A policy research analyst that reads the actual bill, not the press release
It reads the actual bill, not the press release — a sourced brief with section citations, agency dates, and the dissenting view.
Hand it a question, get back a sourced brief with section citations, agency dates, and the dissenting view — not a confident paragraph of nothing.
What a junior analyst spends three days on
A markup gets scheduled, a rule drops on regulations.gov, a committee posts an amended version at 4:40pm — and someone on your team now has to read 90 pages, cross-check what changed against the introduced version, and tell the principal what it means before the morning call. That work is real, but most of it is fetching, not thinking.
The agent does the fetching. Give it a bill number or a question and it browses the full text on Congress.gov or your state legislature site, pulls the CRS summary if one exists, reads the agency NPRM and the comment count, and checks who the listed sponsors and committee are. It synthesizes a brief with inline citations that link back to the exact source, so when your director asks "where does it say that," you have the section number, not a vibe.
It works the way a research desk works:
- Reads primary documents end to end — bill text, Federal Register notices, dockets, CBO scores — not just headlines that summarize them
- Cites every claim to a URL and a section, and flags when a source is secondary or a source can't be found
- Tracks a bill or docket across versions and tells you what changed between the introduced and reported text
- Calls legislative and Federal Register APIs directly when you have a key, so it pulls structured status instead of scraping
- Drafts the brief into Notes or a shared Sheet, and emails it to the distribution list on a schedule you set
It will not pretend to know. When the comment period dates aren't posted yet, or a fiscal note hasn't been filed, it says so and tells you where it looked. A research product you can't trace is worse than no product, so traceability is the whole design.
Related pages in this workflow
These pages cover the adjacent jobs buyers usually compare before choosing an AI agent.
- AI Procurement Research Analyst — A sourcing request, back as a cited supplier brief: shortlisted vendors, pricing signals, and lead times, with a source behind each.
- AI News Briefing Agent — One briefing every morning, built around your world: your industry, your tickers, your topics — scanned, summarized, and waiting in your inbox.
- AI Research Agent — Hours of digging, back in minutes: it reads the full sources and hands you a structured report with every claim cited.
- AI Research Analyst — Hand it a question and a deadline, get a brief that shows its work — figures, links, and the page each number came from.
- AI Legal Associate — The first-pass workload, done: it reads a whole bundle of contracts or case files, writes one summary, drafts the brief, and files it to the right contact.
- AI Legal Research Analyst — A memo where every claim has a link you can check: it searches the open web and reads the actual opinions and filings.
- AI Real Estate Market Analyst — A neighborhood read in one sourced doc — comps, days-on-market, price per square foot, inventory trends — instead of twelve open tabs.
- AI Market Sizing Analyst — TAM/SAM/SOM that shows its work — top-down and bottom-up, every figure traced back to the source it came from.
Frequently asked questions
Does it actually read the bill or just summarize the summary?
It reads the full text. It will pull the official summary or CRS overview as one input, but the claims in the brief are checked against the bill or rule itself, and the citation points to the section, not the summary page.
How do I know it isn't making citations up?
It cites a URL and section for each claim and browses to that source to ground it. When it can't find a source for something, it says so explicitly rather than filling the gap. You can click every link to verify before circulating.
Can it cover state legislatures, not just Congress?
Yes. It browses state legislature sites and regulatory portals the same way it browses Congress.gov. For sites with an API and a key, it can call that directly for cleaner status data.
Can it watch a bill and tell me when it moves?
Set a scheduled sweep or a trigger on the docket. When an amended version or a new Federal Register notice appears, it reads the change, diffs it, and emails you what's different — without you checking manually.
Will it just give me both-sides mush?
No. It states what the text does, then lays out the positions on record for and against with attribution. Where the evidence is one-sided, it says that instead of inventing a counterweight.
How much does it cost?
$49/month or $490/year — cancel anytime. Setup takes about four minutes — name the agent, tell it your topics and jurisdictions, paste any API keys you have, and it's researching. It's in public beta.
AI research analysts by field
One research agent that searches, reads, and cites — pointed at the sources and questions your field cares about.
- AI Research Analyst (hub) — Hand it a question and a deadline, get a brief that shows its work — figures, links, and the page each number came from.
- AI Academic Research Analyst — It reads the papers, not just the abstracts — a synthesized brief with inline citations, source links, and the contradictions flagged.
- AI Brand Monitoring Analyst — Reads the whole thread, not the headline: it checks the sites you watch each morning and briefs you on what actually changed overnight.
- AI Competitive Intelligence Analyst — The whole market, read for you: it watches competitor pricing, changelogs, and job boards and drops a cited brief the moment something moves.
- AI Data Analyst Agent — A data team without the headcount: point it at a CSV or Sheet and ask in plain language — it cleans, charts, and answers, with full undo.
- AI Due Diligence Analyst — Hand it a company name, get a brief you can trust: founders, funding, and litigation flags, each with a citation behind it.
- AI Equity Research Analyst — Hand it a ticker, get a cited brief — built from the 10-K, the latest call, and live multiples, every claim linked to its source.
- AI Grant Research Analyst — Finds the grants you actually qualify for: it reads the RFP, checks your eligibility, and hands you a sourced one-pager with the ask and deadline.
- AI Healthcare Research Analyst — Ask a clinical question, get an answer you can defend: a structured brief where every claim links to the paper, trial, or guideline.
- AI Junior Consultant — Brief to deliverable in a single session: it runs the research, organizes the findings, and builds the deck and the client-ready PDF.
- AI Legal Research Analyst — A memo where every claim has a link you can check: it searches the open web and reads the actual opinions and filings.
- AI Market Research Agent — A cited competitive briefing in your inbox on the cadence you set — competitors watched, trends tracked, a source behind every line.
- AI Market Sizing Analyst — TAM/SAM/SOM that shows its work — top-down and bottom-up, every figure traced back to the source it came from.
- AI News Briefing Agent — One briefing every morning, built around your world: your industry, your tickers, your topics — scanned, summarized, and waiting in your inbox.
- AI Procurement Research Analyst — A sourcing request, back as a cited supplier brief: shortlisted vendors, pricing signals, and lead times, with a source behind each.
- AI Real Estate Market Analyst — A neighborhood read in one sourced doc — comps, days-on-market, price per square foot, inventory trends — instead of twelve open tabs.
- AI Research Agent — Hours of digging, back in minutes: it reads the full sources and hands you a structured report with every claim cited.