The home for brains
A brain for everything you know.
Alive, and usable by your AI.
Notes, projects, customers, decisions, contracts: yours or your whole company's, in one place your own AI can read, update, and work across. One brain instead of ten subscriptions. Your software stays alive.
The home is free.
Built on db.md, the open standard. Open files you own, never locked in. Bring your own AI.Or start a brain in your repo →Your knowledge is scattered, and your AI can't see it
Ten tools, each holding a slice of what you know. None of them talking. And the AI that could finally use all of it has nowhere to read it from.
“What did we agree with Acme?” is split across an email, a call recording, and a Slack thread, so nobody actually knows.
Your CRM is three weeks stale because updating it by hand is the chore everyone skips.
The one person who knows how billing works is on vacation, and the process was never written down.
How it works
One brain. Plain files. Your AI operates it.
Your AI doesn't connect to it through an API. It lives in it: plain text is what a model reads natively, so the brain is home ground for any agent you choose.
The brain holds everything you know
One place, in plain files: raw sources (emails, transcripts, exports) and the typed records your AI keeps current (customers, decisions, meetings). Readable text, not a schema you can't see into.
Your own AI keeps it current
Point your agent at the brain: Claude Code, Codex, whatever you run. The first session migrates what you already have. After that it keeps up, filing what's new from wherever your work lives.
Sevra keeps it alive
We keep the brain organized, indexed, and reachable: fast to query, ready for your whole team, and in time, connected to other brains.
Your agent pulls these in on a schedule. Sevra keeps whatever lands organized and fast to find.
One brain instead of ten subscriptions
Most business software is a database with a screen on top. Your AI builds those on the brain, to your exact spec, and rebuilds them when your needs change. The data is the asset. The software is a cache.
| Customer | Renewal | Terms | Note |
|---|---|---|---|
| Acme Inc. | Sep 1 | Net-60 | 12% uplift agreed |
| Borealis GmbH | Oct 15 | Net-30 | QBR due |
| Karyo Labs | Jan 8 |
Proof
A studio that runs on its brain
Northwind Studio is a demo brain we keep public. Its site is published straight from the records: the client list, a pricing decision, an onboarding playbook, an RSVP page with a working form. Both cards below are read from that brain, not drawn for this page.
Atlas Foods
Packaging across twelve SKUs and a storefront for a regional food brand.
Lumio
A B2B analytics startup. New brand, a design system, and a marketing site in eleven weeks.
Decision: price by outcome, not by hour
Why we price by outcome, not by the hour.
Client onboarding playbook
What our first week with a new client looks like, day by day.
Your AI works across everything at once
A siloed app can only answer about its own slice. Your brain has no walls inside it, so the agent reads across customers, contracts, calls, and decisions together.
The more your brain holds, the more your AI can do with it.
Who runs on a brain
One primitive, any scale
For you
A brain of your own. Notes, projects, the context of your work, in plain files, with an AI that actually uses them.
For your company
One brain for the whole team. Institutional memory, bespoke tools, and daily operations, in one place that compounds.
For agencies
A brain per client, run from one place. Resell brains to the clients you serve. Brains all the way down.
Built on db.md. Owned by you, not us.
Your brain is a db.md store: an open standard for keeping a database in plain files, Apache-2.0. Export it, read it in any editor, or self-host it any day. The neutral layer, never the lock.
db.md
The database in plain files. Sources, records, frontmatter, and the dbmd CLI your agent operates directly. The format every brain is built on.
Open standard · Apache-2.0 · live
link.md
The open standard for the interconnect: how brains address, grant access, and exchange. Your brain talking to your accountant's, an agency's to its clients'. The internet of brains.
Open standard · Apache-2.0 · v0 draft
The standards are neutral and free. Sevra is the managed home on top. Forking the format never forks your data.
Pricing
The home is free
A brain costs almost nothing to host, so the home costs you nothing: indexed, always alive, ready to share. Paid plans add three things: work Sevra runs for you, the organization around a shared brain, and human support. Never per person, never per brain. Your own AI is never metered.
