Every Applicant Scored.By The Same Rubric. On The Same Day.
Selection quality collapses under deadline volume. Zurvek runs your applications through one fixed adversarial rubric, ranks them by percentile inside your own cohort, and leaves a timestamped reason behind every score — so the shortlist survives the question "why this company and not that one?"
The Bandwidth Problem: why India's accelerators can't keep up with their own deal flow
India is not short of programmes or applicants. It is short of the reviewer hours to assess what arrives — and that constraint, not selectivity, is what decides who gets into a cohort.
Active incubators in India
Applications to seats, one 2026 AI cohort
Of Indian startups are ever incubated
Six Things That BreakIn Every Selection Cycle.
THE DEADLINE SPIKE
Applications arrive in a two-week wall. Your team reads the first sixty carefully and the last two hundred at a glance. The order a deck arrived in should not decide who gets an interview — right now it partly does.
SCORING DRIFT BETWEEN REVIEWERS
Two reviewers score the same deck four points apart, and neither is wrong — they weighted different things. Without a fixed rubric applied identically to every applicant, your shortlist reflects reviewer temperament as much as company quality.
NO PAPER TRAIL FOR A REJECTION
A rejected founder asks why. A sponsor asks how the cohort was picked. Most programmes can produce a spreadsheet cell, not a reasoned record. Every Zurvek evaluation is timestamped and cites the page it came from.
THE THIN APPLICATION PROBLEM
Half your applicants have no revenue, no cap table, no audited anything. Scoring them on completeness punishes the earliest companies — often the ones your programme exists for. Missing data has to be flagged, not guessed at and not counted against them.
HEURISTIC FILTERS IN THE MIDDLE OF THE PACK
Cutting the obvious no's on pattern-match is fast and defensible. The problem is the large middle of the distribution, where the fast filter is a coin toss and the differentiated company looks like everything else at a glance.
REPORTING COHORT QUALITY UPWARD
Sponsors, university partners and LPs want evidence that batch six was stronger than batch five. Reviewer notes do not aggregate. Consistent structured scores across every applicant do.
From Intake To ShortlistIn Five Steps.
UPLOAD A LOT
Drop an XLSX or CSV of applicant rows — one row per company, columns read straight from the Zurvek template — or upload pitch decks as PDFs (10 MB each). Text is extracted in your browser before anything is sent. Current lot size is 20 companies; run as many lots as your plan's evaluation count allows.
TRIAGE EVERY APPLICANT
Each company gets a triage verdict and score against the same rubric — same weights, same questions, same order, regardless of when the application arrived. Missing fields are flagged as dark data under the Zero-Inference Protocol: never estimated, never silently averaged.
RANK THE COHORT
Ranked table with sort, verdict filter and a per-company review drawer. Switch to cohort mode to read percentile position inside your own batch — top 10%, top 25%, top half — so a weak intake does not produce inflated absolute scores.
ASSIGN AND REVIEW
Assign companies to reviewers, mark them reviewed with a note, and track batch health as the panel works through the list. Export the whole batch as CSV for your own committee pack.
ESCALATE YOUR FINALISTS
Open a shortlisted company and run the full five-agent adversarial analysis for one evaluation. That produces the interview questions written against that company's unresolved flags, and the six-part Investment Memo you take into selection committee.
Spreadsheet Intake
Most programmes already hold their applications in a sheet. Export it, map the columns to the Zurvek intake template, and upload the file — XLSX, XLS or CSV. Each row becomes an applicant and is triaged against the same rubric as a full deck. Applicants who submitted decks can be uploaded as PDFs in the same workflow.
SUPPORTED
XLSX · XLS · CSV · PDF
PER LOT TODAY
20 companies · unlimited lots
PER PDF
10 MB · text extracted in-browser
Download the intake template — the same column set the batch modal expects.
What a lot looks like after triage — ranked, banded by cohort percentile, flagged, and ready to review.
BATCH
COHORT 07 INTAKE
SOURCE
XLSX · 20 ROWS
INVEST
1
WATCH
3
PASS
2
DARK DATA
23 FIELDS
"Clinical partnerships are real and verifiable; the revenue line is a pilot line presented as recurring."
Strengths
- Two named hospital deployments
- Regulatory pathway documented
Deal-breakers
- Pilot revenue booked as recurring
- No churn data at all
Dark data
- Cap table · not supplied
- Burn rate · not supplied
Illustrative lot. Cohort bands are percentile positions inside your own batch, not a global curve. Escalating a company to full five-agent analysis consumes one evaluation from your plan.
What You GetToday.
BATCH TRIAGE
20 companies per lot, unlimited lots. XLSX, CSV or PDF intake. Ranked scoring table with a review drawer per company.
