AI Dispatch — Daily Dispatch, September 26, 2026

# AI Dispatch — Daily Dispatch, September 26, 2026

**Dek:** Two wire items ran today, both bearing on the same question from opposite sides: enterprise platform Writer discloses SOC 2 and three ISO certifications for its agents but no insurance product, while Munich Re's aiSure — one of eight insurance and liability carriers this index tracks — added a named live deployment to its case-study material without publishing the coverage limit or payout trigger behind it. One entity, SellerClaw, crossed the publication floor in the last 24 hours; the index holds at 804 published. The standing rubric still shows 115 of 116 human-scored published entities (99.1%) at zero or one on `insurance_indemnity`.

---

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.

## What published in the last 24 hours

Two wire items ran:

- **Writer**, an enterprise AI platform, publishes SOC 2 Type II and ISO/IEC 27001, 27701 and 42001 certifications for its agents, alongside a governed execution layer: "Role-based permissions and least-privilege access helps ensure every action is appropriately authorized, while audit logs surface how Writer is used across your organization." No insurance product, indemnification term, or liability cap appears on the same trust page — `insurance_available` is recorded as `no_public_information`. ([full item](https://aidispatch.news/articles/writer-soc2-iso-certifications-no-insurance-2026-09-26)) - **Munich Re's aiSure** added a named live deployment to its case-study material: MKIII, an AI loan-decision model used by credit unions, is "backed by Munich Re's aiSure™ insurance, covering losses from prediction errors, calibration or data drift, unexpected deviations, or shifts in borrower behavior." The claims mechanism is described as "parametric-like" for "fast, objective claims" — but no coverage limit, premium, or payout trigger is published anywhere in the capture. aiSure remains an evidenced, unpublished candidate; more below. ([full item](https://aidispatch.news/articles/munich-re-aisure-mkiii-parametric-claims-2026-09-26))

## What changed in the index this week

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

- **Published entities: 804** (773 agent, 31 vendor). Candidates: 2,299. Archived: 2. - **6 candidate-to-published promotions, 1 published-to-candidate demotion** in the trailing seven days (net +5), per `entity_overwrite_log`. In the last 24 hours: one promotion (SellerClaw, an AI sales agent), zero demotions. - **544 `entity_fields` rows touched** in the trailing seven days — 382 new or corrected `present` values, 162 flipped to `no_public_information`. - **153 `change_queue` items resolved against 2,440 newly detected** in the same seven days. The pending backlog is now **4,149**. - **Zero published entities carry the withdrawn NSFW subtype.** No `job_heartbeats` feed is currently stalled.

## One assurance/insurance development: a named deployment, still no published number

Munich Re's aiSure is the clearest test case this index has for what "AI insurance" means in practice, and today's capture moves it forward on the product side while leaving the pricing side dark. The new material names an actual covered deployment — MKIII, a credit-union loan-decision model — rather than describing coverage in the abstract, and it states a claims mechanism: settlement is "parametric-like," meant to pay on an objective trigger rather than a negotiated loss assessment. That is a real answer on `outcome_settlement`, scored 2 in this index's rubric — held short of 3 because "no trigger metric, threshold or payout basis is published on any page checked, so a reader cannot tell what verified outcome releases payment," per the scorecard rationale on file.

What is still missing is the number a buyer would actually need: no coverage limit, no premium, no stated trigger threshold. `coverage_limits` remains `no_public_information` after this capture, unchanged from the correction this outlet reported in last Friday's issue. aiSure has sat as an evidenced, unpublished candidate since 2026-07-31 and remains short of this index's 12-field promotion bar.

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

- **`insurance_indemnity`**: 116 scored, **115 (99.1%) sit at 0-1** — 112 at a flat 0. The one published entity ever to clear a 1 on this dimension remains Armilla AI, an AI insurance vendor, at 2 — the rubric's top score requires a named carrier, a published limit, and a customer-indemnity term, none of which Armilla's own page states. - **`outcome_settlement`**: 111 scored, 108 (97.3%) at 0-1; 103 at a flat 0. - **`audit_trail`**: 112 scored, 95 (84.8%) at 0-1; 69 at a flat 0. - `compliance` and `runtime_governance` — named for contrast, not part of the standing three — sit at 0-1 for 83.8% of 117 scored entities and 71.4% of 112, respectively.

Full five-dimension human-scored coverage: **111 of 804 published entities (13.8%)**.

The competitor wedge holds, reconfirmed at draft time: zero rows in `competitor_listings` match any of the eight named carriers this index tracks (Armilla, Munich Re aiSure, Lloyd's, Chaucer, Embroker, Mosaic, Relm, Testudo), across both mirrored competitor catalogues. 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.*