The Lab-to-Spinout Watchlist: Grant Signals to a Dated Founder Queue
Build a thesis-scoped watchlist of 20 to 60 pre-incorporation academic teams, each with a named founder, licence status, formation window and contact route.
Key takeaways
- 57% of entrepreneurial leads on National I-Corps teams in FY23-24 were graduate students, so the award PI is usually the professor who stays in the lab, not the person who will run the company.
- A published university patent application reflects a filing made about 18 months earlier, so it confirms IP that may already be optioned and is not an early signal.
- Exclude I-Corps at NIH teams from a pre-incorporation queue: that programme funds existing small companies through SBIR or STTR supplements, which is why its startups-per-team figure reads 96%.
- Companies formed by FY2019-20 NSF I-Corps teams rose 20%, from 302 to 361, between reports, so a watchlist that drops cohorts after 12 months misses the late formers.
- In Refolk's index, only 142 of 1,242 US licensing and tech-transfer titles (11.4%) sit at higher-education employers, which makes the tech-transfer contact route narrow and findable.
- An option has a hard expiry, such as Cornell's 6-month option extendable once for $2,500, and that expiry turns licence status into a dated formation window.
This playbook is for early-stage investors, fund platform and talent partners, and angels who want to meet academic founders before the company is incorporated and the round is shopped. It turns public commercialization-grant awards, proof-of-concept fund awards, university-assigned patent filings and tech-transfer listings into a thesis-scoped watchlist of 20 to 60 teams. Each row names the likely full-time founder, the licence status, a projected formation window and a first-contact route. Setup takes one day. After that, keeping the list current takes about one analyst-day a week.
What signals show a lab team is about to spin out?
Four public signals show it. In rough order from earliest to latest, they are a commercialization-training award such as NSF I-Corps, a proof-of-concept fund award, a tech-transfer listing, and a published patent application assigned to the university. None of them proves a company is coming on its own. My threshold for a watchlist row is two signals on the same team plus a named student or postdoc.
A commercialization-training award is grant money a university team gets to test whether its research has a market, mostly through customer-discovery interviews. A proof-of-concept award is a university's own money spent to de-risk a technology before it is licensed. A tech-transfer office (TTO) is the university unit that owns, patents and licenses inventions made in its labs.
| Signal | What it proves | How it lies |
|---|---|---|
| National I-Corps Teams award | A team with a named business lead is testing a market | The PI on the award is not the founder |
| Regional Hub or Site training | The team is months ahead of a national cohort | Rosters may not name the business lead |
| Proof-of-concept award | The university is spending its own money on it | Names a project, not a founder |
| TTO available-technology listing | The university holds rights it wants to license | Nobody publishes how often these pages update |
| Published assigned patent application | The university owns filed IP in your area | The filing is about 18 months old |
One structural fact makes the I-Corps signal cleaner than it looks: NSF I-Corps will not typically fund teams whose IP is assigned to a third party. So a national award usually means the IP has not already gone to an outside company.
When each signal fires
- Regional trainingHub and Site rosters appear about six months before a national cohort
- National I-Corps awardLands in NSF Award Search with institution, PI and award date
- Proof-of-concept awardPublished by the fund after each round
- Option or licenceRights reserved for a dated term with a hard expiry
- Patent publicationConfirms a filing made about 18 months earlier
Why the patent layer is always late
US patent applications publish promptly once 18 months have passed since the earliest filing date. Provisional applications, which are cheap placeholder filings, are never published. A provisional counts as abandoned 12 months after filing unless it is converted. If a team converts at month 12, the non-provisional publishes about 6 months later. So the earliest public trace of a university patent is roughly a year and a half after the first filing. By then, an option may already be signed.
Why the training layer is early
There is now roughly a six-month gap between when a team applies to a national I-Corps cohort and when it takes part. Teams without prior NSF funding become eligible through regional Hub or Site training. So regional announcements come before the national award, and that six-month window costs you nothing to use.
Which commercialization programmes belong in the queue?
NSF National I-Corps and Energy I-Corps teams belong in a pre-incorporation queue. I-Corps at NIH teams do not. That programme is funded as an administrative supplement to small companies that already hold an active Phase I SBIR or STTR grant, so its teams are companies already.
| Programme | Teams trained | Startups | Startups per team |
|---|---|---|---|
| NSF I-Corps | 2,546 | 1,380 | 54% |
| I-Corps at NIH | 317 | 305 | 96% (already SBIR companies) |
| Energy I-Corps | ~188 | 19 | ~10% |
Source: NSF I-Corps biennial reporting. The per-team column is derived.
