AI Intelligence Brief · Weekly Edition
Your Workers Use AI Daily — Your Training Budget Just Shrank
AI adoption is outrunning the organizations deploying it: workers use tools without training, agents run on bad data, and governance varies by state line.
Past editionRead the current oneAll editions
This week's ranked trends
What moved, and why it matters
1
Rank
Your Workers Use AI More, but You Train Them Less
Inflection↑ up 7 to #1High impactT1 confirmed↑ Accelerating
PillarPeople2ndLeadershipLeverTalent
Builds onPwC 2025 Global Workforce Hopes and Fears Survey(Nov 2025)
What's happeningPwC's 2026 Global Workforce survey of nearly 50,000 workers found 64% used AI at work this past year, up 10 points. Daily generative-AI use jumped from 14% to 22%. Yet access to learning resources fell from 59% to 51%. This follows PwC's 2025 survey flagging similar gaps.
Why it mattersYour team is self-teaching AI on the job without guardrails. Every unguided workaround that becomes habit is a future rework cost you will pay to undo at scale.
One questionCan you name the specific AI skills your top 50 people learned this quarter and who paid for them?
What to doMandate that every business unit with more than 50 people submit a quarterly AI-skills plan tied to actual job tasks, not generic courses. Have HR deliver a unit-by-unit count of AI training hours per employee against current tool usage rates within 30 days.
3 signals underneath · momentum
3 / 10
Freshness
+28
Acceleration
4 sources
Corroboration
Medium
Novelty
Week 10
Tracking
Oct 2
KPMG: 24% of Firms Now Redefine Promotion Criteria Around AI Literacyoperating_model_shift
KPMG's Q3 2026 survey found that 24% of organizations are actively redefining promotion criteria to prioritize AI literacy and effective agent delegation, up from negligible levels in 2025. Additionally, 44% report significant workforce adoption of AI agents, up from 10% in 2025.
ProcessesT1Med
Sep 29
PwC: 64% of Workers Now Use AI at Work, but Access to Upskilling Has Droppedskills_program
PwC's 2026 Global Workforce Hopes and Fears Survey (nearly 50,000 workers) finds 64% used AI at work in the past year, up 10 points YoY, with daily GenAI use jumping from 14% to 22%. However, access to learning/development resources fell from 59% to 51%, creating a growing skills divide between front-runners and the broader workforce.
PeopleT1High
Sep 29
McKinsey: 11 Million US Workers May Need to Switch Occupations by 2035workforce_restructuring
McKinsey Global Institute's 'Workforce in Motion' report estimates approximately 11 million US workers may need to transition to different occupations as AI reshapes the economy, yet only one in seven currently has a clear pathway. Demand for 'AI fluency' has grown elevenfold, now valued over pure programming skills.
PeopleT1High
2
Rank
The Average Enterprise Now Runs 13 AI Agents in Production
Inflection↑ up 2 to #2High impactT1 confirmed↑ Accelerating
PillarPlatforms2ndProcessesLeverCompetitive exposure
What's happeningKPMG's Q3 2026 AI Pulse Survey reports that 25% of large enterprises now run multi-agent AI systems in production. That is up from 6% a year ago. The average enterprise operates 13 agents (AI programs that coordinate tasks autonomously), up from 5 in 2025. This follows KPMG's Q2 finding that multi-agent use had doubled to 18%.
Why it mattersCompetitors embedding agents into core operations are compressing cycle times you still staff with people. Every quarter you delay widens a gap that gets harder to close with hiring alone.
One questionHow many AI agents run inside your operations today, and who approved each one's authority to act?
What to doName one executive accountable for an enterprise-wide agent registry that logs every agent, its scope, and its data access. Have engineering and IT inventory every active AI agent, its owner, and what it can do. Deliver the written list in two weeks.
5 signals underneath · momentum
3 / 10
Freshness
+16
Acceleration
5 sources
Corroboration
Medium
Novelty
Week 11
Tracking
Oct 2
KPMG: Multi-Agent Deployments Quadruple to 25% of Large Enterprisesagent_deployment
KPMG's Q3 2026 AI Pulse Survey reports that 25% of large enterprises now run multi-agent AI systems in production, up from 6% a year earlier. 62% of firms with $1B+ revenue have deployed or are actively developing AI agents, and the average enterprise runs 13 agents in production, up from 5 in 2025.
