AI Dispatch — Daily Dispatch, September 7, 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 of any tier or status, checked directly at draft time. There is nothing to disclose because nothing has been sold.

## Published today

Six candidates cleared today's field-by-field fact-check audit and were promoted: StressLess AI (16 present fields, 5 scored dimensions), PostPlanify (16, 5), Kommunicate (15, 5), Avina (14, 5), MirrorFly AI Voice Agent (14, 5) and Predis.ai (12, 5). All five score dimensions — including insurance/indemnity — were hand-scored against the published rubric before promotion, not estimated.

Getting there cost 45 rejected fields, all with verbatim quotes that failed on what the quote actually supported rather than on whether it appeared in the source. Nineteen were interface furniture read as fact — nav buttons and footer links recorded as pricing, policy and support-channel facts. Twelve were category errors, the largest being MirrorFly's compliance-certs field, built from "meeting key regulatory standards, including SOC 2, HIPAA, GDPR, and PCI DSS" — a claim about what customers can build on the platform, not a certification MirrorFly itself holds. The rest were duplicated quotes filed under multiple field keys, unshipped features recorded as live, and one summary rewritten because it restated a price-table field that had itself overreached its source.

A live correction. BRYTER Extract — indexed since 2026-08-14 against bryter.com/extract/ — was set back to candidate today. That URL now serves a different product: "BEAMON Extract allows teams to export results for reporting, collaboration, or further automation." Every field attached to the entity, including a SOC 2 Type II certification claim that ran under BRYTER's name for 24 days, in fact describes BEAMON. AI Dispatch has not verified any relationship between the two companies and states none; the entity is held as a candidate pending a re-seed under its correct current identity.

Published entities: 777 → 782 (net +5: six promotions, one correction).

## Wire items today

Three items ran, each on an agent the Researcher surfaced in the last 24 hours with at least one assurance-dimension field backed by an evidenced quote:

- HarnessRouter — an agent-hosting platform that names "Launch Partner: Used by Stanford Medicine." "Every session runs in its own isolated sandbox" is its only disclosed control; it discloses no compliance certification, audit trail or insurance. - YundaDesk — a customer-support agent that discloses two governance controls still rare among vendors AI Dispatch has reviewed: "Every reply writes a trace: which skill, which knowledge, delivered or not," and "Refunds, compensation and order changes always require human approval." It scores 2 of 3 on both audit trail and runtime governance — one of the only agents reviewed to date to score above zero on either. It names no insurance or compliance certification. - Fieldproxy — field-service agents deployed into fire-safety compliance inspections across hundreds of facilities, operating under a runtime rule that "a manager must approve any invoice over $5,000." No compliance certification or insurance is disclosed.

## What changed in the index this week

Queried directly against the live database at draft time:

- Published entities: 782 (753 agent, 29 vendor; 2,298 candidates, 2 archived). - 189 new entity_fields rows in the trailing seven days. - 13 candidate-to-published promotions, 7 published-to-candidate demotions in the trailing seven days (net +6). - 204 change_queue items resolved against 1,500 newly detected in the same seven days — the backlog is still growing faster than it's being worked. - Full five-dimension scorecard coverage, human-scored dimensions only: 82 of 782 published entities (10.5%).

## One assurance/insurance development

Re-queried fresh, restricted to human-scored dimensions only — an automated scorer running alongside the Fact-Checker cannot write a 0 or a 3, so a score blending both would overstate coverage:

- insurance_indemnity: 82 of 82 human-scored published entities (100%) sit at 0 or 1. No published entity has ever scored above 1 on this dimension. - outcome_settlement: 81 of 82 (98.8%) at 0-1, 1 at 2. - compliance: 66 of 82 (80.5%) at 0-1, 16 at 2 or 3. - audit_trail: 70 of 82 (85.4%) at 0-1, 12 at 2 or 3. - runtime_governance: 57 of 82 (69.5%) at 0-1, 25 at 2 or 3.

The competitor wedge also holds: across theaiagentindex.com (370 listed) and aiagentsdirectory.com (3,001 listed) — 3,371 combined — zero name any of the insurance or liability carriers we track (Armilla, Munich Re aiSure, Lloyd's, Chaucer, Embroker, Mosaic, Relm, Vouch, Coalition). Neither 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. Today is Monday; the brief specifies digest format Monday–Thursday and a deep edition Friday.

---

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.