Verified legal research,
inside your AI assistant.
Connect Affine to Claude, ChatGPT or Microsoft 365 Copilot and your assistant stops answering from memory. It researches the legislation and case law for your jurisdictions, and shows you the documents behind every claim.
Works with
- Claude
- ChatGPT
- Microsoft 365 Copilot
- and more
One connection, nothing to install — and any other MCP-compatible assistant.
A confident answer is not the same as a correct one.
01Memory only
Answered from memory
Ask an assistant about tax, employment or property and it will answer fluently — from training data that predates the last amendments. Rates, deadlines and article numbers are exactly what a model remembers badly, and nothing on screen tells you which part to double-check.
Sources
None
02With a web search
Answered from web search
Turn browsing on and it does better — it finds something. But what it finds is what ranks today: a firm’s summary, a news item, a PDF of the original text with none of the amendments applied. Closer, and still not the law in force.
Sources
- “Fixed-term contracts — what changed”Law-firm blog post · undatedSecondary
- Lei n.º 7/2009 — original PDFAs published, no amendments appliedNot consolidated
03With Affine
Answered from the primary sources
The same question is routed to Affine, which runs a full research pass over the public legal databases for your jurisdictions. The answer comes back synthesized and cited, with every document it relied on attached — so verifying it takes one click, not another afternoon.
What Affine cites
- 1Código do Trabalho — Art. 344.ºLei n.º 7/2009 · consolidated textLaw
- 2Código do Trabalho — Art. 345.ºLei n.º 7/2009 · consolidated textLaw
- 3Acórdão do Supremo Tribunal de JustiçaProc. n.º 1234/21.5T8LSBCase law
Ask in the chat. Get research, not a guess.
Your assistant hands the question to Affine, waits for the research to finish, and answers with the sources attached.
@Affine can we end a fixed-term employment contract before its term, and what does the employee receive?
Affine · research
donesearching legislation and case law · reading 14 documents · synthesizing
A fixed-term contract may be terminated before its term, but the grounds and the compensation due differ depending on who ends it and when1. Where the employer chooses not to renew, written notice within the statutory period is required, and compensation is calculated on the employee's base pay and length of service2. Courts have consistently treated a late or missing notice as conversion into an open-ended contract3.
Sources
1 Código do Trabalho — Art. 344.º
LawLei n.º 7/2009 · consolidated text
2 Código do Trabalho — Art. 345.º
LawLei n.º 7/2009 · consolidated text
3 Acórdão do Supremo Tribunal de Justiça
Case lawProc. n.º 1234/21.5T8LSB
Research, not retrieval
One question triggers a full search, read and synthesis pass over your sources — the same research agent that powers the Affine app, not a keyword lookup that hands back links.
Citations you can open
Every result arrives with a ready-made citation back to the article, ruling or statute it came from, so nothing leaves your desk unverified.
Down to the line
Your assistant can open a cited document, jump straight to the right article and quote it exactly, with the section and lines it came from.
Illustrative conversation. Sources shown are examples.
Five tools. One workflow.
Every connected user gets the same five tools. Research answers the question; the other four are there to check it.
01Research
The front door for any legal, tax or regulatory question. Searches, reads and synthesizes across the public sources for your jurisdictions, then returns a report with inline citations and every document behind it.
02Find a document
Look up a specific law, decree or ruling by name or number — “Código Civil”, “Lei n.º 7/2009” — for when you want that document rather than an answer.
03Read a document
Open the full text with its section outline, and page through long statutes without losing your place or the numbering.
04Search inside a document
Rank the passages of a single document by meaning, so a long statute or a lengthy ruling can answer one narrow question.
05Find an exact term
Search one document for an exact word, phrase or pattern, with the surrounding lines for context — for when only the precise wording will do.
Read-only, by design
The tools only read. Nothing in Affine can be changed or deleted from a conversation, and your private documents are never in reach from the assistant.
It sees exactly what you see.
The connection is not a new door into your data. Every call carries your own Affine sign-in and is answered under your own permissions.
Runs as you
Each call is made with your verified login. The assistant reaches the public sources for your account's jurisdictions — no more, no less.
Read-only
Nothing can be written, changed or deleted in Affine from a conversation. There is no write path at all.
Public sources only
From the assistant, Affine researches public legal databases. Your organization's private documents stay inside the Affine app.
No training on your data
Your questions and documents are never used to train models, shared with other clients, or aggregated in any way.
Questions, answered
What is MCP?expand_more
The Model Context Protocol is an open standard — started by Anthropic and now supported by OpenAI, Microsoft and others — for connecting AI assistants to outside tools and data. Affine runs a hosted MCP server, so a single connection works across every compatible assistant.
Which assistants can I use it with?expand_more
Claude, ChatGPT and Microsoft 365 Copilot today, plus any other MCP-compatible client. Because the connection is a URL rather than an app, new assistants work as they adopt the protocol.
Do I need to install anything?expand_more
No. You add a URL to your assistant and sign in with your Affine account. On team and enterprise plans an administrator adds the connector once for the organization, and members then connect individually.
Can it read my firm's private documents?expand_more
The tools cover public legal sources only. Research over your internal documents stays in the Affine app, where the permission model lives.
Why does a researched answer take longer than a normal reply?expand_more
Because it is not a lookup. Affine searches, reads and synthesizes across your sources before it answers, which can take a few minutes on a substantial question. Ask once and let it finish.
Which jurisdictions does it cover?expand_more
The same sources your Affine account covers, set by the jurisdictions you chose during setup. Our Sources page lists every database currently connected.
Start pulling the thread
Trusted by 100+ organizations