PlatformsT1High
Oct 1
IBM Launches Self-Hosted Option for 'IBM Bob' Agentic Dev Platformagent_platform_release
IBM introduced a self-hosted deployment option for 'IBM Bob,' its agentic software development platform, enabling highly regulated enterprises to run AI coding agents within air-gapped or sovereign-cloud environments. This addresses a key deployment barrier for financial services, government, and defense customers.
SourceIBM
PlatformsT2Med
Sep 30
Agentic AI Reshapes ERP: Oracle, IFS, and Infor Embed Multi-Agent Orchestrationworkflow_redesign
Industry analysis published September 30 details how major ERP platforms — including Oracle Fusion Cloud (600+ embedded GenAI agents), IFS Cloud, and Infor CloudSuite — are deeply embedding multi-agent orchestration into core offerings. Agents now autonomously monitor supply chains, reroute shipments, and adjust production schedules with minimal human oversight.
SourceAIMultiple
ProcessesT2Med
Sep 28
Instinct Raises $1B at $10B Valuation With Only 14 Employeesfunding_round
AI agent startup Instinct closed a $1B Series C at a $10B valuation, quadrupling its valuation in a single month. The company has just 14 employees. The round underscores the massive market premium investors place on autonomous agentic execution platforms over traditional software.
T1High
Sep 28
Netskope Ships MCP Security Broker to Govern Machine-to-Machine Agent Trafficagent_platform_release
Netskope released its Skylight Agentic Broker, which decodes Model Context Protocol (MCP) traffic in real time to identify active AI agents, tool requests, and session responses. Traditional human-centric security controls fail when agents communicate machine-to-machine; this tool aims to prevent autonomous data exfiltration.
SourceNetskope
PlatformsT2Med
3
Rank
New Chips Built Just for AI Agents Are Already Shipping
Inflection↑ up 3 to #3High impactT1 confirmed↑ Accelerating
PillarLeadership2ndPlatformsLeverCost
Builds onNVIDIA Launches Vera CPU (Announcement / GTC)(Mar 2026)
What's happeningCoreWeave began deploying NVIDIA's Vera CPU. It is a processor built for agentic AI workloads — the computing layer that lets AI agents run, test code, and learn in parallel. A single rack supports over 11,000 concurrent agent environments. NVIDIA first announced Vera at GTC in March.
Why it mattersAgent-specific hardware is dropping the cost floor for running autonomous AI. Any multi-year compute contract signed before these chips reach market pricing locks you into rates that will look expensive within 18 months.
One questionDoes your largest cloud or compute contract include a price-reduction clause tied to hardware improvements?
What to doCap every new compute commitment at 12 months unless the vendor includes a ratchet clause that passes through hardware cost declines. Have procurement pull all AI infrastructure contracts above $250K and flag any with fixed pricing beyond Q3 2027.
2 signals underneath · momentum
2 / 10
Freshness
+27
Acceleration
3 sources
Corroboration
Medium
Novelty
Week 11
Tracking
Sep 30
CoreWeave Deploys NVIDIA Vera — the First CPU Designed for AI Agentscapability_milestone
CoreWeave announced the deployment of the NVIDIA Vera CPU, marketed as the first CPU built specifically for agentic AI workloads. While GPUs handle model training and inference, agent loops require massive parallel CPU environments for sandboxing, code execution, and reinforcement learning. Rack-scale Vera supports 11,000+ concurrent environments.
SourceCoreWeave
PlatformsT1High
Sep 29
Bain Forecasts $6 Trillion AI Market by 2031, Warns Infrastructure Built Ahead of Demandcapex_investment
Bain & Company's 7th Global Technology Report projects the global AI market could reach $6T in annual revenue by 2031, with $4.2T needing to come from novel revenue streams (autonomous systems, physical AI, drug discovery) beyond current consumer/enterprise applications. The report notes infrastructure is being built well ahead of the demand curve.
