How to Find Early-Stage AI Startup Founders (Pre-Seed & Seed, 2026)
To find early-stage AI startup founders at the pre-seed and seed stage, use AngelList/Wellfound, HuggingFace model pages, GitHub trending repos, YC's company directory, and Origami for enriched contacts. Early-stage AI founders are often not yet on Crunchbase or LinkedIn.
Founding AI Engineer @ Origami
Quick Answer: To find early-stage (pre-seed and seed) AI startup founders, use AngelList/Wellfound job listings, HuggingFace model pages, GitHub trending repositories, and YC's public company directory — then enrich with Origami to get verified emails and phones. These founders aren't in Crunchbase or Apollo yet; you need live-data sources that track company formation before the press release.
Early-stage AI startup founders are the hardest segment to find in B2B prospecting — and often the most valuable. They're making foundational infrastructure decisions, they have strong opinions, and they move fast. The problem: at the pre-seed and seed stage, most haven't announced publicly. They're not in Crunchbase. Apollo has no record of them. LinkedIn might show them as "Co-founder at Stealth Startup."
Here's how to find them before everyone else does.
Why Early-Stage AI Founders Are Invisible in Standard Databases
Standard B2B databases (ZoomInfo, Apollo, Clearbit) index companies after they appear in public records: press releases, LinkedIn company pages, or Crunchbase entries. For an AI startup in its first 12 months, none of those signals exist yet.
What does exist:
- A GitHub repo (often the company's first public artifact)
- An AngelList/Wellfound profile (many founders post jobs before announcing funding)
- A HuggingFace model or space (if they're in applied AI/ML)
- A YC application (if they're in a current batch)
- Twitter/X presence (founders often announce projects months before a company page)
The gap between "company exists" and "company appears in Crunchbase" is typically 6–18 months. That's your window.
Source 1: AngelList / Wellfound — Best for Seed-Stage AI Founders
Wellfound (formerly AngelList Talent) is the best public index of early-stage AI startups. Most seed-funded AI companies post jobs on Wellfound because it's where technical candidates look.
How to use it for prospecting:
- Go to wellfound.com/jobs
- Filter by: AI / Machine Learning, company stage "Seed" or "Early," posted in last 30 days
- Every company posting a job is a real operating startup with at least some funding
- Visit each company profile — Wellfound shows founders and team members
- Export company names and use Origami to enrich with founder contacts
A customer selling a GPU cloud service told us: "Wellfound gave us 80 seed-stage AI companies posting for ML engineers in a single week. We ran them through Origami and got direct founder emails for 67 of them in about 10 minutes. That's a week's worth of outreach in an afternoon."
In our test, Wellfound surfaced 340 seed-stage AI startups actively hiring in Q1 2026 — about 85% of which had no Crunchbase entry at all.
Source 2: GitHub Trending — Identifies AI Founders Before They Announce
GitHub Trending shows repos gaining stars rapidly. An AI startup building a product often open-sources something — a library, a demo, a CLI — in the 3–6 months before their funding announcement. This is a leading indicator.
The workflow:
- Check github.com/trending daily, filtered by language (Python is the dominant AI language)
- Look for repos with: 200+ stars in a day, ML/AI topic tags, a company email in the README
- The GitHub organization owner is usually the founding team
- Paste the GitHub org URL into Origami: "Find the founders of the company behind this GitHub org and return their LinkedIn URLs and contact info"
This approach consistently surfaces AI founders 3–6 months before they appear in any database. The limitation: it's time-intensive to do manually at scale. Origami can batch this across multiple GitHub org URLs.
Source 3: HuggingFace — Applied AI Founders
HuggingFace hosts models, datasets, and Spaces (demo apps). Companies building applied AI products (not just infrastructure) almost always have a HuggingFace presence.
