Speaker introduction

Why we did this research

Will Bass Jr. — Co-Founder, Nexus Core · ran all 50 interviews · West Point · 10 years U.S. Army · operations leadership in banking and consulting · NSF I-Corps mentor
Priya Kundu — Co-Founder & CTO, Nexus Core · building the Founder Continuity Record: ingestion, identity resolution, outcome linkage, consent
50 interviews, March–August 2026, across three continents · 39–40 with verbatim transcripts
Research first. Product second. Scope cut against the evidence before a line of product code.

"We sat in every seat in this ecosystem before we studied it — and we let the evidence cut our own product."

50 interviews · three continents · March–August 2026

35+

of 50 interviewees confirmed the graduation cliff: within weeks of graduating, the institution that invested in a founder no longer knows where that founder is.

The reporting problem isn't a reporting problem. It's an incentive problem.

The evidence

The industry runs on a data model that mistakes optimism for progress

Signal saturation across corpus (n = 50 · counts from verbatim transcripts)

Graduation cliff

35+

Checkboxes over outcomes

dominant

Self-reporting broken

dominant

Data fragmentation

strong

Mentor sessions don't roll up

strong

MD bottleneck

strong

The top three signals are one underlying mechanism, not three problems. Three dominant signals held their threshold through the September reconciliation of the corpus to 50; exact per-signal tallies are being re-run against the verbatim set and are shown here as bands, not decimals.

The reframe

Self-reporting and the graduation cliff are one problem, not two

"The program only knew what I wanted them to know, and usually all the best things. Mentors knew the bullshit. You're constantly getting assessed, wrapped and stacked. I would see founders that would let all the dirty laundry out, and it definitely affected their standing in the program."

— Founder · dual participant in two nationally recognized accelerators

Good news gets volunteered. Bad news goes dark. The cliff is the assessment dynamic losing its last enforcement mechanism.

Falsification

Behavioral fixes have been tried. They don't work.

Intervention	Confirming source	Result
Equity stake	Regional venture-fund partner — "Ironically, no. It's basically the same latency."	Did not close gap
Military accountability	Federal service-academy faculty — structural exit severs the link regardless of motivation	Did not close gap
Federal compliance	NSF I-Corps ecosystem operators — mandated reporting produces compliance, not visibility	Did not close gap

Competitive shift

AI has ended the accelerator-software category

"We just build our own" — three of our most experienced operators, independently
The closest platform attempt was acquired by a major bank and shut down
Purpose-built ecosystem SaaS has failed to win the category
The competitor is not another vendor. It's the operator with a weekend and an AI assistant.

Category viability under AI-assisted self-build

Cost to replicate with AI → Differentiation →

The open wedge General-purpose incumbents Homegrown builds Purpose-built SaaS · dead zone

The buyer

The person feeling the pain is not the person with the budget

Who feels the pain — the operator

Who holds the budget and the mandate — the sponsor

Loses the founder the day the program ends

Funds many programs and cannot see across them

Assembles every funder answer by hand — "hero work, every single time"

Is measured on founder outcomes it cannot currently produce

Counts what it can count: attendance, presenters, volunteers

Must defend renewals, boards and audits with evidence, not anecdote

Has no budget line for another tool — and will build its own

Buys for political defensibility, not operational efficiency

The category has been sold to the wrong seat for a decade. The seat that can pay has a completely different reason to buy.

Segmentation

The pain is not universal — it is concentrated

Four low-pain counter-examples in the corpus — a single-state foundation, a state agency, a high-density community network, a multi-year-continuity accelerator
They share personal-knowledge substitution: small footprint with high network density, or continuity built into the program model
They are not the customer. They are the boundary condition — preserved at full weight, not discarded.

