You explain something once — a decision, a correction, the way you like things done — and after that it's simply true. Your agent starts every session already knowing it. Every prompt, the matching memory is pulled in before the answer. You never explain your project twice, not to any agent, not on any machine.
Free tier · no credit card · setup in 60 seconds · recall is never metered or capped
There's a reason most "memory" tools end up unused. They're a drawer: you have to remember to open them. You have to recall that the drawer exists, and that the thing you need is in it, and then you have to make the model's next step be "go look in the drawer." That only works when you're diligent enough to ask — which is exactly the moment you won't. Brethof-brain is not a drawer. It does the work itself, whether or not anyone remembers to ask.
When a session starts, your agent is already briefed — the project's rules, where things stand, what was decided last week. And on every single prompt, the memory that matters is pulled in before the first word of the answer. Your agent spends no tool calls, no turns and no hunting. The memory is simply there when it answers.
After every exchange, a curator reads the turn and rewrites your memory. Decisions saved. Changed facts folded in. Claims that reality has overtaken — deleted. It happens in seconds, not sessions. You never file anything, and you never organize anything.
A curator watching one exchange can keep a single fact current. But it can't see the whole. Once a week, a healer reads all of your memory at once and does what no single-turn view can: merges records that forked apart, absorbs duplicates written months apart, and reconciles the places where your memory disagrees with itself.
In March you killed a solution. It was the right call, and you explained why. An agent steering toward it again is stopped mid-prompt. The reversed decision stays in the graph, flagged as a dead end, so the same dead end is never proposed to you twice.
You said "I'll follow up on that" and you meant it. Then the session ended and it was gone. Brethf-brain watches for those intentions. Every promise you make becomes a row in the Ledger, waiting for you. Say "check the ledger" and your agent walks the rows with a verdict already attached — done, with the evidence; pointless now; still worth doing — and you answer in single words. Fifteen rows takes about two minutes. The ones you keep become goal briefs your agent runs to completion on its own. You review in the evening; the work happens while you sleep.
The deploy, the translation run, the weekly report. Write the procedure down once, as a markdown playbook, and it's stored for your whole account. Any of your agents, in any project, on any machine, can fetch it and run it. Projects are separate; procedures are shared.
Most memory systems work by extraction rules and filters — a model is told what to keep, what to drop, how to file. That's caging it.
This memory is run by a model taught the way you'd teach a new hire: this is what's true, this is what changed, this is what we decided. It grows into the shape of your work instead of being fitted with a shape you have to maintain.
Don’t gate models. Teach them.
How we build everything at Brethof AI.
Summaries are what you have when you can't afford to keep the original. The summary is good until the detail you didn't think to keep is the detail you needed. This memory doesn't summarize. It keeps the complete transcript of every session, and it makes it searchable. The curated memory is the current truth — dense and up to date. The conversation history is the complete raw recording. Notes are fast but lossy; the recording is the insurance.
This is a single plugin with nothing to run — no Docker, no database, no config files. The client is a few hundred lines, and you can read every one of them. That's how you verify for yourself what leaves your machine: your messages, your agent's replies, its thinking, and one-line markers saying which tool it used. The contents of files it read and the output of commands it ran never leave. If it can't reach the service, it fails quietly and your session carries on. The one thing it will not be quiet about is a rejected key, because silently stopping would mean losing history without knowing.
Read →Projects are just names. Say one when saving or searching, and it exists from that moment. No setup, no limit. Your app's memory is just another project. Rules are the one thing that loads into every session automatically. Set them for everything you do, or for one project, and they shape how your agents work.
Read →Install. You need no IT knowledge. None. Tell your agent "install brethf-brain" and it's done: it fetches the client, wires the hooks, and checks its own work. Every other developer tool hands you a wall of commands. This one you ask for. To verify, ask your agent to search memory for anything. A well-formed answer — even "no results" — means it's connected.
Install →Plans meter only how much your memory holds and how much it learns each month. Searching, session briefings, per-prompt recall and graph lookups are never metered and never stop at a limit — on any plan, including the free one. Your stored memory keeps answering.
See plans →The free plan needs no credit card. Launch prices are founding-member prices, locked for as long as the subscription stays active. And you'll see where you stand right now — what you've used, what your plan allows, what's left — with an email before you reach the limit you set. Nothing happens to your account quietly. Get started free
Local speech-to-text that learns your voice. Perpetual licence. Our flagship.
PAID · flagship
Long-term memory for your AI agents — full-text + vector + graph. Briefed at session start, recalled on every prompt, archived automatically.
PAID · free tier
Print-ready digital models. GLB and OBJ included. Lifetime access.
PAID · digital catalog
Five printers and real capacity. No self-serve shop yet — tell us what you need and we quote it by email.
BY ARRANGEMENT · email us
Our YouTube channel. A cyber-tiger host walks through local AI tools and what they actually do.
CHANNEL · live
Curated GitHub lists for AI coding agents, MCP servers, local AI and Linux for AI. Every entry carries a source link.
FREE · curated
Long-form how-tos for local AI on Linux, Windows and macOS, with the configuration files included.
FREE · live
Negative-curation: practices and tools that waste your time, ranked. Receipts required.
FREE · live
Who we are, why we build privacy-first AI, and what we won't do.