2026-W10: March 2–8, 2026

Weekly AI Intelligence Digest

Week of March 2–8, 2026 | Your Conversation Map for the Week Ahead

DRAFT — NOT YET REVIEWED: This digest was generated from daily briefings that have not been annotated by the reviewer. It should not be distributed to ELT until human review is complete.

_Retroactive backfill generated 2026-05-28 to close /intelligence/digests gap. Written in original-cadence voice (Sunday 2026-03-08) for format integrity._

The Week in One Breath

The SI channel race W09 flagged became operational. Cognizant deployed Devin at 350K-employee scale (3/5); Anthropic confirmed Infosys (3/2); Microsoft embedded Claude into M365 via Copilot Cowork at a new $99/user E7 tier (3/7) — strongest external validation of our multi-vendor principle to date, and a structural convergence creating a fresh Glean overlap. Cursor Automations, Codex Windows, and Google Antigravity arrived as production agentic entrants outside our radar. Governance evidence crossed the buying threshold in 48 hours: 97%/38% gap survey, MIT monitoring study, CNBC silent-failure warning, Unleash $35M raise. The Enterprise AI Governance Offering can no longer credibly stay "Proposed."

Conversations to Have This Week

1. Cognizant Just Made Our Devin Pilots a Critical Path, Not a Differentiator

What happened: Cognizant partnered with Cognition on 3/5 to deploy Devin across its 350K-employee base with bundled governance. Stacks on Anthropic-Infosys (3/2) and Amdocs's "Agentic Services" launch — three SI-level offerings in one week using the stack we are piloting. Cursor Automations, Codex Windows, and Antigravity added three production entrants outside our radar.

Why it matters to us: W09 called this the operational urgency. This week converted the warning into a deadline. Our three Devin pilots are the only remaining proof points that could distinguish us from Cognizant's scale or Infosys's rollout. The pivot is no longer "we have the partnership" — it's "we have the radar methodology, governance rigor, and pilot outcomes."

The question to ask: Are we willing to publish a Wave 1 methodology — Windsurf services enablement plus Devin pilot #1 — before all three pilots complete, or do we accept that the shipping window may close first?

Our current stance: The pursuit lists three Devin pilots as critical path with no published start dates. Cognizant, Infosys, and Amdocs have moved past Endava in scale and named identity since the pursuit was last updated.

2. Microsoft Just Embedded Claude in M365 — Our Best Multi-Vendor Argument, and Our Hardest Glean Conversation

What happened: Microsoft launched Copilot Cowork on 3/7, embedding Claude into M365 with background task execution and a new $99/user E7 tier. Three downstream questions opened unresolved: does E7-bundled Claude change the ROI on a standalone Claude pilot? Does Cowork's M365-wide knowledge capability compete with Glean? Does WWT have an advisory pricing model for E7 vs. standalone Claude vs. Azure AI?

Why it matters to us: When the largest OpenAI investor bundles Claude into its productivity suite, single-vendor consolidation arguments lose most of their force. But the launch creates a partnership-overlap problem on Glean and a licensing-advisory revenue opportunity we are not built for.

The question to ask: Is WWT advising clients on Microsoft's multi-model platform, or being displaced by it — and which Glean engagements should we re-scope before clients raise the E7 comparison?

Our current stance: The multi-model position treats principled diversification as our differentiation. The glean file does not yet anticipate Cowork as a competitive overlap. No internal pricing artifact exists for E7.

3. The Governance Offering Has Buyers — We Just Haven't Made It Active

What happened: Four independent governance signals in 48 hours: a 500-tech-leader survey found 97% concerned about unethical AI use but only 38% with formal oversight (3/4); an MIT-led study found most law-firm agent systems lack monitoring, transparency, and stop controls (3/4); CNBC experts called for kill switches against "silent failures at scale" (3/4); Unleash raised $35M for feature flags and kill switches on AI-generated code (3/6). Stacked on Colorado 6/30/2026 from W09.

Why it matters to us: The pursuit has been "Proposed" while evidence has crystallized for three weeks. The 97%/38% figure is the sharpest market-sizing data we've had. Goldman Sachs (3/4) found AI productivity is real but concentrated in software at ~30% — failures are most acute exactly where deployment is accelerating. Unleash is the commercial mirror of what we would offer. We do not need more evidence.

The question to ask: What does the first paid Governance Offering engagement look like — and can we name a target client and an owner by end of week?

Our current stance: The pursuit is "Proposed." The ai-governance-and-risk position does not yet incorporate "silent failure," kill-switch patterns, or the 97%/38% gap. Colorado deadline ~16 weeks out.

Where We're Well-Positioned

• Multi-model principle has its strongest external endorsement. Microsoft embedding Claude in Cowork is the proof point clients arguing single-vendor consolidation now have to refute. See knowledge/positions/multi-model-multi-vendor.md.
• Anthropic commercial trajectory de-risks the partnership. Claude Code at $2.5B ARR with users doubling since January weakens the Pentagon downside in knowledge/decision-log/2026-02-27-anthropic-pentagon-vendor-posture.md.
• Radar evaluation rigor remains IP no SI has matched. Cognizant, Infosys, Amdocs have named offerings but no comparable framework. See knowledge/positions/ai-assisted-development-tooling.md.

Where We're Exposed

• No published agentic delivery methodology vs. three named SI competitors: Cognizant, Infosys, Amdocs — High (upgraded from W09).
• Radar gap on OpenAI/Google agentic platforms: GPT-5.4, Codex Windows, Antigravity in production without evaluation — High.
• Copilot Cowork / Glean overlap unresolved: Three [REVIEW NEEDED] flags from 3/7 open — Medium.
• Governance Offering still "Proposed" with quantified demand: No named owner — High.
• NIST AI Agent Standards owner unassigned, five briefings: Shaping window closing — Medium.

Real-World Connections

Partnership & Pursuit Spotlight

Partnerships Affected

Pursuits Affected

Decisions Needed This Week

• Methodology Wave 1 publication. "Wait for all three pilots" is untenable post-Cognizant. Name owner, set scope, pick date by Friday. See knowledge/pursuits/agentic-coding-delivery-methodology.md.
• Governance Offering — Proposed to Active. Four demand signals plus Colorado 6/30 deadline. Assign owner; identify reference client. See knowledge/pursuits/enterprise-ai-governance-offering.md.
• Copilot Cowork advisory readiness. Working session on E7 pricing, Glean positioning, standalone Claude pilot ROI. Do not advise on E7 until resolved.
• Glean partnership-tier engagement. Glean will restructure SI channel in 60–90 days. Re-engage now.
• NIST AI Agent Standards owner. Five briefings without assignment. Escalate to leadership.
• OpenAI + Google radar sprint. GPT-5.4, Codex Windows, Antigravity in production without evaluation.

On the Radar

• March/April — NIST AI Agent Standards comment deadline. Shapes the baseline our governance offering will reference.
• June 30, 2026 — Colorado high-risk AI impact-assessment requirement. First hard deadline creating a concrete client entry point.
• Microsoft 30% CSP Copilot promotion — time-limited. Activate for existing pilot clients before expiry.
• Anthropic enterprise expansion. Plugins, agents, Claude Code Security + $2.5B ARR signal a land-and-expand that could displace traditional enterprise software we integrate — long-horizon watch.
• Cursor capability vs. governance. $2B ARR + Automations create tension against Watch downgrade. Consider a comparison brief without changing rating.

Synthesized from ~70 source items across 6 daily briefings (March 2, 4–8, 2026 — no briefing March 3). 24 items flagged high-relevance. Reviewer annotations not present — see DRAFT banner.