AI Dispatch — Daily Dispatch, October 10, 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 listing is sold in this issue. The `listings` table itself holds zero rows at draft time — there is no sponsor inventory yet, let alone an active featured placement. There is nothing to disclose because nothing has been sold.

## What published in the last 24 hours

Three wire items ran:

- **Chat Data**, a chatbot and agent platform the company says is "Trusted by 1000+ businesses worldwide," publishes a customer-facing audit log of agent changes — "See who changed your agent's settings and when, with before/after diffs — and track every webhook delivery with status, latency, and one-click resend," per its changelog — alongside third-party compliance claims including "GDPR EU Data Protection SOC 2 AICPA Compliant" and a described HIPAA program, according to an AI Dispatch review completed Oct. 10. Its pricing page describes a usage-based option: "You can directly meter the agent usage through your Stripe account and charge your customers directly" — a term for how a Chat Data customer bills its own end users, not one under which Chat Data itself is paid on verified performance. A full-text search of its homepage capture found no occurrence of "insurance," "indemnif," "warranty" or "liabilit[y]." AI Dispatch's trust-gap rubric found nothing meeting either the outcome-settlement or insurance/indemnity dimension. ([full item](https://aidispatch.news/articles/chat-data-audit-trail-no-insurance-trust-gap)) - **The Federal Trade Commission** said Oct. 8 it is sending more than $15.8 million to consumers it says were deceived by cash-advance app company Cleo AI over "the amounts and timing of available cash advances." The action lands in a liability category the AI insurance market has begun to underwrite by name: Armilla AI, which describes itself as a Lloyd's Coverholder, advertises coverage "with limits up to $25 million" spanning "AI Model Output Liability" and "AI Model Regulatory Violations." AI Dispatch found no public link between Cleo AI and any liability carrier — the FTC listing names none, and Armilla's site does not name Cleo AI. ([full item](https://aidispatch.news/articles/ftc-cleo-ai-cash-advance-redress)) - **Advibly**, an AI ad-creative generator, and **Kai CSV Profiler**, a CSV-analysis tool, both entered the index this week with support for the Model Context Protocol — and opposite defaults. Advibly's site states "Advibly ships an MCP server, so agents in Claude, ChatGPT, Cursor, Claude Code, VS Code, and other MCP-compatible tools can generate image ads, social posts, UGC videos, talking head videos, carousels, and campaign assets directly from your brand context," while Kai CSV Profiler's states plainly "No account or MCP client is required." Neither discloses governance or insurance terms; both remain candidates pending the evidence review that precedes publication. ([full item](https://aidispatch.news/articles/mcp-advibly-kai-csv-profiler-new-listings))

## What changed in the index this week

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

- **Published entities: 809** (775 agent, 34 vendor). Candidates: 2,409. Archived: 2. - **13 candidate-to-published promotions, 6 published-to-candidate demotions** (net +7) in the trailing seven days, per `entity_overwrite_log`. - **1,721 `entity_fields` rows touched** in the trailing seven days — 725 new or corrected `present` values against 996 flipped to `no_public_information`, skewed toward absence this week as the Fact-Checker worked through a backlog of miscategorized assurance claims. - **2,779 `change_queue` items detected, 1,312 resolved** in the same seven days. The pending backlog stands at **5,140**. - **Zero published entities carry the withdrawn NSFW subtype.**

## One assurance/insurance development: a real liability event, no carrier named

No tracked carrier — Armilla, Munich Re aiSure, Testudo, Relm, Embroker, Mosaic, Lloyd's or Chaucer — produced new public evidence of its own this week. What moved instead is on the liability side: the FTC's $15.8 million redress action against Cleo AI, reported above, is the kind of event this index's rubric is built to track under its outcome-settlement and compliance dimensions — a real financial consequence tied to a gap between an AI-branded product's claims and its actual performance, the exact category of risk Armilla's "AI Model Output Liability" and "AI Model Regulatory Violations" coverage lines are priced to cover. No public source connects the two companies; this index found none and reports that absence rather than implying one.

The standing rubric, human-scored dimensions only, published entities only, queried live at draft time:

- **insurance_indemnity**: 139 scored, **137 (98.6%) sit at 0-1** — 133 at a flat 0. - **outcome_settlement**: 136 scored, 131 (96.3%) at 0-1 — 123 at a flat 0. - **audit_trail**: 147 scored, 129 (87.8%) at 0-1 — 91 at a flat 0. - `compliance` and `runtime_governance` — named for contrast, not part of the standing three — sit at 0-1 for 140 of 161 scored entities (87.0%) and 103 of 138 (74.6%), respectively.

Full five-dimension human-scored coverage: **134 of 809 published entities (16.6%)**.

The competitor wedge holds, reconfirmed at draft time: zero rows in `competitor_listings` (408 theaiagentindex.com, swept today; 3,155 aiagentsdirectory.com, last swept Sept. 29 — now 11 days stale) 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.*