For students going after startup internships

Startup internships, found.
Cold emails, drafted.

AI agents find startups that are hiring, check your fit, and draft a fact-checked email to each one. You just hit send.

Product preview with sample data: you describe the internship you want, see three matching startups with a match score and a reason for each, then get a draft cold email to review and send yourself.

01 · Meet the crew

Six agents, one job each.

Judgment goes to agents. Everything else (dedupe, ranking, budgets) is plain code you can read.

  1. Lead

    Turns your goal into a search plan, reviews what comes back, and picks the shortlist.

    3 structured calls · plan · review · shortlist

  2. Scouts×3 in parallel

    Search the web at the same time, each from a different angle, for startups that are hiring.

    tool loop · web_search · fetch_page

  3. Researcher

    Reads each shortlisted company's site, job posts and GitHub so the email has real specifics.

    tool loop · web_search · fetch_page · github_lookup

  4. Fit check

    Scores how well you match, names your strengths and gaps, and asks follow-ups when unsure.

    structured call · sub-scores · strengths · gaps

  5. Writer

    Drafts a short, specific cold email built from the research and your resume.

    structured call · subject + body

  6. Critic

    Checks every claim in the draft against the research and your resume. Anything unsupported gets sent back.

    structured verdict · approve or reject with reasons

02 · Under the hood

Watch it work. Trust what it writes.

Live trace

Every step, as it happens

Starting the sample run…

Parallel scouts

Three searches at once

  • scout-12 leadsSeed-stage AI infrastructure startups hiring ML or backend interns, Bay Area or remote.
  • scout-22 leadsRecent YC batch companies with open internship or 'we're hiring' pages.
  • scout-32 leadsOpen-source dev-tools companies with active GitHub orgs and good-first-issue tickets.
Critic

Checks every claim

Draft 1 · rejected

Hi team, I saw Tessellate AI just raised a $40M Series B and I'd love to help. I've built FastAPI services and contributed to open-source Python tools. Would you be open to a short chat about internships this summer?

The funding claim contradicts the research (seed, not Series B).

Draft 2 · approved

Hi team, I've been following the Tessellate AI serving runtime on GitHub and I'd love to help. I've built FastAPI services and contributed to open-source Python tools. Would you be open to a short chat about internships this summer?

Budget caps

It never runs away

Searches9 / 12

Caps on searches, pages and cost for every run. Hit one and the crew stops cleanly and tells you why.

Control

You stay in charge

  • Stop any runCancel mid-run. The crew stops after its current step.
  • See every stepEvery search, page and decision is on the trace.
  • You send itCopy the draft into your own inbox and change anything.

Examples above are from a recorded sample run.

What it does not do

Enforced in code, not policy

  • Never sends emailIt drafts. You copy, edit and send from your own inbox.
  • Never applies for youNo forms, no applications, no accounts made in your name.
  • Public sources onlyCompany sites, job posts, blogs and public GitHub.
  • Never scrapes LinkedInLinkedIn and similar sites are blocked outright.

03 · FAQ

Questions

Does it send emails for me?

No. It only drafts. You review each email, change what you like, and send it yourself.

Where does the information come from?

Public pages only: company sites, job posts, blogs and public GitHub. Never LinkedIn.

What do I need to give it?

A short goal (role, stage, location) and your resume. The resume stays private to your account.

How long does a run take?

A few minutes. You can watch every step live, or leave and come back to the results.

Who can use it?

Access is invite-only for now. Request access and we will get back to you.