Chokepoint Intel AI
Chokepoint Intel AI
Find what constrains AI, and test who controls it.
Most AI-investing content tells you what a famous voice believes. Chokepoint Intel AI shows you what the evidence supports. Give it a theme, a company, or a claim you saw online, and get a structured, source-tagged research brief you can check line by line.
What it does
| Workflow | You give | You get |
|---|---|---|
| Chokepoint Scan | A theme (AI power demand, HBM, optical interconnect, cooling) | Supply-chain map, 6-factor bottleneck scores, research candidates, failure modes, verification plan |
| Claim Audit | A thread, article or thesis | Atomic claims, red flags, verdict and confidence grade per claim |
| Company Dossier | A company name or ticker | What it really sells into AI, exposure purity, financial reality checks, risks |
| Watchlist | Any scan or dossier | Observable indicators that would strengthen or weaken the thesis |
How scoring works
Each candidate bottleneck is scored 0-5 on: concentration, expansion lead time, substitutability, criticality, pricing evidence, demand visibility. Missing evidence is scored Unknown, never rounded up.
Evidence tags on every fact
Verified · Reported · Claimed · Background · Inference · Unknown
If live data is not available in a session, the agent says so and gives you a verification checklist instead of pretending to have checked.
Try these prompts
Find the bottlenecks behind AI data center power demand.Audit this claim: "Company Z controls 95% of component Q and is sold out through 2030."Dossier on <company>: how pure is its exposure to AI cooling?Turn the open items into a watchlist.
Why it is different
- Checks the thesis, not just the ticker
- Transparent rubric: you can see why a layer scored what it did
- Separates a real bottleneck from a good outcome for any single company
- Assesses claims, not people
- Works in English, Arabic and French
Important limits
- Research and education only. Not investment, legal or tax advice. No buy/sell/hold calls, no price targets, no position sizing, no predictions.
- Outputs are AI-generated and can contain errors. Verify against primary sources before acting.
- Data freshness depends on the runtime. Without live sources, facts are labeled as background or claimed.
- Not affiliated with any company, analyst or brand mentioned in outputs.
Privacy
Your messages are processed by the language model provider configured by the publisher, through Capafy's relay. Do not paste passwords, account numbers or personal identifiers.
