Solutions / MCP server
Ten engines as tool calls inside Claude
GEO Lens ships as a Model Context Protocol server. Claude, ChatGPT connectors and custom agents can query the full Chinese and Western engine roster and run share-of-voice audits natively, inside the workflows your team already runs.
Three tools, full roster
list_enginesThe roster with region, underlying model and cost, so agents can choose their panel.ask_enginesOne question across any set of engines, Chinese prompt to Chinese engines, English to Western, with optional source citations.geo_auditA full share-of-voice audit: brands with bilingual aliases, a prompt set, and back come mention rates, hemisphere splits, first-mention ranks and every underlying answer.
What that unlocks
Your comms agent checks how Doubao describes your CEO before the announcement goes out. Your research agent runs the same category question across ten engines and tables the disagreement. Your reporting pipeline appends a China-versus-West visibility section no other vendor can produce.
Deployment is remote MCP over HTTPS with per-client bearer tokens, or on-premises for teams with data-residency requirements.
# Any MCP client, one audit call geo_audit( brands={"YourBrand": ["YourBrand", "你的品牌"], "Rival": ["Rival", "对手品牌"]}, prompts=[{"id": "shortlist", "q_en": "Which vendors should I shortlist for X?", "q_zh": "X领域应该重点考虑哪些供应商？"}] ) # returns mention_rate, cn_rate, west_rate, share_of_voice, # avg_first_mention_rank + every verbatim answer
Access
The MCP server is offered under commercial license with per-client tokens and usage quotas. Write to rio.chen@mpowerlabs.ai for a token and a scoping call.