Read the table carefully. The NIH figure of 96% does not mean those teams are good bets. It means they were companies before training started. In that programme, all entrepreneurial leads are C-level executives. The Energy figure covers teams based at National Laboratories, not universities, and it converts at about a tenth the NSF rate. Keep Energy rows only if your thesis is in that domain.
Cadence and volume
- NSF runs 12 national I-Corps Teams cohorts a year, with about 24 teams in each.
- The program grew from 509 teams in FY2017-18 to 685 teams in FY2023-24.
- 52% of National I-Corps teams have been linked to startups.
- One secondary source reports that essentially all accepted teams receive the full $50,000. I have not confirmed this against NSF, so check it on the award record before you rely on it.
Late formers are real
No public source gives a median time from finishing a cohort to incorporating. The best timing evidence is a recount between reports: companies formed by FY2019-20 teams rose 20%, from 302 to 361. NSF's conclusion is that startups keep launching years after I-Corps training. The mechanism is simple. Many founders incorporate after they graduate, not during the cohort.
Proof-of-concept funds publish on a calendar
Proof-of-concept funds name their awardees after each round, which makes them easy to schedule.
- MIT's Deshpande Center made 8 grants totalling about $600,000 in academic year 2025. It reports that about 62 projects have spun out.
- Yale Ventures publishes Blavatnik Fund awardees by name and now runs funding twice a year. Accelerator awards feed the early pipeline. As of 2021, the fund had backed 33 projects, which produced eleven funded ventures that had raised $107 million.
For scale, AUTM's 2023 survey reported over 25,000 invention disclosures, and startup creation fell 11% from 2022. In Canada in 2023, there were 683 licences and options and 134 startups. Disclosures vastly outnumber companies, so the watchlist has to filter hard. That is why two signals, not one, is the bar.
How do you tell the full-time founder from the faculty co-founder?
The award PI is the technical lead, and that is usually the professor who stays in the lab. The entrepreneurial lead (EL) is the person on an I-Corps team responsible for the business side, typically a PhD student or postdoc who works directly with the technology in the PI's lab. The EL is the likely full-time founder. Name both on every row.
NSF's own FAQ separates the roles. The technical lead is a role on the I-Corps team. The principal investigator is a role on the NSF proposal and award. The award database records the PI. Nobody has publicly documented whether the EL appears as a structured field, so treat any EL name you pull from an abstract as unverified until a second source confirms it.
Where the EL's name actually turns up
- Award abstracts, which sometimes name the team.
- Regional Hub or Site announcements and cohort pages.
- Lab people pages, for current PhD students and postdocs.
- Co-inventor lists on university-assigned patent applications.
- The person's own professional profile or code activity, which often shows the career stage.
Each row needs an EL name, a career stage (PhD student, postdoc, research staff) and an expected graduation or contract-end date. That end date matters more than the award date. It is the point at which the EL can actually leave the lab to run something.
The PI on the award stays in the lab. The entrepreneurial lead is the one who leaves to run the company.
When the EL is not the CEO
Brandeis notes that students and postdocs rarely have the experience or business skills to be a company's sole management. Flag rows where a hired CEO looks likely. Typical cases are an EL early in a PhD, a team with secured rights but no committed lead, or a technology in a heavily regulated field. For a platform or talent partner, a flagged row is an opening: you may be the one who introduces the CEO.
Assembling these names by hand means cross-reading award abstracts, lab pages and patent inventor lists, then matching each person to a public profile. Refolk collapses that into one plain-English request across the open web, LinkedIn and the public GitHub graph. It returns named people with their current roles, which you can then check against the award record.
Founder commitment against licence status
How do licence status and the formation window fit together?
Every row gets one of four licence states, and the state sets the formation window more than the grant date does. An option carries a hard expiry, and the team has to incorporate or lose its reserved rights before that date.
| Status | What it means | Evidence | Window effect |
|---|---|---|---|
| None | No known invention disclosure | Award or lab page only | Open-ended; recheck weekly |
| Disclosed | Invention reported to the TTO, maybe a provisional filed | TTO listing, conversation | Provisional abandons at 12 months unless converted |
| Optioned | University agrees not to license elsewhere for a set term | TTO confirmation | Formation due before expiry |
| Licensed | Rights signed to an entity | TTO startup page | Company likely exists; hand off |
Options make the window datable
An option is a time-limited promise from the university not to license the technology to anyone else while the team gets ready. One practitioner write-up describes the university agreeing not to license for 6 to 24 months. Cornell is a concrete example. It offers a simple 6-month option for a $1,000 fee, waived if the inventors are on the founding team, and the option can be extended for another 6 months for $2,500. A Cornell-style optioned row therefore has a known worst-case end date, at most 12 months from the start.