SPREADSHEET INTAKE
Export your application form to XLSX or CSV, map it to the Zurvek template, upload. No re-keying, no per-deck upload for text-only applications.
COHORT-RELATIVE RANKING
Percentile position within your own batch alongside the absolute score, so cohort strength is read in context rather than against a global curve.
ZERO-INFERENCE PROTOCOL
Every missing data point is flagged as dark data. No assumed averages, no invented figures, no penalty for being early — and the absence list is itself a screening signal.
ADVERSARIAL INTERVIEW QUESTIONS
For each escalated finalist, questions generated from that company's specific unresolved flags — not a generic template your panel has asked forty times.
INVESTMENT MEMO
Six-part committee-ready memo per finalist: verdict, opportunity, forensic findings, risk matrix, conditions, analyst note. Exportable as PDF.
SECTOR AND STAGE CONTEXT
Scores are read against stage and sector benchmarks, so a pre-seed deeptech application is not measured against a Series A SaaS company.
REVIEWER ASSIGNMENT
Assign applicants to named reviewers, track what has been reviewed, and see batch progress and health while the intake window is still open.
BATCH ANALYTICS EXPORT
Score distribution, verdict split and flag density for the whole lot, downloadable as CSV for your own reporting.
IMMUTABLE AUDIT TRAIL
Every evaluation is retained with its inputs, its verdict hash and its timestamp, so a selection decision can be reconstructed months later exactly as it was made.
What ZurvekDoes Not Do.
- It does not tell you to admit or reject anyone. It produces a scored, reasoned record; the decision stays with your committee.
- It does not predict returns or survival. Nobody can, and the research on accelerator selection is explicit about that.
- It does not certify compliance with any scheme or regulator. It is structured to support documentation practice, not to replace it.
- It does not fill in missing data. An unknown is reported as unknown.
CustomisedFor Your Programme.
Three things we set up per programme, quoted individually. Write to support@zurvek.com with your intake volume and cohort calendar.
CUSTOM RUBRIC WEIGHTS
Your selection criteria mapped onto the scoring dimensions instead of the default weighting — so a deeptech programme is not scored like a consumer accelerator.
Not a new scoring engine: the same evaluation dimensions, reweighted to your published criteria.
PROGRAMME-BRANDED COHORT REPORT
The batch analytics output rendered as a report you can hand to sponsors, university partners or LPs — score distribution, verdict split, dark-data density across the intake.
Reports what your batches actually contained. It does not benchmark you against other programmes.
ASSISTED INTAKE SETUP
We set up a historical cohort as your first batch, map your application form to the intake template, and walk the divergence between the ranking and the cohort you actually picked.
A working session, not a consulting engagement. Scoped per programme.
Early Access Rollout
Rolling out- 100 companies per spreadsheet lot — larger XLSX intake for programme accounts, being enabled progressively. Today's lot size is 20.
- Bulk escalation — select several finalists and run full analysis across all of them in one action.
- Application form ingestion — direct intake from your application platform instead of a manual export.
Progressively enabled for early access accounts. Not generally available yet.
Programmes on Syndicate can ask what is currently enabled on their account before a cohort intake window opens.
The QuestionsProgramme Directors Ask.
What does this cost per application?
On the Syndicate plan at ₹34,999/month you get 300 evaluations — about ₹117 per company evaluated. A batch triage entry and a full escalation each consume from that count.
Half our applicants barely have a deck. Does that break it?
No. Thin applications are the normal case at accelerator stage. Missing fields are marked as dark data and reported as unknown. The score reflects what was evidenced, and the report tells you exactly what was absent — which is itself a useful screening signal.
Does this replace our reviewers?
It replaces the first pass. Zurvek produces a consistent scored ranking and a documented reason for every applicant; your team spends its hours on the shortlist and the interviews, not on reading deck two hundred at midnight.
Our applications live in a spreadsheet, not in decks. Does that work?
That is the faster path. Export your form responses to XLSX or CSV, map the columns to the intake template, and upload the sheet directly. Each row becomes an applicant with the same rubric applied.
What happens to the decks we upload?
Deck text is extracted in your browser. Documents are processed for your evaluation only, are not shared with third parties, and are not used to train third-party models. Full detail is on the Security page.
Can we try it on a batch we have already decided?
That is the recommended way to start. Run last cycle's applications, compare the ranking to the cohort you actually picked, and see where the two disagree. Write to support@zurvek.com to set that up.
Syndicate — ₹34,999/month
300 evaluations per month, roughly ₹117 per company evaluated. Sized for a programme running two to three cohorts a year. Start with a batch you have already decided and compare the ranking against the cohort you picked.