SourceBain & Company
LeadershipT1High
4
Rank
Bad Data Is the Top Reason AI Agents Fail in Production
Inflection↓ down 1 to #4High impactT1 confirmed↑ Accelerating
PillarData2ndProcessesLeverRisk & liability
What's happeningKPMG's Q3 2026 survey found that 53% of enterprises cite data readiness — quality, access, and governance — as the number-one barrier to deploying AI agents. Agents that act on their own magnify the damage of dirty or incomplete data. This follows KPMG's Q2 report identifying the shift from scaling AI to making it operationally sustainable.
Why it mattersEvery agent you deploy on unreliable data makes faster, more confident wrong decisions. The cost isn't a failed pilot — it's an autonomous system acting on bad information at production speed.
One questionWhich three data sets power your highest-stakes AI tools, and when were they last audited for accuracy?
What to doDecide that no AI agent moves to production until the data it depends on has a named owner and a documented quality score. Have data engineering deliver a written quality assessment of every data source feeding a live AI agent within 30 days.
1 signal underneath · momentum
6 / 10
Freshness
+14
Acceleration
3 sources
Corroboration
Medium
Novelty
Week 11
Tracking
Oct 2
53% of Enterprises Cite Data Readiness as Top Barrier to Agent Deploymentdata_readiness_evidence
Despite surging agent adoption, KPMG's Q3 2026 survey identifies data readiness — encompassing data quality, accessibility, and governance — as the number-one barrier to AI agent deployment, cited by more than half of respondents. Agentic systems that act autonomously magnify the consequences of poor data quality.
DataT1High
5
Rank
Most Firms Now Report Measurable AI Returns for the First Time
Established↓ down 3 to #5High impactT1 confirmed→ Steady
PillarLeadershipLeverRevenue
What's happeningKPMG's Q3 2026 survey finds 58% of enterprise leaders now see measurable business value from AI. They cite productivity gains (56%), better decisions (49%), and direct financial outcomes (37%). Two-thirds say they would maintain AI spending through a recession. This follows KPMG's Q1 2026 finding that nearly two-thirds saw meaningful value.
Why it mattersYour board will benchmark your AI returns against the 58% of peers already reporting value. If you cannot show a number, you no longer look cautious — you look behind.
One questionWhat is the dollar figure your AI spending returned last quarter, and who presented it to the board?
What to doSet a standing board-report line item that shows AI spend against measurable output every quarter, starting this one. Have finance and department heads co-produce a one-page AI ROI summary with actual productivity and revenue metrics for the next board packet.
2 signals underneath · momentum
6 / 10
Freshness
+1
Acceleration
5 sources
Corroboration
Low
Novelty
Week 11
Tracking
Oct 2
58% of Enterprise Leaders Report Measurable AI ROI; 67% Call Budgets Recession-Proofvalue_evidence
KPMG's Q3 2026 survey finds 58% of global enterprise leaders now see measurable business value from AI, distributed across productivity (56%), decision-making (49%), and financial outcomes (37%). Separately, 67% say they would maintain AI spending even through a recession, positioning AI as a core strategic line item.
LeadershipT1High
Oct 2
US Chamber of Commerce: 66% of Small Businesses Now Use AI; 47% Have Added Jobsproductivity_result
The US Chamber of Commerce's Empowering Small Business report (October 2, 2026) finds 66% of US small businesses use AI, with top use cases in marketing (68%), customer comms (52%), and admin (47%). Notably, 47% of SMBs using AI have added headcount, versus only 6% reporting cuts — countering immediate displacement fears.
PeopleT1Med
6
Rank
Google and a Startup Bid $12.5M for One Company's Old Emails
EstablishedHigh impactT2 supported• Emerging
PillarData2ndLeadershipLeverRisk & liability
What's happeningGoogle bid $10 million at a bankruptcy auction for Spirit Airlines' internal data. The trove included 100 million emails, 500 million Teams messages, and 30 million lines of code. AI startup Micro1 then outbid Google at $12.5 million, per the Washington Post and Time.
Why it mattersYour internal emails, documents, and chat logs now carry a market price you have never calculated. That value may already be walking out through a vendor agreement you signed.
One questionDo any of your current AI vendor contracts grant the vendor rights to retain, train on, or resell your company's data?