The order of events is contested
Sources disagree on what happens first. One practitioner says founders generally need to incorporate before any IP can be licensed. Toronto describes option, licence and investment negotiations running in parallel. The Cornell option suggests an option can come before incorporation. Record which pattern each target institution follows, because it changes whether "optioned" means "about to incorporate" or "incorporating now".
What is not established
AUTM does not publish the time from disclosure to option or licence, so that figure is not established publicly. AUTM also stopped collecting the closest proxy, the number of licences and options executed by startups, after 2022. The one practitioner write-up I have describes 18 to 24 months as total outliers. Do not project formation windows to the month. Use quarter ranges, anchored on the earliest of the option expiry and the EL's end date.
Equity terms change what you are pricing into
| Regime | University stake | Anti-dilution | Royalty |
|---|---|---|---|
| USIT, life sciences | 10-25% | Dilutable | ~0.5-5% |
| USIT for Software | 5-10% | None before first round | 0-2% |
| US ~5% model | ~5% | Through Series A | Yes |
Sources: TenU USIT Guide and its 2024 software supplement; published commentary on UK and US spinout terms.
Anti-dilution protection means the university keeps its percentage when new shares are issued, so the dilution falls on everyone else. Under the US model of about 5%, the university's stake stays protected until Series A closes, which pushes dilution onto founders and early investors. Ask the TTO which policy applies before you discuss terms. A pre-seed cheque behaves differently under each regime.
The weekly watchlist procedure
The procedure runs in eight steps. After one setup day, it takes about half a day of award pulls, 2 to 4 hours of patent search and 2 hours of refresh each week, plus about 2 hours per proof-of-concept funding cycle. Done looks like a dated table of 20 to 60 rows, each with source signals, a named EL, a licence state, a formation window and a contact route.
Lab to dated founder queue
- Scope the thesisThe partner and an analyst agree a keyword list and 15 to 40 target institutions. One day; done is a written list of both.
- Pull training awardsQuery NSF Award Search for I-Corps awards from the last 18 months and pull Energy I-Corps cohort pages, excluding I-Corps at NIH. Half a day a week; done is one row per team with institution, PI and award date.
- Add proof-of-concept awardsAfter each round, add named awardees from funds such as Deshpande and Blavatnik, matched to existing rows or opened as new ones. About 2 hours per cycle.
- Add the patent layerSearch published applications assigned to target universities in the thesis classes and attach them by inventor and institution. 2 to 4 hours a week; log each as a lagging confirmation of a filing about 18 months old.
- Split the founder rolesName the EL separately from the technical lead or PI. Done is an EL name, career stage, and expected graduation or contract-end date on every row, with a hired-CEO flag where it applies.
- Assign licence statusThe partner sets none, disclosed, optioned or licensed from TTO listings and conversations, recording option start and expiry dates.
- Set the formation window and contact routeThe partner assigns a quarter-range window and a named first contact, defaulting to TTO new-ventures or licensing staff, and notes the institution's ordering of option and incorporation.
- Refresh weeklyRerun the award and patent pulls, move rows between stages, and retire rows that incorporate or go cold. About 2 hours; done is a dated list where every status reflects the latest evidence.
The row schema below keeps the list sortable and makes each claim traceable to a source. Paste it as the header of a spreadsheet or database table.
row_id: institution: lab_pi_name: (technical lead, stays in lab) el_name: (entrepreneurial lead, likely full-time founder) el_career_stage: PhD student | postdoc | research staff | unknown el_end_date: expected graduation or contract end, YYYY-MM el_confirmed_by: two sources, e.g. award abstract + lab page hired_ceo_likely: yes | no | unclear signal_icorps: award ID + award date | none signal_regional_training: Hub or Site name + date | none signal_poc_award: fund + round + date | none signal_tto_listing: listing title + date seen | none signal_patent_pub: publication number + pub date + est. filing date | none licence_status: none | disclosed | optioned | licensed option_start / option_expiry: YYYY-MM / YYYY-MM institution_order: incorporate-then-licence | option-first | parallel | unknown formation_window: quarter range, e.g. Q2-Q3 of next year contact_route: named TTO staff member + title | lab | none yet last_reviewed: YYYY-MM-DD stage: watch | ready | contacted | incorporated | cold
Keep the allowed values exactly as written so rows sort and filter cleanly; add thesis-specific columns at the end.
Who to contact first, and how?
