Product Intelligence
Product Intelligence
When Your Source Is Your Customer's Press Release

Okta dropped three announcements in June claiming ownership of AI agent security. The independent security community describes identical risks using completely different vocabulary. That gap is a live stress test for any Anchor deployment where the customer is also the dominant source in the research pipeline.
This piece traces the problem to a product-level finding: Anchor's research configuration has no way to tag whether a source is vendor-produced or independently calibrated, and no mechanism to flag when the same concept—here, MCP authorization—surfaces across the Okta and TinyFish deployments under different names.
When Your Source Is Your Customer's Press Release
Okta dropped three announcements in June claiming ownership of AI agent security. The independent security community describes identical risks using completely different vocabulary. That gap is a live stress test for any Anchor deployment where the customer is also the dominant source in the research pipeline.
This piece traces the problem to a product-level finding: Anchor's research configuration has no way to tag whether a source is vendor-produced or independently calibrated, and no mechanism to flag when the same concept—here, MCP authorization—surfaces across the Okta and TinyFish deployments under different names.
Clinical Evidence Gap

Clear Path currently uses Elixir's clinical page as its evidence layer. It should be using it as the claim layer, because that's what it is. Elixir's BIOADAPTOR RCT page reports favorable TLF (target lesion failure: the composite of cardiac death, heart attack in the treated vessel, and repeat intervention at the stent site), CE Mark status (European device clearance), and a U.S. investigational disclaimer. The numbers are real. They appear in peer-reviewed journals and conference coverage. But Elixir selects which endpoints to emphasize. When TCTMD covered the same four-year data in May 2026, outside cardiologists noted that the components driving the TLF gap were only numerically lower, not statistically significant. Elixir's page omits this. Three sources close the gap: ClinicalTrials.gov for trial state, the FDA PMA database (the FDA's most stringent pathway for high-risk devices) for U.S. authorization, and TCTMD for clinical interpretation. Without them, Clear Path is Elixir's narrative with better prose, and a customer can get that from any competent medical writer.
The Public-Signal Ceiling

Shade Tree Fund is the deployment where Anchor's public-monitoring model reaches a limit that better research configuration cannot fix. The variables that determine whether Docket creates dependency — whether a donor update changed a giving decision, whether an expert recommendation shifted a grant, whether a grantee hit a milestone — are private by design and by regulation. No amount of source expansion closes this gap.
It would be tempting to treat Shade Tree like Elixir: find the right public registry, reconfigure research priorities, sharpen the publication. But Elixir's domain has a public primary record (ClinicalTrials.gov) that Anchor wasn't using well. Shade Tree's domain has no equivalent. ProPublica shows zero filings for Shade Tree's EIN, and even when filings appear, IRS rules exempt donor identities from disclosure. A small climate philanthropy with a March 2025 ruling date has a quiet public footprint because that is the normal operating posture of the category.
The product decision is binary. Either Anchor builds a private-input mechanism for customers whose value chain is invisible from public signals, or it draws an explicit boundary around which customer types the current model can serve. Both positions are defensible as long as they're stated.
If Shade Tree begins publishing named grantee outcomes, advisory rosters, or donor impact reports on its public site, the category boundary softens. Until then, Docket's irreplaceability cannot be assessed at even moderate confidence through public monitoring alone.