What to doForbid any vendor contract that allows training on or resale of your proprietary data without explicit, revocable consent. Have legal review every active AI vendor agreement and flag any clause granting data-use rights beyond the contracted service. Report back in two weeks.
1 signal underneath · momentum
4 / 10
Freshness
n/a
Acceleration
4 sources
Corroboration
High
Novelty
Week 1
Tracking
Oct 1
Google and Micro1 Bid Up to $12.5M for Spirit Airlines' Bankruptcy Datadata_licensing_deal
Google won a bankruptcy auction for Spirit Airlines' internal data (100M emails, 500M Teams messages, 30M lines of code) at $10M, then was outbid by AI startup Micro1 at $12.5M. The bidding war signals that corporate digital exhaust is now a high-value commodity as public training data nears exhaustion.
DataT2High
7
Rank
Robotics and Physical AI Startups Are Raising Billion-Dollar Rounds
EstablishedMedium impactT1 confirmed• Emerging
PillarProcesses2ndPlatformsLeverCompetitive exposure
Builds onSkild AI Valued at >$14B(Jan 2026)NEURA Robotics $1.4B Round (Europe)(Dec 2025)Physical Intelligence ~$11B Valuation(Nov 2025)
What's happeningFieldAI reportedly sought $700 million at a $10 billion valuation to scale mapless autonomous navigation, according to SiliconAngle. SiMa.ai closed $150 million at a $1.45 billion valuation for physical AI on edge devices — industrial robots, drones, and vehicles. This follows Skild AI reaching a $14 billion-plus valuation earlier this year.
Why it mattersCapital is flooding into AI that moves in the real world — warehouses, fleets, factories. If your operations depend on physical labor or logistics, funded competitors will automate those same tasks within two to three years.
One questionWhich of your most labor-intensive physical operations could a well-funded competitor automate first?
What to doChoose one high-labor operation — warehouse, fleet, or field service — and allocate budget to scope what physical-AI vendors can deliver today. Have operations identify the three highest-cost manual workflows and have engineering assess commercially available robotics or autonomous solutions for each.
2 signals underneath · momentum
4 / 10
Freshness
n/a
Acceleration
3 sources
Corroboration
High
Novelty
Week 1
Tracking
Oct 3
FieldAI Seeks $700M at $10B Valuation for Mapless Autonomous Navigationfunding_round
FieldAI reportedly sought $700M in funding at a $10B valuation to scale its mapless autonomous navigation models for physical AI applications. The round reflects surging investor conviction in the transition from digital AI agents to physical-world autonomous systems.
SourceSiliconAngle
T3Med
Sep 28
SiMa.ai Closes $150M Series C at $1.45B for Edge Physical AIfunding_round
SiMa.ai secured $150M in Series C financing at a $1.45B valuation to scale physical AI applications on edge devices, targeting industrial robotics, drones, and automotive systems.
SourceSiMa.ai
T1Med
8
Rank
Three Governments Now Want Three Different AI Rule Books
Inflection↓ down 1 to #8High impactT2 supported↑ Accelerating
PillarLeadershipLeverRisk & liability
What's happeningChina published its AI Safety Governance Framework 3.0, mandating training-data rules, algorithm filing, and censorship. The US federal government has signaled deregulation while over 250 state-level AI laws create a domestic patchwork. The EU faces internal pressure to soften AI Act enforcement. Each regime is pulling in a different direction, as reported by Freshfields and CSIS.
Why it mattersIf you operate across borders, there is no single compliance playbook. Each market now carries its own regulatory cost, and a product compliant in one jurisdiction may violate rules in another.
One questionCan you list which AI regulations apply to your operations in your top five markets by revenue?
What to doMandate a jurisdiction-by-jurisdiction AI compliance map covering every market where you sell or operate AI-enabled products. Have legal and compliance deliver a written matrix of applicable AI regulations per market, with gaps flagged, within 45 days.