Make first contact through the TTO's new-ventures or licensing staff, not the student. At Stanford, student inventors or faculty involved in a start-up may not negotiate with the university for the company unless they are on leave. No public source sets etiquette for an investor's first contact with a TTO, so the conservative route is the safe default.
The TTO route also answers your most important diligence question. Toronto notes that investors usually want to be sure the entrepreneur has an option or licence before investing. The office that holds the rights is the one that can tell you where they stand.
How findable the TTO route is
In Refolk's index of professional profiles, licensing and tech-transfer titles are a small population. University offices are a minority of it unless you filter.
| Market and filter | People with the title | University employers in a 25-profile top-employer sample |
|---|---|---|
| US, all industries | 1,242 | 3 of 25 (Stanford OTL 2, Columbia Technology Ventures 1) |
| UK, all industries | 789 | 2 of 25 (Aston, Brunel) |
| US, Higher Education | 142 | Majority of sample |
Source: Refolk's index. A UK Higher Education run was not made.
The university population is also concentrated. In the 25-profile Higher Education sample, Stanford OTL and the University of Virginia Licensing & Ventures Group supply 4 each, 8 of 25 between them, with the University of Idaho next at 3. "Licensing Associate" is the most common title, at 12 of 25. Large offices staff that role, and it is the most findable first contact. The US market has 1.57 times as many of these titles as the UK.
Building this list by hand means a title search, an industry filter and a check that each person still holds the role. With Refolk, I ask for licensing associates at university technology licensing offices that cover my thesis area and region, and I get back named people at the target institutions. I map them to the rows that need a contact route.
Subject: Early-stage investor interest in [technology area] at [university] Hello [first name], I invest at the pre-seed stage in [thesis area], and I follow work coming out of [university] in that field. I noticed [the I-Corps award / proof-of-concept award / published application] connected to [PI name]'s lab. I'm not asking for confidential information. I'd like to know whether the office would be open to an introduction to the team when the timing suits them, and what process you prefer investors to follow for teams that may be forming a company. If there is an option or licence process under way, I'm happy to wait until the team and the office are ready. A short call at your convenience would help me understand how your office works with outside investors. Thank you, [name, fund, one-line thesis]
Send to one named person per office; replace the bracketed fields with specifics from the watchlist row before sending.
How does the watchlist go wrong?
It goes wrong in seven predictable ways. Most come from treating the earliest-visible name or the latest-visible date as the real one. Each failure below has a cheap check. Run the checks before a row moves to "ready".
| Failure | What it looks like | Check |
|---|---|---|
| NIH I-Corps row is already a company | Award lists a firm, and the EL is a C-level executive | Search for a Phase I SBIR award to an existing company before adding the row |
| PI mistaken for the founder | Row names the professor as founder | Find the EL in the abstract or Hub announcements |
| EL assumed to be CEO | Early PhD student slotted as sole management | Flag rows where a hired CEO is likely |
| Patent date read as fresh | Approach timed from the publication date | Back-date about 18 months and ask the TTO about an option |
| Title search returns noise | Non-university employers in results | Always filter by industry |
| Stale AUTM proxy | Citing startup option counts as current | That question was discontinued after 2022 |
| Student contacted mid-negotiation | Direct message to an enrolled inventor | Route the first contact through TTO new-ventures staff |
The noise problem in detail
Title searches for licensing roles return people who share the words but not the job. In the UK sample from Refolk's index, 5 of the 25 top-employer results were Cheshire Police. Only 2 were universities. Without an industry filter, the contact-route column fills with people who cannot help you.
False positives on the signal side
- One-signal rows. A lone tech-transfer listing or a lone patent publication shows the university holds IP. It does not show a team intends to form a company. Keep these out of "ready" until a second signal or a named EL appears.
- Unconfirmed EL names. A name pulled from one abstract might be a past team member or a mentor. Confirm it in two places.
- Programme mix-ups. An Energy I-Corps team at a National Laboratory is not a university spinout, and its startup conversion is far lower. Label the source programme on every row.
False negatives
- Dropped late formers. Teams that incorporate after graduating show up as missing if you only watch the last 12 months. Keep the late-former tab.
- Provisional-only IP. A team with only a provisional filing has no public patent trace at all. Training and proof-of-concept awards are the only way you will see it early.
How do you keep the watchlist current?
Refresh it weekly, move rows by evidence rather than by age, and hand a row off once the company exists. This guide covers the stage before incorporation. Once a row incorporates, it belongs in formation-filing and founder-resolution tracking.
Stage moves to make at each refresh
- Watch to ready: two signals, a confirmed EL, and a licence status of disclosed or optioned.
- Ready to contacted: a named TTO contact has received the first note.
