See what's holding your stack hostage
Eight quick questions about your AI stack, then a radar of exactly where the lock-in lives — model, data, tooling, infra. Every red flag names its mechanism, and every mechanism gets a concrete swap.
Eight honest questions about your current AI stack. The radar fills in live.
0/8 answeredCould you serve tomorrow's traffic from a different model than today's?
Do you depend on a capability that exists only inside one provider?
If you switched embedding models, what happens to your vector store?
Do you control which region your prompts and outputs are processed in?
How are your tool / function definitions written?
Where do your prompts live?
How does your code actually call the model?
Have you ever rehearsed leaving your main provider?
Bigger shape = more locked in.
Answer all 8 questions to get your sovereignty score.
Every red flag names the exact mechanism creating the lock-in — a proprietary tool schema, a re-embedding bill, a closed wire format — instead of a vague avoid-vendor-lock-in slogan. The reveal is structural, not motivational.
What actually creates lock-in in an AI stack?
Named mechanisms, not vibes: tool definitions written in one vendor's proprietary schema, prompts tuned to one model's quirks, embeddings that would need a full re-embed to move, and SDK calls scattered through the codebase. Each one is a separate exit cost you can measure.
How does MCP reduce tooling lock-in?
MCP is an open interface: you define a tool once, in a neutral format, and any client that speaks the protocol can use it. Your tools stop being written in one provider's dialect, so they outlive any single vendor decision.
What is exit cost, and why score it before you need it?
Exit cost is the time and money it would take to move a workload to another provider or an open-weights model. Measured cold — mid price-hike or mid deprecation-notice — it is always higher. A rehearsed exit is also negotiating leverage, even if you never use it.
This playable is part of the registry: its tool page
Next: catch the confident fake
Spot the Hallucination deals you fluent AI-style passages where exactly one sentence is made up. Practice the reflex that separates fluent from true.