VentureLens

Real-time billing: a $299 charge auto-classified as related-party (not arms-length) by our Finance Agent, per competition rules.
Public /impact page: 13 arms-length customers, $2,887 revenue, 165 applications screened by Gemini — verifiable via the JSON API.
AI Operations dashboard: 34,579 Gemini-agent actions, 99.9% decided by AI (L2–L3), 142.9 human hours displaced, 175 Gemini API calls.
Live Gemini screening: PayLink MENA scored 93/100 with a structured rubric breakdown across Team, Market, Traction, and Innovation.

Inspiration

Incubators and accelerators receive hundreds of startup applications every cohort. Most programs still review them manually, spreadsheets, email threads, and weeks of committee meetings. Founders wait too long for answers. Reviewers burn out. Smaller programs without enterprise budgets are locked out of AI-powered tools entirely.

Our team has spent years building incubation software through Gohorto, serving 500+ organizations and 20,000+ startups. When the Build with Gemini XPRIZE opened, we asked: What if we built a focused, Gemini-native product that solves the single hardest bottleneck, application screening, for programs that can't afford a full platform?

That question became VentureLens.

What it does

VentureLens is an AI-native SaaS tool that helps innovation programs evaluate startup applications in minutes instead of weeks.

Program managers can:

Accept applications via web forms or PDF/pitch deck upload.
Run every submission through Gemini for structured scoring, risk flags, and plain-language summaries.
View every applicant's Gemini score, risk flags, and plain-language summary in a structured admin dashboard, so committees can review and compare candidates without digging through raw submissions.
Send AI-drafted feedback emails to founders (reviewed by staff before sending).
Track screening throughput, API usage, and agent execution logs in a production dashboard.

Every inbound application is processed by Gemini before a human reviewer sees it, AI runs the operation, not just a feature on the side.

How we built it

VentureLens was built as a new product during the hackathon window (May - August 2026) by a team with deep incubator domain expertise.

Stack:

Gemini API: core screening engine (at least one LLM call per application in production).
Google Cloud Run: API and web app hosting.
Google Cloud SQL: application and evaluation data.
Google Cloud Storage: pitch deck and document uploads.
Laravel + Vue.js: backend and frontend.
Stripe: subscription and cohort-package billing.

AI-native operations:

Gemini screens 100% of inbound applications automatically.
AI generates committee briefs and founder communication drafts.
Production dashboards log every Gemini API call, screening result, and agent action.

Pre-existing work disclosure: Team members bring incubator domain knowledge and open-source framework experience from Gohorto. VentureLens is a separate product, new brand, new codebase, new customer relationships, created after the submission period began. Standard frameworks were used as boilerplate; all Gemini integration and screening logic is original work from this build window.

Challenges we ran into

Fair, explainable scoring: Startup evaluation is subjective. We had to design rubrics that work across different program types while giving Gemini enough structure to score consistently and enough flexibility for program managers to override.
Multilingual applications: Our target market includes MENA and Turkey. Applications arrive in English, Arabic, and Turkish. Gemini handles multilingual input well, but normalizing outputs for mixed-language committees required careful prompt engineering.
Trusting AI with real decisions: Program directors are accountable for who gets into a cohort. We learned that AI must "assist" decisions, not replace them, so every score comes with reasoning, and humans always have final say.
Building a real business in 90 days: The competition requires real users and real revenue, not just a demo. That pushed us to ship fast, talk to customers early, and iterate on pricing (cohort packages vs. monthly subscriptions) within the first weeks.

Accomplishments that we're proud of

Shipped a production-deployed product on Google Cloud with Gemini running core screening workflows, not a prototype, not a slide deck.
Turned years of incubation domain expertise into a focused product that programs can adopt in days, not months.
Built an AI-native operation where Gemini handles screening, reporting, and communication drafting end-to-end.
On track to serve real programs with real paying customers during the competition window.
Made AI-powered application screening accessible to smaller programs that could never afford enterprise incubation software.

What we learned

AI-native means AI runs the business: Judges don't want a chatbot bolted onto a form. They want to see Gemini executing key operational decisions in production, with logs to prove it.
Domain expertise is a moat: Knowing how incubators actually run cohorts, evaluation criteria, committee dynamics, founder communication, let us build something useful in weeks that a generalist team couldn't in months.
Revenue validates the problem: When a program director pays $199 for a cohort screening package, that's stronger proof than any benchmark score.
Gemini's document understanding is a game-changer: Parsing unstructured pitch decks and extracting evaluable content multimodally is exactly what manual reviewers struggle with most.

What's next for VentureLens

Scale pilots: across incubators in Turkey, MENA, and Europe through direct outreach and incubator network partnerships.
Freemium tier: 5 free screenings to onboard new programs, converting to paid cohort packages ($199–$799/mo).
Expand AI operations: mentor matching post-selection, automated progress check-ins, and AI-generated program impact reports.
Integrate with larger ecosystems: VentureLens as the lightweight entry point; Gohorto as the full platform for programs that outgrow it.
Category impact: help 1,000+ programs screen 100,000+ applications fairly and fast, so more founders get the chance to build businesses and create jobs.