- Any stage to incorporated: the TTO startup page, a licence announcement or a formation record shows an entity. Hand it off.
- Any stage to cold: an option expired with no entity, or the EL's end date passed and they took another role. Move it to the late-former tab, not the bin.
Re-checking the facts this guide depends on
Several inputs change over time. Re-check the mechanism, not a remembered value.
- I-Corps cadence and lag. Confirm cohort counts and the gap between application and cohort on the NSF I-Corps pages each year.
- Startup counts. NSF recounts older cohorts in each biennial report. Watch for upward revisions as evidence of late formation.
- AUTM survey fields. Check whether the startup licence and option question has returned before you cite it.
- Proof-of-concept calendars. Confirm each target fund's round schedule, since Blavatnik moved to twice a year.
- Institution option terms. Re-read each target TTO's startup guide for option length, fees and extensions.
- TTO staffing. Re-verify named contacts before each outreach, because licensing associates move between offices.
Before calling the watchlist done
- The list holds 20 to 60 rows scoped to a written keyword list and 15 to 40 named institutions.
- No row comes from I-Corps at NIH, and each row is labelled with its source programme.
- Every row in "ready" has at least two independent signals.
- Every row names an entrepreneurial lead separately from the PI, confirmed in two sources.
- Every EL has a career stage and an expected graduation or contract-end date.
- Every row carries a licence status of none, disclosed, optioned or licensed, and optioned rows have an expiry date.
- Every patent signal records its estimated filing date, about 18 months before publication.
- Every formation window is a quarter range, not a single month.
- Every contact route names a TTO person found with an industry filter, not an enrolled student.
- The list carries a last-reviewed date from this week, and a late-former tab exists for older cohorts.
Questions practitioners ask
Where can investors find a list of I-Corps teams?
Start with NSF Award Search, which NSF itself points to for finding former I-Corps participants, and add Energy I-Corps cohort pages. Past I-Corps at NIH participants can be found in NIH RePORTER, but those are existing companies. The award record lists the PI, not reliably the entrepreneurial lead, and nobody has publicly documented whether the EL appears as a structured field. Read abstracts and regional Hub announcements for the EL's name and confirm it in a second source.
How do I identify the entrepreneurial lead behind a university startup?
Look past the award PI, who is the technical lead, and find the PhD student or postdoc who works directly with the technology in that lab. In FY23-24, 57% of National I-Corps entrepreneurial leads were graduate students. Check award abstracts, regional Hub or Site announcements, lab pages, and co-inventor lists on university-assigned patent applications. Record the person's career stage and expected graduation or contract-end date, because that date drives the formation window.
How early should I contact a university spinout team?
Make contact once a team has at least two signals and a named entrepreneurial lead, and route the first contact through the tech-transfer office's new-ventures or licensing staff. At Stanford, student inventors may not negotiate for a start-up while enrolled and not on leave. No public source sets investor etiquette for tech-transfer contact, so the conservative route is the safe default. It also tells you where the licence stands.
Are university patent applications an early signal of a spinout?
No. They are a lagging confirmation. US applications publish about 18 months after the earliest filing date, and provisional applications are never published. By the time an assigned application appears, the team may already have an option signed. Use patent publications to confirm IP ownership and to find student co-inventors, not to time your approach.
How much equity does a university take in a spinout?
It depends on the regime. The UK USIT guide sets starting university stakes of 10-25% for life sciences, and its software supplement recommends 5-10% with no anti-dilution before the first round and royalties of 0-2%. Some US universities use a model of about 5%, protected against dilution through Series A, plus royalties. Ask the tech-transfer office which policy applies before you price anything.
Try it on the search you came here for
Stop building boolean strings. Just describe the person.
Type one sentence. I plan the search, read GitHub, public LinkedIn and Crunchbase records, and the open web as it is right now, and hand back a ranked list with the reason next to every name.
01Describe them
One plain sentence. Role, city, stack, stage, whatever matters to you.
02I read the web live
GitHub, public LinkedIn and Crunchbase records, the open web. Not a database that went stale last quarter.
03You read the shortlist
Ranked, with the reasoning under every name. Open a profile, ask a follow-up, narrow it down.
- Staff backend engineers in NYC who shipped Rust in production
- Series A fintechs in SF under 50 people, growing headcount this year
- Maintainers of fast-growing Rust web frameworks on GitHub
- No boolean, no filters, no seat to buy. One box.
- Read at search time, so a profile updated yesterday counts today.
- Every step visible as it runs, every name with its reason.
500 free credits on sign-up. No card, no demo call. See real searches.