HIRE CLAIRE.YOURDIGITALBUSINESS ANALYST.
Claire is Digital Crew's AI business analyst. Drop in any URL and she runs a 3-tier scrape — HTTP, stealth browser, full Playwright — then returns a structured profile: what the company does, who they sell to, and a proof point. The brain behind every Crew agent's account research.
Claire Sterling
Analysts cost €80K.
And take weeks.
Hiring a business analyst is broken. Senior ones take three months to ramp. Junior ones miss half the data. The good ones cost a fortune. There's a third option.
- €80k / year all-in. Six months to fluency. Days per company profiled.
- Works 40 hours / week. Sleeps. Takes vacations. Misses sources.
- Half the research lives in their inbox or a stale Notion doc.
- Doesn't scale. Need 100 accounts profiled this quarter? Hire 10× the analysts.
- Every report has a different shape. You spend hours normalizing them.
- €1.5k / month. Onboards in 10 minutes. Profiles in ~30 seconds.
- Works 24/7. Never sleeps. Never misses a tier-3 escalation.
- Every job lands as structured JSON in Supabase. Searchable forever.
- Clone her. Spin up 50 Claire instances and profile 50 accounts in parallel.
- MIT licensed. Self-host, fork, modify. She's yours forever.
From URL to Profile
in Six Steps.
Same flow as briefing a real analyst — recruit, onboard, brief, work, review, deepen. Except Claire does 80% of each step herself.
- 01Recruit
Choose Your Hire
Claire is your digital business analyst. Account research, market briefs, competitive deep dives — open source, self-hosted, yours forever.
- 02Onboard
Brief Your New Hire
10-minute conversation, not a setup wizard. Claire learns your ICP, your domain language, and the questions you actually need answered.
- 03Brief
Drop a URL
POST any company URL to /research. Claire kicks off a 3-tier scrape — fast HTTP, then stealth Camoufox, then full Playwright if the site fights back.
- 04Work
Claire Researches the Account
Scrapling pulls the page. OpenRouter's gpt-4o-mini extracts structure. browser-use Cloud is the AI fallback when nothing else cracks the site.
- 05Review
Read the Profile
Three fields you can act on — what_you_do, who_you_sell_to, proof_point — stored in Supabase and polled over a single GET endpoint.
- 06Deepen
Run Deep Research
Need more than one page? Spin up Claire's deep-research mode for multi-page narratives, competitor maps, and longer briefs.
Everything a junior analyst does. Just faster. And structured.
Briefs in conversation, not forms
Tell Claire what to research in plain English — or just paste a URL. She translates the brief into a research plan, runs the scrape, and shows the structured result before it lands in Supabase.
https://stripe.com — what do they do, who do they sell to, and one proof point?
Plugs into the stack you already run
Scrapling for the scrape, OpenRouter for the LLM, browser-use Cloud for the AI fallback, Supabase for storage. No new vendors, no rip-and-replace.
Reports in like a real employee
Every job emits a structured profile — what_you_do, who_you_sell_to, proof_point — with status transitions (pending → running → complete) you can poll over a single REST endpoint.
Profiles dozens of accounts in parallel
Claire runs each research job as an independent worker. Need 50 profiles by Friday? Spin up 50 instances. Same code, same output schema, different URLs.
Stripe — Profiled
Brex
Notion
Datadog
Linear
Vercel
HubSpot
Asana
Webflow
Ramp
Twilio
Klaviyo
+ URL
Claire levels up.
The longer she works, the sharper she gets.
Every URL teaches Claire something. Every domain primes her extraction model. Every reviewed profile sharpens her judgment. By month six, she's a Level 5 — running the full pipeline while you supervise.
- 1
Rookie
Day 1–7Onboarded. Knows your ICP and domain vocabulary.
- 2
Junior
Week 2–4First 100 profiles delivered. Tier-1 covers most.
- 3
Apprentice
Month 2–3Optimizing tier escalation. Learning sites that fight back.
