01 · Search
The entry point is intentionally minimal: one search field, suggested decisions, and a clear statement of what Vera does.
Founder · AI Product · 2026
Vera.
An agreement engine designed to show where the internet agrees—and where it does not—before someone makes a decision.
The project
A new way to understand what the internet actually agrees on.
Search engines return links. Review sites return rankings. AI assistants often return a confident answer. Vera was built to do something different: examine evidence across the web and reveal whether a real consensus exists.
The product is intentionally direct. A user asks a question, Vera gathers and evaluates evidence, then returns a clear consensus classification, an explanation, leading contenders, and the sources behind the result.
Role
Founder, product strategy, UX, visual design, and development
Stack
Next.js, OpenAI, Tavily, Supabase, Vercel
Focus
Consensus classification, evidence quality, trust, and
decision support
Status
Live product
The engine
From an open-ended question to an evidence-backed verdict.
Vera uses one central consensus engine with category-specific validation rules layered around it. The goal is not to produce the most confident answer. The goal is to produce the most truthful answer the available evidence supports.
01
Understand the intent
Vera determines what kind of decision the user is making and routes the question through the correct evidence rules.
02
Retrieve and validate evidence
The system gathers sources, balances domains, extracts real contenders, removes duplicates, and rejects invalid entities.
03
Classify the consensus
Vera determines whether the evidence supports a clear consensus, a split decision, or no reliable consensus.
04
Explain the result
The final page presents the verdict, reasoning, leading contenders, trust signals, and supporting sources.
Product screens
A quiet interface built around one clear decision.
02 · Verdict
Vera does not force a winner. When credible evidence supports multiple options, the result says so clearly and explains why.
03 · Contenders
Evidence-backed contenders remain useful even when the internet is divided. Each option can be explored, understood, and saved.
My contribution
The full product—from positioning to production.
Vera required more than a polished interface. I designed the product concept, trust model, evidence pipeline, consensus logic, category rules, responsive experience, brand, infrastructure, and launch system.
The product was repeatedly tested against real queries across software, products, local recommendations, travel, destinations, providers, and comparisons. Every refinement was made to improve usefulness without weakening trust.
Live now