AI Atlas

Daily updates on real-worldAI deployments worldwide.

← All posts
7 Sep 2026 – 13 Sep 2026

AI Atlas Weekly Report — 2026 Week 37

WeeklyWarningSoftware
Share
13
New Use Cases
30
New Organizations
5
Countries/Regions
3
Industries

Use Case Highlights

90% task success rate · month-long workflows compressed to days · 50–60 specialized model calls per request — Salesforce's Informatica unit replaced single-agent designs with the CLAIRE multi-agent system inside Intelligent Data Management Cloud, coordinated by an orchestration agent with validation checkpoints and deterministic tool routing.

SalesforceUnited StatesSoftware

US DoD Program of Record (March 2026) · primary AI platform for targeting/intelligence/battlefield decision support — Palantir's Maven AI is the military's official AI stack.

Palantir TechnologiesUnited StatesSoftware

Claude as default model for ServiceNow Build Agent — ServiceNow deepened its Anthropic partnership to power enterprise app development, enabling developers to build agentic workflows that reason, act, and execute autonomously.

ServiceNowUnited StatesSoftware

Swiss Armed Forces technology partner integration · 2025 ARC-AGI-2 leader (>27%) — Giotto.ai's portable reasoning system to be integrated into RUAG's LLARA AI platform following April 2026 cooperation agreement.

Giotto.aiSwitzerlandSoftware

150+ U.S. health systems · 47% fewer documentation errors · 22% more relevant findings captured — Microsoft's DAX Copilot provides ambient clinical documentation across the Veterans Health Administration and other large systems.

MicrosoftUnited StatesSoftware

Trends

Multi-agent orchestration is the new default for enterprise data workflows

CLAIRE (Salesforce), Maven AI (Palantir), Claude-Powered ServiceNow Build Agent, and the agentic architecture in 85% of this week's 13 published use cases — control-plane-plus-specialized-agents pattern replacing single-agent designs.

Software industry dominates — 85% concentration (11 of 13 published UCs)

Insurance (UNIQA SEE) and Consumer Finance (Klarna) are the only other industries with new published UCs this week; Software industry cross-ref pool reached 113 published UCs.

Under the Hood

Pipeline and operations behind the headline numbers—search tuning, data checks, internal notes, next steps, and linked case IDs.

Search Strategy

Query Performance

QueryHitNotes
Claude Code agent security review enterprise 2026HighFigma dual-model Claude Code + Codex case hit at confidence 0.9
multi-agent enterprise workflow orchestrationHighCLAIRE pick; 4 Salesforce candidates at confidence 0.9 all hit this query
agentic security investigation Director-Experts-CriticMediumSlack multi-agent security investigation case; lower volume but high quality
AI claims automation insurance AzureMediumUNIQA SEE — single Austria hit; Insurance industry sparse

New Queries Added

multi-agent enterprise data workflow orchestration 2026AI security investigation Director-Experts-Critic architectureClaude Code + Codex dual-model security review

Data Quality

0

Zero coordinate records

n/a (no occurrences)

0

Missing company_id

n/a (no occurrences)

0

Short content (<500 chars)

n/a (no occurrences)

Observations

  • Daily-ai-push cron was offline the entire reporting window (2026-09-07 → 2026-09-13). Pipeline metrics last entry is 2026-08-26, an 18-day gap. The 13 published use cases and 30 new companies in this report came via the Claude Code admin path instead.

  • Weekly pick is Informatica CLAIRE — multi-agent enterprise data workflow orchestration with 90% task success rate and 50–60 specialized model calls per request. Selected over Nokia Cursor, Atlassian Rovo harness, Figma dual-model security reviewer via cross-ref score (19 vs 14/11/10).

  • Software industry dominates: 11 of 13 published use cases (85%) are Software — same-industry pool at 113 published UCs. Insurance (UNIQA SEE) and Consumer Finance (Klarna) are the only other industries with new published UCs this week.

  • 4 multi-deployment pattern companies this week (Atlassian, Salesforce, Google DeepMind, Bayer each with 4 published UCs in last 90 days) — Salesforce appears both as the pick's company and as a multi-deployer, anchoring both same-company and multi-deployment cross-refs.

  • News pipeline produced only 1 thematically relevant article (Box CEO Aaron Levie's Fable 5.1 enterprise data eval); CLAIRE/Informatica-specific news absent. CLAIRE pick relies entirely on use case + admin knowledge, no corroborating news.

  • useCaseStatusBreakdown 13 published / 6 pending / 48 archived = 67 total — 72% archived rate. Step 3 admin path is producing high quality at low volume, suggesting filter rules may be over-tight or pipeline path differs structurally from cron path.

Next Steps

  • 1highskills/daily-ai-push/SKILL.md

    Cron offline 18+ days (last successful run 2026-08-26); new ingestion path through Claude Code admin replacing daily push

    Investigate cron health and decide whether to (a) restore daily-ai-push cron, (b) formalise Claude Code admin path as new primary ingestion, or (c) run both in parallel with deduplication

  • 2highskills/daily-ai-push/SKILL.md

    Use case status breakdown: 13 published / 6 pending / 48 archived (72% archived rate) — strict curation yielding low volume

    Review strict filter rules; consider relaxing vendor_self_marketing and content_length thresholds if they are over-rejecting real deployments

  • 3mediumskills/weekly-report/SKILL.md

    News coverage thin — only 1 thematically relevant article found for CLAIRE/Informatica pick this week (Box CEO Fable 5.1 eval)

    Add CLAIRE/Informatica and enterprise data agent workflow keywords to the daily-ai-push L2 query pool to widen future news coverage

Related Use Cases