Built in the open by Tuaha and Farzan. Marketing site landed two weeks ago. Self-serve beta opens after paid checkout approval, with the roadmap a quarter ahead so you can plan around what's landing.
Pulled from the build doc. Sequencing follows V2 then V3 in the product bible, with telemetry from each shipment refining priority for the next one.
Single Report and Builder Pack open for self-serve founders once paid checkout is approved. Everything we shipped pre-launch is in the box on day one.
Single Report
Paid verdicts at $14.99. The full five-stage pipeline, a named falsifier on the core claims, and a software-generated digital report.
Builder Pack
Three verdicts at $39.99. Run several ideas through the same five-stage pipeline. Side-by-side comparison is on the roadmap.
Voice-first intake
Answer brief questions by voice, with transcription as you go. Useful when typing slows you down.
Reasoning trace UI
Watch the pipeline live as sources are checked and claims held or dropped.
Founder workspace dashboard
The post-verdict home. Recently viewed verdicts and in-flight reports from one screen. Mobile layout included.
Sample verdict
A full report you can read before paying. Same structure, same source rigor, same defensible numbers as the version you'll buy.
Methodology, documented end to end
Five-stage pipeline on OpenAI's GPT-5.4 family, with Anthropic used for live web search. A tier-graded source library and the quality gates that run on every report. Bottom-up TAM rules and a named falsifier on the core claims.
Tier-graded source library
Official sources graded Tier 1 through Tier 3 by reliability, alongside open web retrieval. Reports typically cite 40+ sources.
Sub-processor list
Every third-party processor named on the security page with status, region, and purpose. We rely on our model vendors' standard API terms, accessed via the Vercel AI Gateway, under which API payloads are not used to train their models.
First wave after the beta opens. Telemetry from the launch cohort sets priority for what lands first.
Verdikt Agents
Five to nine named specialists matched to the idea you just validated. One-line install, ready to help you start the build the day the verdict lands.
Refinement chat
Ask follow-ups after the verdict lands. What if I targeted a different ICP. What if I dropped the second feature.
Credit packs
Pre-purchase verdicts in bulk and use them when you need them. Volume discount on a one-time payment. No subscriptions, no recurring billing.
Email and Slack notifications
Progress nudges as the pipeline runs. Off by default. The opposite of a notification-spam product.
The defensible second half of the year. Built on top of agency-agents, the validator council, and the API surface.
AI Validator Council
Seven named specialists each return an independent vote and reasoning. Disagreements get flagged, not buried under one confident summary.
Public report sharing
Custom slug and OG image so one link briefs a cofounder or collaborator in thirty seconds. Permissions-aware, not just a URL.
Pivot mode
Re-validate after a pivot with a side-by-side delta against the original verdict. Shows what moved and why.
Share mode
A sanitized version of any report for sharing with cofounders and collaborators. Numbers and citations intact, internal notes stripped.
API access
Programmatic verdicts for accelerators and platform partners. Same pipeline, structured output, your billing.
Exploring. No dates yet.
Things we'll commit to once V3 is in production and we've heard from the second cohort of teams using Verdikt as a primary research tool.
Every pre-launch ship since the marketing site went up. No retro-dating. No silent rollbacks. The version number is the date it shipped to production.
First look at the post-verdict home. Recently viewed verdicts and in-flight reports from one screen. Mobile layout ships at the same time so you can check status from anywhere.
Shipped by Tuaha & Farzan
First public view of the build doc. Six pre-launch ships tracked here, three quarters of roadmap, the same dates we work to internally.
Shipped by Tuaha & Farzan
Reports run on OpenAI's GPT-5.4 family, with Anthropic used for live web search. Five-stage pipeline, a tier-graded source library, and the quality gates that run on every report. The bottom-up TAM rules and our aim of a named falsifier for the core claims are now public.
Shipped by Tuaha & Farzan
New Tier 1 and Tier 2 official sources added to the graded library, alongside open web retrieval. Reports typically cite 40+ sources.
Shipped by Tuaha & Farzan
A full report goes public so founders and AI builders can read the output before they buy. Same structure as the paid version, same source rigor, same defensible numbers.
Shipped by Tuaha & Farzan
Every third-party processor in the stack listed on the security page with status, region, and purpose. Reports run on OpenAI's GPT-5.4 family, with Anthropic used for live web search. We rely on our model vendors' standard API terms, accessed via the Vercel AI Gateway, under which API payloads are not used to train their models.
Shipped by Tuaha & Farzan
Pricing, about, methodology, security, blog. The first version readable by founders and AI builders. Built by Tuaha Jawaid and Farzan Ansari.
Shipped by Tuaha & Farzan
If a teammate would pay for something on this page sooner, tell us. Requests from paying founders and builders move ahead of internal priorities.
