Voygr Launches Business Validation API for Fresh Place Intelligence

Voygr (YC W26) has launched API access for their real-world place intelligence platform, starting with a Business Validation API that addresses stale place data in AI applications. The founders, Yarik and Vlad, have backgrounds at Google Maps, Apple, Meta, and ridesharing/travel companies.
What Voygr Solves
The core problem: ~40% of Google searches and up to 20% of LLM prompts involve local context, but 25-30% of places churn every year. Traditional maps APIs provide fixed snapshots without detecting real-time changes like closures, rebranding, or operational status changes. Voygr's benchmark showed even the best LLMs get 1 in 12 local queries wrong.
Current API Capabilities
- Business Validation API: Tells you whether a business is actually operating, closed, rebranded, or invalid
- Input requirements: POI/business name and its address (exact address needed for businesses within plazas)
- Processing scale: Tens of thousands of places per day for enterprise customers
- Methodology: Aggregates multiple data sources, detects conflicting signals, returns structured verdict
The team describes their approach as "continuous integration, but for the physical world." They're not yet providing opening times or pictures - those features are in development under an experimental enrichment API not yet open to the public.
Technical Details
Voygr started with "one of the hardest parts - knowing whether a place is even real." Their system actively detects changes since "the world doesn't emit structured 'I closed' events." The API is available at github.com/voygr-tech/dev-tools with instructions defined for agents to onboard.
This infrastructure becomes increasingly critical as AI agents start searching, booking, and shopping in the real world - the founders estimate this problem gets 10x bigger with agent adoption.
📖 Read the full source: HN AI Agents
👀 See Also

Flash-MoE: Running 397B Parameter Qwen Model on MacBook Pro with Pure C/Metal
Flash-MoE is a pure C/Metal inference engine that runs Qwen3.5-397B-A17B, a 397 billion parameter Mixture-of-Experts model, on a MacBook Pro with 48GB RAM at 4.4+ tokens/second. The 209GB model streams from SSD through custom Metal compute shaders with no Python or frameworks.

Nelson: A Claude Code Plugin for Coordinating AI Agents Like a Naval Fleet
Nelson is a Claude Code plugin that structures AI agent coordination using naval fleet principles, featuring three execution modes, a risk classification system, hull integrity monitoring, and standing order gates to prevent common anti-patterns.

Focusmo macOS app adds local MCP server for Claude AI integration
Focusmo, a macOS focus app, now includes a local MCP server that allows Claude AI to access real focus data for weekly reviews and planning. The server runs locally on Mac with no external servers required, keeping all data on-device.

Drop-in OAuth Provider for Personal FastMCP Servers on All Claude Platforms
A developer created a single-file Python OAuth provider that enables personal FastMCP servers to work on Claude.ai web, mobile, and Desktop platforms without requiring external identity services like Auth0 or Google.