MetaTuner (competitor)
Alias / mishearing: Granola renders MetaTuner as “Meta View” / “Meta Tuner” / “Metaviewer” — same product.
Current truth
- Solo/part-time, Rotterdam-based (Netherlands). Founder is a “reasonably credible guy who understands the space” (Elliot’s read 2026-06-22); LinkedIn finally tracked down (premium profile, connected to some of the Alpic team). Only other person linked to it is an advisor/investor who is “on board.” Assessed as likely not his full-time job — but credible.
- Copied the Ghost Team website directly (the scrolling brand icons), which itself echoes Alpic’s. Confirms he’s watching us closely.
- What he actually does (our read): post-connection invocation optimization, NOT organic discovery. Despite website copy claiming he optimizes for the “app store” and “organic pickers,” he is not running live client sessions — he almost certainly calls the MCP server directly via API and tests prompts there. His loop: test a live app → read tool-call / invocation data → recommend changes to improve invocation → iterate.
- Product surface: an SDK / tracking tag you install in your MCP server; an auditor (paste an MCP URL or JSON → flags “critical issues,” e.g. “description must start with…”); golden-prompt sets (labeled positive/negative dataset graded on precision/recall — “tool fired when it should,” “negative prompts handled correctly”); invocation analysis / tool optimization that “tracks real-world selections” and spots drop-off. Developer-focused.
- Audit quality: thin / old-school per our review (“description must start with use this…” = a stupid, dated rule). We believe canonical description rules don’t hold (see Positioning).
Positioning vs us
- Different loop, complementary surface. MetaTuner optimizes “Used” (invocation quality once a connector is already connected); Ghost Team measures “Found” (organic pre-connection discovery / connection-picker visibility), which is only observable by running live client sessions — invisible in logs. He is thin on / not really doing organic discovery.
- This is exactly the post-connection optimization layer we have NOT built — a roadmap gap we now own (see roadmap). MetaTuner + Alpic’s intent→invocation feature are the comparables for it.
- Not viewed as a threat to lead-position; “good that someone else is thinking about this.” No interest in reaching out (we’re the category leader); Vincent floated coopetition / cofounder-genius possibility, Elliot declined.
Open questions
- Founder’s name + whether this is a real venture or a side project (LinkedIn screenshots TODO — web lookup deferred).
- Does he ever move into true organic-discovery measurement (live client sessions)? Currently no.
Timeline
- [2026-06-22] (internal discussion — Vincent + Elliot, website/strategy session) Re-found MetaTuner; mapped his loop as post-connection invocation optimization (SDK + golden prompts + tool-call data), not organic discovery. Noted the direct website copy. Used as the prompt to add post-connection optimization to our roadmap. Key shared insight: you can’t write canonical optimization rules — models keep changing and Anthropic itself doesn’t publish golden tips; optimization must be model-scoped (e.g. Haiku).