Concentrated severity — where the ICP sits

Belief vs. evidence

Five industry beliefs the evidence contradicts

What the industry believes

What 50 interviews revealed

Better dashboards will fix visibility

Dashboards are not the constraint — founder participation is, and it drops under assessment

Alumni loss is about operator bandwidth

It's about assessment ending — no residual reason to disclose

Equity stake improves reporting

Tested and falsified — same latency either way

Post-graduation is when the data gets interesting

It's when the data goes dark

The pain justifies purchase

Chronic and absorbed, not acute — "nice to have"

The open space

Five things nothing on the market delivers

Continuity across the graduation boundaryNo vendor delivers

Identity of the same founder across programs, cities and company namesNo vendor delivers

Outcome evidence from sources the founder does not controlNo vendor delivers

Over-the-top integration that doesn't demand migrationNo vendor delivers

The sponsor-of-multiple-programs viewAttempted once for venture-backed cos · shut down

The gap is coherent. All five cluster into one thesis.

The honest counter-case

Three findings that cut against the commercial premise

⚠

WTP is unproven

The most credible interviewees were the most skeptical anyone will pay. Enthusiasm is not purchase intent.

⚠

Self-build is strengthening

The moat is collapsing in real time. A weekend and an AI assistant replicate most feature sets.

⚠

The buyer is fractured

Funders sometimes have structural reasons not to want founder-level data.

One investor was pitched this product and passed: "We didn't think it would make that much of a difference." — angel network board member, Midwest

What to do

Three strategic moves — one for each seat

Operators

Invert the data flow

Extractive systems get gamed. Give founders a reason to disclose — value back, not surveillance out.

Funders

Ask for outcomes you did not have to request

Attribution is unprovable. Independently evidenced outcomes are not. Fund continuity, not cohorts.

Founders

Keep your own record

Programs that ask for weekly data without giving anything back are not partners.

The category resets when all three seats change their default.

Discussion

The conversation matters more than the consensus

If you run a program

What percentage of your cohort is telling you the full picture?

If you fund programs

What would you do with outcome evidence you can't do with activity reports?

If you went through one

How much of what you told your MD was true?

If you build in this space

How do you defend a position AI can now replicate?

Appendix A1 · held for Q&A

How the research was conducted

Persona IDOperator · funder · investor · mentor · founder · skeptic

Tailored questionnaireOpen-ended, non-leading, no product shown

Recorded interview50 sessions, Mar–Aug 2026 · 39–40 verbatim

Primary transcriptSynthesis from re-read, not derived summaries

Signal taggingConfidence-calibrated: validated / emergent / falsified

Several apparent "buyer validations" in early materials were found on re-read never to have had the purchase question put to them — and were downgraded to problem-alignment.

Appendix A2 · held for Q&A

Who was interviewed

50
interviews
roles overlap

Program operators / MDs / EDs — 11 Ecosystem builders — 10 Institutional funders (state · federal · foundation) — 9 Founders (active + wound-down) — 8 Mentors / EIRs — 4 · Investors — 3 Methodology review — 1 · unconfirmed short sessions — 4

Geography: US Northeast, Mid-Atlantic, South, Midwest · United Kingdom (Scotland) · Estonia · Latin America. International recurrence rules out a US-only artifact.

Appendix A3 · held for Q&A

How confident we are, finding by finding

VALIDATED — multiple independent confirmations

Graduation cliff 35+ / 50 · saturated ~I30
Checkboxes over outcomes dominant
Self-reporting broken dominant
Fragmentation · mentor black box · MD bottleneck · late discovery strong · re-tally pending

EMERGENT — strong but few-source; probe first

Cross-organization handoff gap · mentor-selection · integrity vetting 1–3 sources
National SBDC channel 2 independent sources · corroborated

FALSIFIED — tested and rejected

Pure-passive collection (3 founder rejections) · equity-fixes-visibility ("same latency") dead
In-program pace measurement as the product — our own July thesis: substrate refused, "the program is 5% of everything," untraceable scores indefensible dead

Appendix A4 · held for Q&A

The five hardest questions — and straight answers

Isn't this selection bias — you interviewed people who have the problem?

