Answer before they file
Residents search everything you've already released, in plain language — and you see the requests that never had to happen.
A FOIA platform for the AI eraFewer requests. Faster responses.
Fewer requests. Faster responses.
FOIA made more intelligent.
Most requests ask for something that's already public. Brandeis answers those on the spot. The rest arrive sorted, assigned, and on your state's legal deadline — AI does the drafting, your staff make every call.
51
public-records laws built in
100%
of actions logged, people and AI alike
0
outside services needed
Residents see what's already public before they file. Staff get requests that arrive organized.
Sunlight is said to be the best of disinfectants.
Louis D. Brandeis, 1913 — the namesake, and the job description. What's already public should answer itself. What isn't should leave only by a person's decision, on the record. That's the whole product.
The problem
More requests. The same one person. Still by hand.
Three trends, one desk.
Staffing hasn't kept pace
Most offices are one person, wedged between other duties. Volume climbs. Headcount doesn't.
AI is driving the volume up
AI tools now write many of the requests — broader, faster, and around the clock.
Most of what's public was never published
The answer is usually already public — just nowhere a resident can find it. So a question becomes a case number.
How we help
Three jobs. One system. No new headcount.
Most records software does only the middle one.
Run every request on rails
Requests arrive sorted and assigned, departments reply from a link with no login, and the deadline comes from your state's law.
A record that stands up
A permanent log no one can edit, redactions that can't be lifted off the page, and every release approved by a named person.
What actually changes
Teams can spend time on what matters.
The same law, a different day at the desk.
How it's always worked
With Brandeis
Every question becomes a case number and a wait.
Most questions get answered on the spot, from records already public.
Staff read every page for exemptions, by hand.
AI flags what the law may let you withhold, citation attached — staff confirm each one.
Deadlines live in a spreadsheet and slip in a busy week.
Every deadline comes from your state's law, and every clock is checked nightly.
A password is all that stands between the internet and the case files.
Every staff sign-in takes a password and an emailed code, and every sign-in is logged.
“Sent” is the only record of what happened.
Every action is logged permanently, and anyone can check that a released file is genuine.
The express lane
Routine requests, one click. Still a named human.
Write the policy once. The lane does the markup.
Your policy, written once
Pick a routine record type, like incident reports or code complaints, and say what a public copy hides.
Arrives ready to send
Matching records open already redacted under the policy, with the letter drafted.
Checked after it ships
One release in ten gets a second reader, and a miss goes straight back to the policy.
City of Riverton · Release reviewExpress lane · code complaints v1
City of Riverton · Code enforcement case file
- Property owner
- Alder Street Holdings LLC
- Complainant
- Address
- Phone · email
- Inspection
- Inspector R. Vance, . Vegetation over 12 inches confirmed; notice of violation issued.
One click by a named human. One release in ten gets a second reader afterward.
The agent era
AI files requests now. Brandeis answers them.
More requests every year are written by AI assistants. Work with them instead of drowning in them.
Same door, same limits
A resident's AI can search your public records and file for them, through the same gate and rate limits as your own form.
You see which requests a machine filed
Your reports split requests by how they arrived, so the AI share is a number, not a hunch.
Floods get caught
Near-identical requests from different people are grouped each night, and one released record can answer them all.
For the technical: every portal speaks MCP, so you can point an AI agent straight at it. Developer docs →
Under the hood
AI agents that work for you.
One at each stage of a request. Each drafts; your staff decide.
Records Check
Answers residents from the records you've already released, with the document linked.
Sees only public records. Nothing else.
Staff Router
Sorts, scopes, and routes each request to the right desk the moment it lands.
A coordinator accepts, edits, or discards.
Auto-Redact
Proposes redactions line by line, scanned pages included; approved ones are sealed into the page for good.
Every redaction confirmed by a person, then checked for leaks.
Hot Spotter
Spots what residents keep asking for and points at the records worth publishing next.
A person decides what gets published, record by record.
Exemption flags, drafting, a staff copilot, and appeal packets ride along too. Every agent action is logged, and with no AI switched on at all Brandeis still runs. How the guardrails work →
A modern interface for residents.
Ask a question, search the archive, track a request — no account, no form to decode.
Ask in plain language
The portal answers from what's already public, links the document, and only drafts a request when it has to.
Search everything released
Every released record lives in a public archive that's searchable, dated, and verifiable by anyone.
Track without an account
Status, deadlines, and messages from a link, plus a public log of every request the office has handled.
What it costs
Your first 1,000 requests are free.
Every office deserves the good version, whatever its size. After the first thousand, you pay for exactly three things:
Usage
The requests you run through the system. No seats, no modules.
Document storage
The records you keep, publish, and serve.
Hands-off help
Optional — our team takes on whatever desk work you'd rather hand off.
Not a procurement
Bring us your worst backlog.
We open your records office with you — and the first 1,000 requests are on us.
Thirty minutes on the real product — your state's rules, your backlog, no slides. The about page has the thinking behind it.