1 signal underneath · momentum
1 / 10
Freshness
+13
Acceleration
4 sources
Corroboration
Medium
Novelty
Week 4
Tracking
Sep 29
China Releases AI Safety Governance Framework 3.0 as US-EU Governance Models Diverge Furtherstandards_framework
China published its AI Safety Governance Framework 3.0, enforcing requirements on training data, algorithm filing, and censorship while promoting its Global AI Governance Initiative. Meanwhile the US federal government signals deregulation and renamed AI to 'Super Intelligence' by executive order, while the EU faces internal pressure to soften AI Act enforcement. Over 250 US state-level AI laws now create a complex patchwork.
LeadershipT2High
9
Rank
Connecticut Now Requires You to Disclose AI in Hiring Decisions
Established↓ down 4 to #9Medium impactT1 confirmed→ Steady
PillarProcesses2ndLeadershipLeverRisk & liability
Builds onConnecticut SB 5 Signed Into Law(May 2026)
What's happeningOn October 1, 2026, Connecticut's AI employment law took effect. Companies must now disclose AI use in hiring and employment decisions in plain language. The law also specifies that using automated tools does not shield employers from discrimination claims, as reported by CT Mirror. This follows Connecticut's signing of SB 5 in May.
Why it mattersEvery AI tool touching a hiring, promotion, or termination decision now creates a disclosure obligation in Connecticut — and other states are drafting similar rules. Non-compliance exposes you to discrimination suits the technology was supposed to prevent.
One questionWhich of your HR and recruiting tools use AI, and does your hiring disclosure language already meet Connecticut's standard?
What to doMandate that every AI tool used in employment decisions across all states carry a plain-language disclosure reviewed by legal. Have HR and legal audit every recruiting, screening, and performance tool for AI components and produce updated disclosure language for each within 30 days.
2 signals underneath · momentum
2 / 10
Freshness
+2
Acceleration
3 sources
Corroboration
Low
Novelty
Week 11
Tracking
Oct 1
Connecticut AI Employment Law Takes Effect, Mandating Disclosure and Preserving Bias Claimsregulation_policy
On October 1, 2026, provisions of Connecticut's comprehensive AI law took effect requiring companies to disclose AI use in employment decisions in plain language and specifying that automated decision-making does not shield employers from discrimination claims. Full implementation frameworks are due by October 2027.
SourceCT Mirror
LeadershipT1Med
Sep 28
US 'AI Kill Switch Act' Debate Highlights Tension Between Security and Innovation Governanceregulation_policy
Lawfare legal analysis published September 28 examined the proposed 'AI Kill Switch Act,' which would empower the Department of Homeland Security to slow or halt frontier AI models. Legal scholars warn against 'cybersecuritization' of AI governance — treating complex societal issues as narrow security threats — arguing it reduces democratic oversight.
SourceLawfare Media
LeadershipT2Med
Signal ticker
All 19 findings this week
Oct 3Market movesFieldAI Seeks $700M at $10B Valuation for Mapless Autonomous Navigation funding_roundsource ↗T3
Oct 2PlatformsKPMG: Multi-Agent Deployments Quadruple to 25% of Large Enterprises agent_deploymentsource ↗T1
Oct 2Leadership58% of Enterprise Leaders Report Measurable AI ROI; 67% Call Budgets Recession-Proof value_evidencesource ↗T1
Oct 2Data53% of Enterprises Cite Data Readiness as Top Barrier to Agent Deployment data_readiness_evidencesource ↗T1
Oct 2PeopleUS Chamber of Commerce: 66% of Small Businesses Now Use AI; 47% Have Added Jobs productivity_resultsource ↗T1
Oct 2ProcessesKPMG: 24% of Firms Now Redefine Promotion Criteria Around AI Literacy operating_model_shiftsource ↗T1
Oct 1DataGoogle and Micro1 Bid Up to $12.5M for Spirit Airlines' Bankruptcy Data data_licensing_dealsource ↗T2
Oct 1LeadershipConnecticut AI Employment Law Takes Effect, Mandating Disclosure and Preserving Bias Claims regulation_policysource ↗T1
Oct 1PlatformsIBM Launches Self-Hosted Option for 'IBM Bob' Agentic Dev Platform agent_platform_releasesource ↗T2
Sep 30PlatformsCoreWeave Deploys NVIDIA Vera — the First CPU Designed for AI Agents capability_milestonesource ↗T1
Sep 30ProcessesAgentic AI Reshapes ERP: Oracle, IFS, and Infor Embed Multi-Agent Orchestration workflow_redesignsource ↗T2