- 4
Pro
Month 4–6Handles JS-heavy sites solo. Fallback only on edge cases.
- 5
Autonomous
Month 6+Runs the full pipeline. Manager reviews weekly.
Built for Power Users & Developers
A research stack you can actually inspect, fork, and run yourself
- Unique
3-Tier Scraping Pipeline
Scrapling routes each URL through HTTP → Camoufox stealth browser → full Playwright. Fast where it can, undetectable where it must.
Structured Output
Every profile returns the same three fields — what_you_do, who_you_sell_to, proof_point. Drop into your CRM, your prompt, your dashboard.
- New
browser-use Fallback
When Scrapling can't crack a site, Claire spins up a browser-use Cloud agent and lets an LLM drive a real browser to extract what's needed.
Multi-Model AI
OpenRouter under the hood. gpt-4o-mini by default; swap in Claude, Mistral, or local models. You're never locked to one provider.
- Developer
Two-Endpoint REST API
POST /research with a URL, get a job_id. GET /research/{job_id} until status is complete. That's the whole API surface.
- New
MCP Server
Connect Claire to any AI tool via the Model Context Protocol. Drive her from Claude, Cursor, or any MCP-compatible client.
Profiles You Can Actually Use
One URL in. Three structured fields out. Same shape every time.
what_you_do
Stripe builds payments infrastructure for the internet — APIs and tools that let businesses accept money online, manage subscriptions, issue cards, and run their financial stack from one platform.
who_you_sell_to — Internet businesses of every size, from indie SaaS founders to public companies. Heavy adoption among startups, e-commerce, marketplaces, and platforms that need to move money.
proof_point — Processed over $1T in total payment volume in 2023, with millions of businesses live on the platform across 195 countries.
— Tier 1 · Scrapling HTTP · 1.8s
What Claire scraped
- Homepage hero + product nav
- Customers & case studies pages
- Investor / press release scrape
“Same schema for every site. Whether it's a landing page or a JS-heavy SPA.”
Claire scrapes, then structures
Your digitalworkforce.You own the code.
Claire is fully open source under MIT. Read every line. Self-host on your own VPS. Audit how she handles the URLs you point her at. Fork her for your stack. We can never lock you in — that's the point.
Installation
~2 min- 1.Clone the repositorygit clone https://github.com/digital-crew-technologies/claire-agent.gitcd claire-agent
- 2.Configure environment variables
Copy
.env.exampleto.envat the repo root and fill inSUPABASE_*,OPENROUTER_API_KEY, andBROWSER_USE_API_KEY. - 3.Run the Supabase migrations
Open the Supabase SQL editor and run, in order,
backend/migrations/001_research_jobs.sqlthen002_research_jobs_v2.sql. - 4.Start the stackdocker compose up -d
Frontend on http://localhost:3000, API on http://localhost:8000.
PAY ZERO TO START. PAY ONLY WHEN CLAIRE DELIVERS.
Self-host the open source for €0. Let us run it and only pay per profile delivered. Or build on top of the API. Same Claire underneath.
OPEN SOURCE
Builders & tinkerersSelf-hosted. MIT licensed. Yours forever.
- Unlimited Claire instances on your infra
- Full 3-tier scraping pipeline (Scrapling + Camoufox + Playwright)
- Bring your own keys (OpenRouter, browser-use, Supabase)
- Community support on GitHub & Discord
- Fork it. Modify it. Resell it.
PAY AS YOU GO
Teams that want profiles, not infraHosted Claire. Pay only per profile delivered.
Only billed on status=complete jobs. No monthly fee.
- Zero upfront — $0 until Claire delivers your first profile
- Hosted Claire, billed per completed research_job
- All three tiers + browser-use fallback included
- Live dashboards & weekly performance review
- Cancel anytime, no contracts, no lock-in
- No credit card required
- Cancel anytime
- MIT licensed core
- EU data residency
Questions Teams Ask
Everything you need to know about pointing Claire at a URL