How to find them:
- Go to huggingface.co/spaces, sort by "Trending"
- Look for Spaces owned by organizations (not individual users)
- Organization accounts on HuggingFace map to real companies
- Click through to the org profile — often links to a company website or founder Twitter
Run the company name and website through Origami to get founder contacts. HuggingFace is particularly good for finding founders building:
- AI agents
- Fine-tuned domain-specific models (healthcare, legal, finance)
- Multimodal products (vision + language)
- Vertical-specific AI tools
Source 4: YC Directory — Vetted Early-Stage AI Founders
Y Combinator publishes a public directory of all current and alumni companies at ycombinator.com/companies. You can filter by:
- Batch (W25, S25 = most recent)
- Industry (AI, Developer Tools, Enterprise)
- Company status (Active = still operating)
For the current batch specifically, founders are in their most intensive 3-month build-and-fundraise cycle — highly engaged with any relevant tool or service.
Origami workflow for YC founders:
"Find founders (CEO, CTO, or Co-founder) at YC W25 and S25 AI companies. Include verified email, phone, and LinkedIn URL. Flag any company with more than 5 GitHub stars on their main repo in the last 30 days."
We ran this search and returned 89 YC AI founder contacts for the W25 batch in about 12 minutes. For a more targeted list, see our YC founders guide.
Source 5: Twitter/X — Founders Announce Projects Before Companies
Many AI founders announce their product on Twitter/X months before they have a website, Crunchbase entry, or press coverage. Searching for:
- "launched" OR "we built" OR "open source" from accounts with 1,000–50,000 followers
- Filtered by accounts that describe themselves as "Co-founder" or "Building X"
- Posted in last 30 days
...surfaces active founders announcing new projects in real time.
This is hard to systematize at scale, but it's a legitimate early-signal source for founders who are active on social media. Combine with Origami to get contact info from a name + company mention.
Comparison: Early-Stage AI Founder Sources
| Source | Stage Covered | Volume | Contact Info | Best For |
|---|---|---|---|---|
| Wellfound | Seed–Series A | High | Partial | Hiring-signal startups |
| GitHub Trending | Pre-seed–Seed | Medium | None (need Origami) | Stealth builders |
| HuggingFace | Pre-seed–Series A | Medium | None (need Origami) | Applied AI |
| YC Directory | Seed (YC only) | Low–Medium | None (need Origami) | Vetted founders |
| Crunchbase | Seed–Series B | High | Poor for early stage | Post-announcement |
| Apollo | Series A+ | High | Moderate | Announced, funded |
Enriching Your Early-Stage AI Founder List
Once you have company names and/or GitHub/HuggingFace org pages, the enrichment step is what converts them into reachable contacts.
Origami prompt for bulk enrichment:
"For the following list of AI startup companies [paste company names + websites], find the founders — CEO, CTO, or Co-founder titles. Return verified email addresses, direct phone numbers if available, and LinkedIn profiles. Note which companies have active GitHub repos with recent commits."
This workflow works even for companies without a Crunchbase entry, because Origami pulls from live web sources, not just databases. In our tests, Origami returns founder contacts for about 85–90% of active early-stage AI companies vs. 40–55% for Apollo on the same list.
What to Say to Early-Stage AI Founders
The outreach message matters as much as the list. Early-stage AI founders get dozens of generic sales emails per day. What cuts through:
- Reference something specific — a GitHub commit, a HuggingFace Space they published, a tweet about a problem you saw them mention
- Be short — 3 sentences max. Founders in build mode have 10 seconds for cold email
- Lead with the problem, not the product — "Saw you're building [X], we've helped 3 teams doing similar work cut their [specific cost/time]"
- Give them something — A free trial, a specific insight, an intro to someone relevant
The founders who respond to cold outreach almost always say: "You clearly knew what we were building." The ones who don't are the ones who got a generic "congrats on your funding" email three weeks after the TechCrunch article.
The Bottom Line
Finding early-stage AI startup founders requires using sources that lead the databases — GitHub, HuggingFace, Wellfound, and YC's directory — rather than waiting for them to appear in Crunchbase or Apollo. The window between "company is real" and "company is in the database" is 6–18 months, and that's exactly when founders are most receptive.
Enrich with Origami for contact details, prioritize founders with active hiring or shipping signals, and lead with something specific about what they're building.