Solutions · Intelligence
Each AI assistant your team uses forgets everything between sessions, and none of them knows what the others learned. We set up one shared memory of how your company works, kept in your own systems, read by every agent, and we run it for you. We built it for ourselves first.
- A team that uses AI assistants every day and re-explains the same rules, clients and decisions in every session.
- A company using more than one AI vendor, with what each one learned stuck inside it.
- An owner who wants AI work to follow the house rules: how we deploy, what we never say to a client, who approves what.
What we do
The work, in plain words.
- 01
One memory, in your hands
Plain text files, one fact each, in a private repository your company owns: readable by people, versioned, and portable between AI vendors.
- 02
The right memory at the right moment
Nothing is loaded wholesale. A small core of rules is always on; the rest is recalled by what the person is asking, and some memories are read at a moment, such as before a deploy.
- 03
Shared across machines and assistants
Every session saves what it learned and every new session starts from it, on every machine and across the assistants your team uses.
- 04
Secrets kept out
Each save is checked for keys, passwords and connection strings; a file that holds one stays local and is named, never committed.
- 05
An audit trail
Every change is a commit: who or what wrote it, when, and what it replaced. Outside writes that fail the checks are set aside for a person to review.
- 06
Run and tuned
We measure how often the right memory reaches a session and tune it, archive what is finished, and keep the setup current as the assistants change.
How it starts
Three steps, each in writing.
- Step 1
A conversation
Which teams use AI, which assistants, and what they keep having to repeat.
- Step 2
A first memory
We write the core rules and the facts that matter with you, and connect your assistants to it.
- Step 3
Run it
On the monthly retainer: sync, checks, tuning and the archive, looked after.
Proof
Running today, not a promise.
Our own company memory
Every AI session at Blue Coast Water, on every machine and across two assistant vendors, reads and writes one shared memory: recall by prompt, a sync after every turn, secrets held back, and an audit trail in git.
Private AI agents
Memory is what makes an agent useful on day two. We build agents into running systems.
Private AI agents →Build & Run
Sold the way we sell everything: set up inside your company and run, not a product to license.
The three ways to engage →
What we do not do
Where we draw the line.
- We do not sell a memory product or keep your memory on our servers. It lives in your systems.
- We do not put secrets in memory, and the checks refuse to.
- We do not claim it replaces documentation people read. It is written so both can.
Questions
What people ask first.
What is company AI memory?
A shared, written record of how your company works (rules, decisions, clients, systems) that every AI assistant reads before it acts and adds to when it learns something.
Which assistants does it work with?
We run it across more than one assistant vendor ourselves. Which ones fit your team is part of the first conversation.
Where is the memory stored?
In a private repository your company owns, as plain text. You can read it, edit it and take it anywhere.
How do you stop secrets leaking into it?
Every save is checked for keys, passwords, tokens and connection strings. A file with one is held back and named, never committed.
Does every session load all of it?
No. A small core is always on; the rest is recalled by what is being asked, so sessions stay fast and focused.
What does it cost?
Set up and run on a retainer, quoted by the founder after the first conversation, in writing.
No prices on this site: the founder quotes in writing after the first conversation. How engagements work.
Tell me what you need.
Philip Radvan, founder. Every message lands with me, and it's me on the call.