Overview
PAVI — the Philippine AI Visibility Index — is the AI visibility intelligence layer for Philippine and SEA brands. It measures how brands appear across 10+ AI models — ChatGPT, Claude, Gemini, Perplexity, and more — and tracks visibility scores, industry rankings, sentiment, and buyer-journey coverage in both English and Tagalog.
Built by ThirdCode as a product, a data asset, a GEO framework, and the clearest proof that the team can architect intelligence systems from scratch.
The problem it solves
Most Philippine brands are invisible to AI. When their customers ask ChatGPT or Gemini for recommendations, they don't appear. Nobody has built a tool to measure this — and without measurement, there's no strategy.
The two technical problems worth naming:
- Prompt distribution. A representative set of real Filipino user prompts, per industry, across language registers — not a translation of US templates.
- Attribution integrity. LLM responses change between runs. The measurement layer has to tolerate stochasticity without smoothing out genuine shifts in visibility.
What makes it different
PAVI is the first AI visibility index built specifically for the Philippines and SEA. It does not just report scores — it generates the strategic intelligence brands need to fix their AI visibility gaps.
- LangGraph orchestrates multi-turn probe conversations per engine.
- FastAPI workers issue the calls and normalize outputs.
- pgvector handles semantic matching across paraphrased mentions.
- Postgres holds the tenant data; React 19 reads with row-level security.
Outcome
PAVI tracks visibility for two dozen Philippine brands across five industry verticals, with measurable shifts already documented after brand-side content changes — the loop works.
The platform shipped as a SaaS at pavi-delta.vercel.app.