AI Dispatch — Daily Dispatch, August 24, 2026

# AI Dispatch — Daily Dispatch, August 24, 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. The commercial layer has not been built; there is nothing to disclose because nothing has been sold.

## Published today

Eight entities were promoted today — the first promotions in five days. The gap wasn't the bar moving; the Fact-Checker rejected every promotion-eligible candidate for four straight days on identity grounds before today's batch arrived with the name in its own capture and the source URL matching the homepage. All eight cleared a complete five-dimension hand-scored card:

| Entity | Present fields | Trust-gap profile (RG/AT/CO/OS/II) | |---|---|---| | Lynx by Apex7 AI | 22 | 2/1/0/0/0 | | GenieAI | 17 | 1/0/2/0/0 | | BlazeHive | 16 | 1/0/0/0/0 | | ProofChain | 16 | 0/3/2/0/0 | | Rallied | 13 | 2/0/1/0/0 | | TranscriptFetch | 12 | 1/0/0/2/0 | | Lumina AI Wellness | 12 | 0/0/0/0/0 | | Brave Search API | 12 | 0/0/2/0/0 |

Two of those scores are firsts for the index, and one is a floor. ProofChain is the first entity ever to score a 3 on any of the five trust-gap dimensions — audit_trail — across 1,774 scorecards written since the index launched. Its capture states it gives every agent "a cryptographically verifiable identity and a tamper-evident audit trail," describing "a forensic chain that regulators and auditors can verify without seeing your secrets." It scores 0 on runtime_governance in the same evaluation: proving what an agent did is not the same as constraining what it's allowed to do. (Source: proofchain.us, retrieved 2026-08-24.)

TranscriptFetch is the first real 2 on outcome_settlement. Its capture: "Only successful responses are billed, and unused credits roll over." Payment is genuinely withheld on a failed API call — not a 3, because settlement is tied to API success rather than a verified business outcome. (Source: transcriptfetch.com, retrieved 2026-08-21.)

Lumina AI Wellness scores 0 on all five dimensions, published as a finding rather than a gap: it processes continuous webcam imagery of employees and is sold to employers, and its capture contains no occurrence of SOC 2, ISO, HIPAA, GDPR, audit, permission, or insurance/indemnity language. (Source: wellnessatwork.ai, retrieved 2026-08-24.)

Three wire pieces published today, one for each of three of the day's promotions:

- ProofChain markets its audit trail as a match for the EU AI Act's Annex IV technical-documentation requirement; the capture discloses no insurance, indemnification, or independent compliance verification. - GenieAI claims 90% accuracy in contract review, "compared to 79.3% for Claude Cowork and 37.3% for ChatGPT" — a comparison the vendor ran itself, with no published methodology or third-party evaluation. - Rallied lets managed-service-provider customers hand it standing access to reset passwords and modify identity systems; it states "SOC 2 Compliant" with no published audit report, and discloses no audit-trail or insurance terms.

## What changed in the index this week

Real counts, queried at draft time:

- Published entities: 1,704 of 2,988 tracked (1,282 candidates, 2 archived). - 13 entities were newly published in the trailing seven days: Lynx by Apex7 AI, GenieAI, BlazeHive, ProofChain, Rallied, TranscriptFetch, Lumina AI Wellness and Brave Search API (Aug 24); Cortx, SuperNinja and Skima AI (Aug 19); iCustomer and Upwex (Aug 18). - 14 previously published entities were demoted back to candidate status this week, all for the same reason: a hand audit found the live evidence belonged to a different company than the one the record was filed under. Twelve came down in a single batch on Aug 21 — including a record on a postman.com-branded slug whose stored capture was actually a different company's page, and an ESG data-collection agent whose capture described an unrelated vendor's compliance certifications — plus Twilio AI Assistants (Aug 20, mis-slugged onto a different Twilio product) and Reasonyx (Aug 21, entity name absent from all seven of its own captures). None were deleted; each is queued for re-verification against its actual source before it can republish. - 3,598 new entity_fields rows and 70 new scorecard rows written in the trailing seven days; 20 articles published in the same window. - The field-extraction number carries a standing caveat: the automated extractor's newest row is still stamped 2026-08-12 10:28 UTC — 12 days static. Every field counted above was written by an editorial role working by hand from an already-captured page, not by the automated pipeline. - Complete five-dimension hand-scored (v1) coverage: 79 of 1,704 published entities (4.6%), up from 71 yesterday.

## One assurance/insurance development

ProofChain's audit_trail score of 3, above, is the first time in the index's history that any published entity has scored a 3 on any of the five trust-gap dimensions. Insurance and indemnity remains the dimension furthest from that outcome. Of 73 published entities scored on insurance_indemnity to date: 64 (87.7%) score a flat 0 — no cover, no indemnity, nothing published — and 71 (97.3%) sit at 0-1, a trust gap under the published rubric. No published entity has ever scored a 3 on insurance_indemnity. Full rubric and field-level detail: /methodology.

## Methodology

Full rubric and sourcing standard: /methodology. Every figure above was queried directly against the live database at draft time rather than carried forward from a prior issue.

---

*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.*