew).
- 05-30 Inv #8 (£4,800) issued · outstanding.
- 06-09 Estimating Assistant acceptance 22/25.
Every answer reconciles from here, not a stale summary.
We run our own company on it first.
This isn't a slide deck. Our own company runs on the exact thing we build, updated every day. Ask the sample company a real question and it answers what's true today, traced back to the actual decision behind it, where a generic AI tool hands you a number that was true six months ago.
We also let our own AI tools handle routine work: they pick up a job, do it, and hand it back to us to check.
See it run live → ProofOverview
Where Meridian stands — today
Monthly retainer
£4,800/mo
was £3,500/mo · click → real source
Who this is for
- ✓You're founder-led, roughly 6 to 15 people, profitable and growing.
- ✓The knowledge that runs the place lives in you and a few experienced heads, and onboarding is slow because of it.
- ✓You believe AI matters, but it keeps handing you generic answers, and you suspect it doesn't actually know your business.
You're right. That's the part we build.
Everything you'd want to know before a call.
No pitch needed. Here are the straight answers, including the ones most people don't put on a website.
What you actually get
How is this different from a wiki or Notion we'd just stop updating?+
Two ways. We build it from how you already work instead of asking you to write it, and we set up a small routine to keep it current. The SOPs you abandoned died because updating them was a second job. This is built to avoid that.
Your time and effort
How does what's in my head get into the system? Do I have to write it all down?+
No. You talk, we write. We record short interviews, pull in what already exists in Slack, email and docs, and capture how the work actually gets done. People will talk for an hour. Nobody writes for one.
Who keeps it current after you're gone?+
We set up a light routine so updates happen as part of the work, not as extra admin. Honest answer: nothing stays current by magic. A small deliberate habit beats the wiki everyone abandons, and that's what we build for.
What it costs
What does the Build cost?+
Builds start at $3k. The exact number is scoped in the first step, so you see a real figure before you commit a cent to the Build. No surprise quote at the end.
Trust and risk
You're one person. What if you get sick, move on, or AIL shuts down?+
Fair, and it's why none of this leaves you dependent on me. The first step delivers a working layer in tooling you own, yours even if we never work together again. The Build lives in tooling you own and can export. If I disappear, you keep everything, in a form your team can run without me.
Your data
I'd be handing over my most sensitive knowledge. How is that handled?+
It's set up in tooling you own, and access is scoped by role, so each person, and any AI, only sees what's relevant to them. Your knowledge is not used to train public AI models. You own the data and can export it at any time.
See it work.
Talk with us. 30 minutes on a call, and we'll show you the operating memory running live on a sample company, and we'll find where your business is losing what it knows. No deck, no homework, no pitch.