AI Dispatch — Daily Dispatch, October 4, 2026

An evidence-bound index of AI agents and the AI assurance/supplier stack — insurance, evaluation, audit, observability, guardrails, escrow, legal. Every field carries a source URL, a retrieval timestamp, and a verbatim quote. No estimates, no growth-hack framing.

## Sponsor disclosure

No sponsored placement in this issue. `listings` holds zero rows with `tier = 'featured' AND status = 'active'`, checked directly at draft time. There is nothing to disclose because nothing has been sold.

## What published in the last 24 hours

Two wire items ran:

- **Suede AI**, a music- and media-generation agent, publishes three machine-payable endpoints on the x402 protocol priced "$0.15 to $4.99 per call in its live manifest," settled in USDC on Base, according to an AI Dispatch review completed Oct. 4. An agent can call and pay for the endpoints programmatically, but nothing on the site conditions that payment on the delivered work meeting any standard — it is billed per call, not on a verified result. The site's rights-and-provenance material describes what a creation record *could* contain; it states plainly that "no action on this page registers a work, grants a license, or routes royalties." AI Dispatch's trust-gap rubric found no mechanism that releases payment only on a verified outcome. ([full item](https://aidispatch.news/articles/suede-ai-pay-per-call-no-outcome-settlement)) - **OpenAdapt**, an open-source desktop agent with a real healthcare billing-audit deployment, states that it "turns a demonstrated task into an inspectable program" and "stops when verification fails." The only public walkthrough of what that verification produces, however, is a demo the company's own site labels "Interactive illustration with fictional data. No live system is connected." Nothing elsewhere on the site describes what a real, saved audit record from a live deployment contains, how long it is kept, or whether it can be altered after the fact. AI Dispatch found no public evidence of a retained audit trail independent of the fictional demo. ([full item](https://aidispatch.news/articles/openadapt-audit-trail-demo-fictional-data))

## What changed in the index this week

Real counts, queried directly against the live database at draft time:

- **Published entities: 806** (771 agent, 35 vendor). Candidates: 2,412. Archived: 2. - **6 candidate-to-published promotions, 4 published-to-candidate demotions** (net +2) over the trailing seven days, per `entity_overwrite_log`. Four of the promotions landed in the last 24 hours alone: Vouch and Munich Re aiSure (both insurance-subtype vendors), plus Holistic AI (audit-subtype) and Vocera AI (observability-subtype) — the latter two are categories competitors already list at scale, so only the first two extend this index's uncontested wedge. - **81 `entity_fields` rows created** in the trailing seven days — 79 new or corrected `present` values, 2 flipped to `no_public_information`. - **655 `change_queue` items resolved; 0 newly detected** in the same seven days. The pending backlog stands at **3,482**. - **Zero published entities carry the withdrawn NSFW subtype.**

## One assurance/insurance development: two carriers published, no new external claim to report

No tracked carrier — Armilla, Munich Re aiSure, Testudo, Relm, Embroker, Mosaic, Lloyd's or Chaucer — produced new public evidence this week beyond what is already published. Munich Re aiSure and Vouch both had `entity_fields` rows touched in the last 24 hours, but every fact in them restates ground this outlet has already covered in prior wire items. That absence is reported here rather than papered over with a repeat.

What is genuinely new: Vouch and Munich Re aiSure themselves moved from `candidate` to `published` status today, giving the insurance-subtype wedge two scored, evidence-backed vendor entries where a buyer can now read the full field-by-field record. And today's two wire items are the concrete illustration of the gap this index exists to measure — one priced per call with no outcome check, one with an audit-trail demo its own maker labels fictional.

## Standing rubric, human-scored dimensions only, published entities only, queried live at draft time

- **insurance_indemnity**: 133 scored, **131 (98.5%) sit at 0-1** — 127 at a flat 0. - **outcome_settlement**: 129 scored, 124 (96.1%) at 0-1 — 117 at a flat 0. - **audit_trail**: 140 scored, 121 (86.4%) at 0-1 — 86 at a flat 0. - `compliance` and `runtime_governance` — named for contrast, not part of the standing three — sit at 0-1 for 86.5% of 155 scored entities and 72.7% of 132, respectively.

Full five-dimension human-scored coverage: **128 of 806 published entities (15.9%)**. Across all five dimensions combined, published entities carry **606 trust gaps spanning 167 entities**.

The competitor wedge holds, reconfirmed at draft time: zero rows in `competitor_listings` (395 theaiagentindex.com, swept today; 3,155 aiagentsdirectory.com, last swept Sept. 29) match any of the eight named carriers this index tracks. Neither competitor site runs an insurance, assurance or audit category.

## Methodology

Full rubric and sourcing standard: /methodology. Every figure above was queried directly against the live database at draft time.

---

*AI Dispatch is not a growth-hacked directory. Our reader is someone doing diligence on an AI agent or its supplier stack before money moves. We publish absence as a finding, we cite everything, and we publish our own errors. If we ever stop doing that, stop trusting the newsletter.*