Getting started: from an empty project to a ranked field
Before you start#
- Access. EvalLens is available through a limited partner program — there is no public sign-up. Access for your team is set up after a partner call.
- A rubric decision. You can run the default six-dimension pitch rubric unchanged, or edit the weights. Decide before the batch runs: weights lock when scoring starts.
- The field. Decks as PDF, PPT, PPTX, or a Google Slides link — one deck per team, up to 50 MB.
- Your judges. Who sets the Jury Score, and when they will be in the room.
Sign in with email and password or Google OAuth. You land in your own workspace: your projects and their status, with a place to start a new one.
Choose the mode before the wizard opens — Pitch Competition runs a six-judge panel, Hackathon runs five. Then work through the five wizard steps: details, criteria and weights, judges, how decks are collected, and a final review.
Full walkthrough: Set up a project.
Two options, and they can be mixed. Add teams by hand from the project page — name, project, deck, and an optional note for the jury. Or publish a submission page and share one link, so teams upload their own decks and sign in with Google.
Full walkthrough: Collect submissions.
The project page shows what is ready, what is incomplete, and what still needs attention. Evaluation can start once at least one entry is ready; the whole batch is processed in parallel rather than deck by deck.
Each deck passes the same five stages — Decoder, AI Judges, Summarizer, Scoring, Report. See Run an evaluation.
Each team comes back with a three-layer report: a project summary, an AI Score Report showing what shaped each dimension, and ranked questions for the live Q&A. Every finding cites the slide it came from.
Start with the decks the report flags: high score with weak evidence, and dimensions where the judges split. See Read a report.
In the Review Board, score each criterion from 0.0 to 10.0. The AI Total Score stays read-only beside your number. Once every startup is scored, generate the ranking — it is built from Jury Scores weighted by your criteria.
Full walkthrough: Score and shortlist.
What you have at the end of a run#
| Output | What it is for |
|---|---|
| Structured workspace | Every application in one place instead of scattered files and threads |
| Deck-level evaluations | Scores across P1–P6 with a confidence signal on each |
| Judge assessments and evidence | Each assessment tied back to the slide it came from |
| Strengths, weaknesses, gaps | Alongside deck-completeness signals per section |
| Leaderboard and comparison | Ranked by your Final Score, not by the AI score |
A safer first run#
If this is the first time EvalLens touches a real result, run it in parallel instead of in place. Send a field you have already judged — last year's cohort, or this year's batch while your screeners work as usual — and compare the panel's read against the placements you already know. Nothing counts until you decide it does.
Next steps#
- Core concepts — the vocabulary the rest of these pages uses, including the three different scores.
- Brief your jury — the one intervention with published evidence behind it.
- What EvalLens does not do — read this before you promise anything to a stakeholder.