From challenge statement to a PoC your business unit actually signs.
EvalLens reads every challenge submission end to end against your criteria, hands your jury evidence-linked one-pagers, and produces a selection memo you can defend upstairs — every pick traced to a quote and a page. Your jury decides; you hold proof of process.
The funnel is fine. The screening is exposure
- 300+submissions a 6-week corporate challenge pulls, in the range we see across programs — screened by the innovation team on top of the day jobthe pilot measures yours
- 6 lensesa jury of business-unit executives scores through six different priorities — and the averaged spreadsheet hides every disagreementcommittee math
- ~1/3of an innovation specialist's week goes to manual diligence on inbound — a range we see across programsthe pilot measures yours
- “What shipped?”the question every steering committee now opens with. Activity metrics stopped counting; defensible selections and PoCs are the scoreboardyour last steering meeting
“Why these five?” The memo answers.
The moment the CFO asks why these startups made the shortlist — and what makes this year different — you open a memo where every pick has a quote and a page number. The “innovation theatre” line dies before it’s spoken.
Walk into the steering committee with receipts, not gut scores from six executives who read three decks on the flight in. Jury disagreements surface — nothing silently averaged. And every startup that didn’t make it gets a written, defensible answer — drafted from the record, approved by your team before it’s sent — including the one your CEO’s contact submitted.
Challenge close → scoped PoC, in six steps
EvalLens prepares the evidence. Your jury judges; the business unit gets a PoC brief the same week.
- 01
The challenge statement becomes the rulebook
Your search field and challenge statement translate into corporate screening criteria: use-case fit for the sponsoring BU, integration effort into your stack, security and compliance posture, references in regulated industries, procurement readiness. Criteria are versioned — when the BU sponsor changes priorities mid-challenge, you re-score against v2 with one action, and both versions stay in the record. You get: one rulebook the whole jury shares, versioned.
- 02
Submissions land from wherever they already land
Your branded challenge page, or a batch export from the Agorize/HYPE-class platform you run — CSV or files in, evaluations out. Applicants change nothing; multi-language submissions are read in the language they arrive in. You get: no platform migration, no retyping.
- 03
Every submission read against the statement
Every page processed, coverage logged. The use-case-fit reviewer cites the exact slide where the startup addresses your challenge statement — or flags that it never does. Submission #1 and #300 get the same standard. You get: your team screening exceptions, not the pile.
- 04
Your jury judges one-pagers, not homework
Independent AI reviewer roles — not people; six adversarial reads that can't anchor on each other — prepare evidence-linked briefs. Executives judge on the briefs; their scores are logged in committee mode, and their disagreements surface for the deliberation instead of vanishing into an average. You get: one exec-hour covering a briefed shortlist.
- 05
The selection memo comes from the live record
Winners approved by people, ranking built from jury scores, and a memo where every pick traces to evidence — generated from the record, not reconstructed the night before the steering committee. You get: board-ready proof of process.
- 06
The evaluation becomes the PoC brief
Strengths, open questions and verification steps hand straight to the business unit as the pilot's starting brief — same week, not next quarter. When evaluation is a standing process instead of a quarterly scramble, PoC approval stops being a committee-cycle problem — and the pilot measures your own before/after. You get: time-to-PoC as your KPI win.
What the pilot measures on your challenge.
Close → scoped PoC
Calendar days from challenge close to an approved PoC scope — before vs after.
Jury-hours per exec
What one executive actually spends to judge the shortlist, briefed vs cold.
Coverage
Share of submitted pages actually read — logged, verifiable.
Record completeness
Can every selection be answered with a quote and a page? Counted, not claimed.
Metrics agreed with you before the pilot starts. The readout is yours to take upstairs.
Your InfoSec review, with a folder — not a handshake.
Security pack up front, before any document moves.
Startup decks are confidential submissions under your program’s terms — and AI vendors are exactly what your vendor-risk process exists for. So the review comes first, with artifacts your team can actually check — the sub-processor list, DPA and security overview are already public on this site:
Two ways in, both procurement-friendly: a fixed-fee managed pilot — one live challenge, 4–6 weeks, run in parallel with your current screening, sized so most teams start it under a single PO — and an Enterprise track with SSO, SLA, your approved models, your servers and your retention policy. See pricing or book a call.
Asked in every steering committee
- Will our executives take scores from a machine?
- They don't have to — AI reads are advisory. Executives judge on evidence one-pagers, their scores build the ranking, and their disagreements surface for deliberation. What they notice is walking into jury day with every deck effectively read and the holes flagged.
- Our challenge platform already has AI scoring.
- Challenge platforms score signals and cluster ideas — useful workflow. They don't read a 20-page deck end to end and return per-criterion findings with quotes your jury can verify. EvalLens is that evaluation layer; keep your platform for workflow and community.
- What does InfoSec get?
- The security pack — architecture, data-flow, sub-processor list, DPA, contractual no-training commitment — before the pilot. Deployment follows your policy: managed perimeter today, your models and your servers on the Enterprise track.
- Procurement takes six months. How do we start?
- The pilot is a fixed-fee engagement sized under most single-PO thresholds, with MSA and DPA templates ready. Most teams run the pilot while procurement processes the production agreement in parallel.
- The BU sponsor changes criteria mid-challenge. Then what?
- Criteria are versioned. You update the rulebook, re-score the pool against v2 in one action, and both versions stay in the record — so the memo explains not just the picks, but the change.
- What do the startups' lawyers say about AI reading their decks?
- Submissions are processed only for your challenge's evaluation, never used for training, with retention you define — and we provide disclosure language for your challenge terms so that's transparent to every applicant upfront.
- Who else runs this?
- We're onboarding a founding cohort of programs; corporate references are shared under NDA where they exist. That's exactly why the engagement starts with a measured parallel pilot on your own challenge — the proof is generated on your data, not borrowed from a logo wall.
Pilot it on your next challenge.
30 minutes with you: we map the challenge statement to criteria, agree the pilot metrics, and hand your InfoSec the security pack the same day. The first run is free through August 31, for batches up to 10 decks.