We ran a deliberate skeptic track and preserved the four no-pain counter-examples at full weight. The boundary condition is in the findings, not buried.

50 interviews — is that enough?

Qualitative saturation reached: the top signal stabilized around interview 30. New interviews now refine segments, not findings.

Isn't the fix just better software?

Three structural tests — equity, military accountability, federal compliance — all failed. The mechanism is incentive, not tooling.

So who pays?

Honest answer: unproven. The sponsor that funds multiple programs is the credible candidate. One national foundation has offered design input and described its own purchase path — appetite, not commitment, and reported as exactly that.

If the opportunity is real, why hasn't anyone built it?

Someone did. A major bank acquired the closest attempt and shut it down. That proves the market attracts serious players and that execution is hard. It also served venture-backed companies only — the sponsor of pre-institutional founders was never its customer.

Appendix A5 · held for Q&A

What this research does not yet know

Limitation	Impact if wrong	Planned next step
Corporate-accelerator seat unrepresented (N≈0)	MED	Entry point identified; interview scheduled
Pure-investor seat under-represented	HIGH	Targeted investor outreach in buyer sprint
Sample skews English-speaking	LOW	Non-anglophone ecosystem interviews queued
WTP not directly tested with every interviewee	HIGH	Dedicated buyer-and-payment discovery sprint
SBDC channel rests on one strong source	MED	National association + platform referrals
Founder sample skews equity-program alumni	MED	Grant-program founder recruitment

Appendix A6 · held for Q&A

The incumbents read the funded few. The open ground is every founder a sponsor touched.

Platform	What it reads	Whom it covers	The question it answers
Aumni (J.P. Morgan · shut down)	Post-closing legal documents — executed SAFEs, priced rounds	◷Venture-backed companies only	"What were the exact terms of what already happened?"
Carta & cap-table ledgers	Administrative records of equity and compliance events	◷Companies with a cap table to administer	"Who owns what, as of the last filing?"
CRM-replacement platforms for economic developers	Whatever operators migrate and re-enter	◷Programs willing to switch systems	"What did we write down this time?"
General-purpose stack	Whatever operators manually enter	◷One program, as of the last entry	"What did we last write down?"
AI-assisted self-build	Whatever an operator wires it into — "the magic is just getting the data right"	◷One program; no identity precision, no external outcomes	"What can I see across my own tools?"
Nexus Core	The sponsor's existing system of record (read-only) + outcome events from sources the founder does not control	⚑Every founder a sponsor's programs touched — across programs, cities and years	"What became of the companies we funded — and can we show the evidence?"

The limit is structural — a system whose substrate is the term sheet cannot see a portfolio that is overwhelmingly pre-institutional.

Appendix A7 · held for Q&A

Self-reported versus independently evidenced — the map that matters

Positioning: coverage × evidence source

Coverage: funded few → every founder touched Evidence: self-reported → independent

Evidenced outcomes, whole portfolio · unoccupied Aumni · Carta — funded few only CRMs · forms · self-report DIY / AI self-build

Strengths

Read-only on systems already in use · identity resolution across programs as the core competence · outcome evidence from independent sources, not self-report · every fact resolves to a source record · consent built for institutional review

Weaknesses

Pre-build — first pilot Q1 2027; the artifact under evaluation is a synthetic seeded tenant · identity precision at the bar a sponsor accepts is the technical risk · external outcome coverage unproven at scale · WTP expressed as appetite, not commitment

Opportunities

The dark zone: the post-graduation window where 35+ of 50 interviewees confirmed visibility collapses — invisible to every incumbent because none reads the founders who never raise · sponsors measured on outcomes they cannot see · one relationship = dozens of programs

Threats

Self-build replicates any feature — defensibility must be demonstrated in identity precision, outcome coverage and consent that passes security review · incumbents bolting an "outcomes" tab onto a ledger · category scar tissue from a prior shutdown whose customer was never this buyer