Ask questions about a contract without sending it anywhere
A contract is precisely the document you cannot paste into a chatbot. It is somebody else's confidential information, it is frequently covered by an NDA that names disclosure to third-party services, and professional-conduct duties on confidentiality do not carve out a convenience exception.
The situation
- The useful question is rarely "summarise this" — it is "what is the notice period", "what happens if we terminate early", "who carries the liability here".
- Answering it means reading forty pages for the three that matter, which is the part that takes the afternoon.
- Any tool that would help has to keep the document on your machine, or it is not usable on client work at all.
How it works
1. Open the agent and load the contract
PDF or plain text, read in the browser. Nothing is posted to a server at any point.
2. It indexes the text on your machine
The document is split into passages and embedded locally, so the agent can find the clause that answers a question instead of skimming the whole file.
3. Ask in plain language
Answers are built from the retrieved passages and quote the actual wording, so you can check the clause rather than trust a paraphrase.
4. Keep it, or throw it away
The index lives in your browser's storage and goes when you clear it. Nothing is retained anywhere else, because nothing was sent anywhere else.
The agents that do this
Contract Plain-Language Explainer
Upload a contract and get it explained in plain language — obligations, fees, deadlines, exit clauses, and red flags — with every answer grounded in the actual text, fully in your browser.
Document Q&A
Upload a PDF or text file and ask questions answered only from that document — fully in your browser, nothing uploaded to a server.
Open one and it runs in this browser — nothing to install, and no account needed to try it. You can also browse everything else people have published.
What never leaves your computer
- The contract is opened, split and indexed by code running in your tab; the text never crosses the network.
- Our servers log the page request, the same as any website. They never see the document, the passages, or your questions.
- Downloaded as a single HTML file, the agent runs with the network off — a useful property when the answer is "it never left the building".
Why this is structural, not a policy
There is no server-side AI on AgentOp: the model runs in your browser tab, so your documents have nowhere to be sent. The full account of what our servers can and cannot see is on the trust page.
What it will not do
- It is not legal advice and does not pretend to be. It explains what the document says; the judgement about what that means for your client is yours.
- Answers are grounded in the passages it retrieves, so a question whose answer is spread across many distant clauses can be answered incompletely. It quotes what it used, so you can see what it read.
- A scanned or photographed contract is limited by on-device OCR (English printed text). A text-based PDF is read exactly.
- The small on-device models are good at finding and restating what a document says; they are not frontier models, and the honest use is locating and explaining clauses rather than drafting.
Will it run on my machine?
The on-device option needs a recent computer with a graphics card, or an Apple Silicon Mac, and Chrome or Edge — phones and iPads cannot run these models at all. Rather than guess, the browser check measures your actual machine in about a second and tells you which models it can run, before anything downloads. If the answer is no, the same agents work with your own OpenAI or Anthropic key instead — at the cost of the text leaving your machine, which for confidential material is the trade-off this page exists to avoid.
Questions
No. It is read, split and indexed inside your browser tab. AgentOp has no server-side inference and stores no agent documents — there is nowhere for the text to go.
No. The agent explains what a document says in plain language and points at the clauses it used. It is a reading aid for a professional, not a substitute for one, and it is instructed to say when something should be confirmed with a qualified professional.
That is the case it is built for, but the judgement is yours: on the on-device option nothing about the document leaves your machine, so no disclosure to a third-party processor occurs. If you instead supply an OpenAI or Anthropic key, the passages do go to that provider, and the usual third-party analysis applies.
Answers are assembled from passages retrieved out of the document itself and the agent is instructed to work strictly from them, quoting the wording rather than paraphrasing from memory. You can read the quoted clause to check it.
Nothing. The on-device model is free to run because the computing is yours. If you choose a cloud provider instead you pay that provider directly with your own key; AgentOp does not sit in between.
Related
Other jobs like this
The same idea applied to other kinds of confidential paperwork.
Will it run here?
A one-second test of your machine, before any model downloads.
Trust & security
Exactly what our servers see, and what they never receive.
Build your own
Change the wording, the fields or the rules — it is a template.