Oops! Something went wrong while submitting the form.
Surface founders building in stealth — before they announce, before they're on anyone's list. Sai reads the signals across LinkedIn, filings and hiring data, then hands you a qualified shortlist.
A stealth startup founder is someone who has already started building — incorporated the entity, pulled in a co-founder, sometimes closed a pre-seed round — but has deliberately said nothing publicly. No launch post, no press, no company page. For investors, this is the most valuable moment to make contact and the hardest moment to detect. By the time a founder is discoverable through the usual channels, three other funds have already had the conversation.
The difficulty is that stealth is not the absence of information. It is information scattered across places nobody checks together.
Founders in stealth are consistent in what they leak. Learning to read these is most of the work:
Any one of these on its own is noise. Two or three of them pointing at the same person, in the same window, is a stealth startup founder.
Each signal lives in a different system, and none of them are searchable in the way you actually need. LinkedIn will not let you query "changed headline to stealth in the last 60 days." Filings are public but unstructured. Job boards do not index by "company has no website." Checking all of it properly for one thesis is an afternoon, and it decays within a week — which is why most funds do it once, produce a list, and never refresh it.
That is the part worth handing to an agent. The judgment of what constitutes a fit stays with you; the sweeping does not.
Sai reads your thesis rather than string-matching it, then works each source in turn: LinkedIn for role and headline changes, search for domain and page activity, funding databases for filings that have not yet become announcements. It cross-references the results to collapse duplicates, drops anyone already in your CRM, and applies your stage, geography and timing filters before writing anything down. What lands in your sheet is one row per founder with the specific signals that flagged them, each linked to its source so a colleague can verify the call in a single click.
Replace each bracketed placeholder with your own parameters. Three adjustments that meaningfully change output quality: narrow the timeframe (a 60-day window surfaces genuinely fresh stealth founders; twelve months returns companies that have already launched), name the specific backgrounds you index on rather than a generic seniority bar, and paste in your anti-portfolio or do-not-contact list so the shortlist arrives already filtered.
A populated Google Sheet — name, prior company, the signals that surfaced them, a one-line fit rationale, and source links — plus drafted first-touch messages for the strongest matches, each referencing what the founder is actually building rather than a generic template. Run it weekly and Sai flags only what changed since the last pass, so the list stays current without being rebuilt.