Sep 29LeadershipBain Forecasts $6 Trillion AI Market by 2031, Warns Infrastructure Built Ahead of Demand capex_investmentsource ↗T1
Sep 29PeoplePwC: 64% of Workers Now Use AI at Work, but Access to Upskilling Has Dropped skills_programsource ↗T1
Sep 29PeopleMcKinsey: 11 Million US Workers May Need to Switch Occupations by 2035 workforce_restructuringsource ↗T1
Sep 29LeadershipChina Releases AI Safety Governance Framework 3.0 as US-EU Governance Models Diverge Further standards_frameworksource ↗T2
Sep 28Market movesInstinct Raises $1B at $10B Valuation With Only 14 Employees funding_roundsource ↗T1
Sep 28PlatformsNetskope Ships MCP Security Broker to Govern Machine-to-Machine Agent Traffic agent_platform_releasesource ↗T2
Sep 28Market movesSiMa.ai Closes $150M Series C at $1.45B for Edge Physical AI funding_roundsource ↗T1
Sep 28LeadershipUS 'AI Kill Switch Act' Debate Highlights Tension Between Security and Innovation Governance regulation_policysource ↗T2
Go deeper
Size up your market
See where AI is actually moving demand, pricing, and competition in your specific market.
Run the AI Market Snapshot →Turn signal into a plan
Build a prioritized, evidence-backed AI roadmap from this week's shifts.
Start an AI Blueprint →How to read this brief
Maturity
EmergingGenuinely new — no earlier precedent found
BuildingGaining ground; we're tracking it develop
EstablishedLatest beat of an ongoing shift — see "Builds on"
InflectionEstablished and accelerating hard right now
Evidence tier
T1Confirmed — primary source
T2Supported — credible secondary corroboration
T3Directional — early / soft signal
Impact & tags
HighMaterially changes a leader's calculus
MedWorth planning around
pricing_changeNamed type from a fixed catalog — how we organize signals
Pillar & lever
PlatformsPrimary pillar — the one dimension that trend's question tests
LeadershipSecondary — also touched, but not what we ask you about
CostThe single business lever the trend moves for you
Methodology
Signals are timely, source-supported findings from the last 7 days. Each is graded by evidence tier (T1 Confirmed / T2 Supported / T3 Directional) and impact, links to its source(s), names the actual company/regulator/vendor involved, and carries a named signal type from a fixed catalog — the type is a soft tag left blank when a finding is genuinely ambiguous. Individual signals are optionally tagged to one of five maturity pillars (People, Processes, Platforms, Data, Leadership) for coverage, and signals that don't fit stay untagged. On a trend the pillar does more work: each card names one primary pillar — the single dimension its "One question" puts to you — plus at most one secondary for context, and one of five business levers (revenue, cost, competitive exposure, risk and liability, talent) that the trend actually moves. Both are labels on the analysis, not the ordering: the brief is still ranked by importance and never grouped by pillar. Across an edition we check that all five pillars come up as a primary at least once and that no single one dominates; when a pillar never does, that is reported as a gap in our source coverage rather than papered over by re-tagging a card. Trends cluster related signals and are ordered by an internal score (intrinsic strength × a momentum multiplier weighted so fresh movement outranks a steady megatrend, with high impact able to override raw momentum) — that score is used only to sort; you see rank + maturity + impact, never a number. Because we only cluster the current window's freshly-published signals, a trend appears only when it has new evidence. To keep "new to us" from reading as "new to the world," a separate web-grounded precedent lookback searches for real, dated earlier instances of the same pattern; when it finds them the trend is labelled Established (or Inflection when accelerating) and shows a dated "Builds on" lineage, rather than Emerging. Precedents are context only — never counted as in-window evidence and never invented. Impact labels are guardrail-gated: a trend built only on Tier-3 signals cannot be rated High, and a demo is never treated as GA. On a baseline edition, no week-over-week movement is claimed. The model writes the prose; it never invents a number, a tier, a source, a type, or a precedent.