Free
The complete home, for one person.
- A real brain, not a trial
- Always alive, fully capable
- Publish pages and apps, invite viewers
- 5 GB asset storage
Common questions
Everything you know, kept in one place your AI can use. For a person: notes, projects, and the context of your work. For a company: customers, decisions, processes, contracts, the knowledge that usually lives in people's heads and old chat threads. It's stored as plain files in the open db.md format, and your own AI reads, updates, and works across all of it.
The open standard your brain is built on: a database kept in plain files instead of a proprietary system. Sources hold raw material like emails and transcripts. Records hold everything your agent writes: typed facts like customers and decisions, plus the synthesis that ties them together. Every file carries simple frontmatter your agent reads and writes. It's Apache-2.0 and yours to keep.
Yours. Point Claude Code, Codex, or another coding agent at your brain and it does the reading, updating, and building. Sevra keeps the brain organized and reachable; you bring the intelligence, and your own agent is never metered. On paid plans Sevra can also run scheduled work for you, billed as run credits at cost.
Your agent does it, starting with what you already have. Point Claude Code, Codex, or any capable agent at a Notion export, an old wiki, or a folder of docs, and the first session turns it into a brain. From there it keeps up: connecting to where your work lives with the connectors those agents already ship, filing what's new on a schedule, on your own AI subscription. Forward, paste, or drop in anything by hand. On paid plans, Sevra can run that schedule for you instead. Whatever lands, Sevra keeps it organized, indexed, and fast to find.
No GitHub needed: the hosted home is free for one person, and Sevra keeps the brain organized and reachable. Technical users can also keep a brain in any Git repo, since the format is open and works anywhere. Either way you direct your AI in plain language. You don't write the files by hand.
A wiki is a place you maintain by hand and read with your eyes. A brain is maintained by your AI and built for your AI to use. It's structured so an agent can answer across all of it at once, and the data is plain files you own, not pages locked in someone's app.
Those remember your chats, in their cloud, for their model. This is your knowledge, in your files, usable by any AI you choose. It's the data layer, not the assistant. You can read it, clone it, and take it anywhere.
No embeddings, no black box, no separate index you can't read. The brain is plain files with frontmatter the agent reads directly, and you can read the exact same files. Recall comes from the agent searching and reasoning over real text, not from a similarity score you have to trust.
A lot of it, over time. The small tools that are really a database with a screen on top, your AI builds those on the brain, to your spec, and rebuilds them when needs change. The hard systems you trust for accounting, payroll, and comms keep their place. What changes is the long tail you used to rent one subscription at a time.
The home and what it publishes. Sevra keeps the brain organized, indexed, and reachable, and runs the surfaces the brain puts up: pages and small apps at your own address, sign-in for readers you invite, forms whose entries land in the brain as evidence, functions, and scheduled runs that keep records current. You bring the AI for everyday work. The brain itself stays open and exportable, yours to take anywhere.
No. Your brain is plain files in an open format. Export it, read it in any text editor, self-host it, or walk away with it any day. Sevra is the neutral layer on top, never the lock.
The open standard where brains connect: how one brain addresses, grants access to, and exchanges with another. Your brain talking to your accountant's, an agency's to its clients'. The v0 draft is public, and the hub already speaks it. db.md comes first.
The home is free: hosting, publishing, sharing with people you invite, for one person with a real brain. Paid plans add three things: work Sevra runs for you, organizations with members and roles, and human support. Sevra-run work is billed as run credits, sold at cost: every paid plan includes a monthly amount, and you can buy more anytime. Your own agent is never metered, and there is no per-person or per-brain pricing. Unlimited brains on every plan.
Teams & agencies
Get your company brain
Org-owned brains, roles, and agency setups are hands-on while we roll them out. Tell us about your company or agency, and we'll set up your brains and migrate what you already have. We take on a few teams at a time. Working solo? Start free above, no waiting.
Built on db.md, the open standard. Your data stays yours.