AI Intelligence Brief · Weekly Edition
GPT-6 Lands, Returns Stay Invisible, and Congress Notices
The gap between AI capability and AI accountability widened again this week — and your exposure grew on both sides.
Past editionRead the current oneAll editions
This week's ranked trends
What moved, and why it matters
1
Rank
GPT-6 Astra Just Made Your AI Contracts Look Expensive
EstablishedHigh impactT1 confirmed• Emerging
PillarPlatforms2ndLeadershipLeverCost
Builds onOpenAI GPT-5.6 Sol frontier model family(Jul 2026)OpenAI GPT-5.5 autonomous computer operator model(Apr 2026)OpenAI GPT-5.4 with autonomous agent prep(Mar 2026)
What's happeningOpenAI released GPT-6 Astra on September 14, 2026. It is not a chatbot but an autonomous computer operator. It writes code, analyzes data, and configures systems on its own. It uses up to 65% fewer output tokens (billable units of AI work) and runs 25 times faster per turn than competitors. Priced at $10 per million input tokens, it undercuts most enterprise AI pricing. This follows OpenAI's GPT-5.6 Sol release in July.
Why it mattersAny AI compute contract signed before this week carries rates that may already be above market. Locked-in pricing that looked reasonable in Q2 now overpays for less capable models.
One questionCan you name every AI vendor contract your company holds, its per-unit price, and when it expires?
What to doCap all new AI compute commitments at 12 months and require automatic price-reset clauses tied to public benchmark pricing. Have procurement pull every active AI contract over $100K and list the per-unit rate, lock-in period, and renegotiation window.
1 signal underneath · momentum
10 / 10
Freshness
n/a
Acceleration
4 sources
Corroboration
High
Novelty
Week 1
Tracking
Sep 14
GPT-6 Astra launches as autonomous computer operator with aggressive pricingmodel_release
OpenAI released GPT-6 Astra on September 14, 2026, a model designed not as a conversational assistant but as an autonomous computer operator capable of long-horizon software engineering, data analysis, and system configuration. It scored 57.9% on Terminal Bench 4.0, 99.9% on ARC-AGI-3, and 100% on ExploitBench. Priced at $10/M input tokens and $50/M output tokens, it uses up to 65% fewer output tokens and achieves 25x lower turn latency than competing models, collapsing the unit economics of enterprise AI deployment.
PlatformsT1High
2
Rank
OpenAI Says Its Own Model Can Find and Exploit Vulnerabilities
Inflection↑ up 6 to #2High impactT1 confirmed↑ Accelerating
PillarLeadership2ndPlatformsLeverRisk & liability
What's happeningOpenAI's system card for GPT-6 Astra, published September 14, 2026, classifies the model at the "critical cybersecurity threshold." It scored 100% on ExploitBench. That means it can autonomously find and exploit zero-day vulnerabilities — flaws no one has patched yet. The same model sold for software engineering can also weaponize security gaps. This follows the NY Department of Financial Services' May 2026 warning on frontier AI cyber risks.
Why it mattersIf you deploy an autonomous agent that can exploit vulnerabilities, you own the liability when it does — whether it targets your systems or someone else's.
One questionHas your security team tested what happens when your own AI agents are pointed at your own unpatched systems?
What to doMandate that no autonomous AI agent goes live without a documented security review signed by your CISO. Have engineering and security red-team your highest-value internal systems using your current AI tools and deliver written findings.
1 signal underneath · momentum
10 / 10
Freshness
+15
Acceleration
3 sources
Corroboration
Medium
Novelty
Week 4
Tracking
Sep 14
GPT-6 Astra reaches 'critical cybersecurity threshold' — can discover and exploit zero-day vulnerabilitiessafety_incident
OpenAI's own system card classifies GPT-6 Astra at the 'critical cybersecurity threshold,' acknowledging the model can autonomously discover and exploit zero-day vulnerabilities (scoring 100% on ExploitBench). This forces enterprise leaders to confront a new category of dual-use risk: the same agent that automates software engineering can also weaponize security flaws.
LeadershipT1High
3
Rank
40% of Large Companies Are Scaling AI Agents — Most Without Guardrails
Established↑ up 1 to #3High impactT1 confirmed→ Steady
PillarProcesses2ndPlatformsLeverCompetitive exposure
Builds onMCP reaches 97M monthly downloads, donated to Linux Foundation AAIF(Dec 2025)
What's happeningMcKinsey's State of AI 2026 report shows 40% of large enterprises now scale agentic AI (AI that acts on its own across systems). That is up from 27% a year ago. Most organizations remain gated by gaps in process documentation and data readiness, not technology.
Why it mattersCompetitors are moving agents from pilot to production. Every quarter you spend debating governance is a quarter they spend learning what works at scale.
One questionWhich of your core business processes has a documented, approved plan for agent-driven automation — not a pilot, but a production rollout?
What to doChoose three revenue-critical workflows and commit to production-scale agent deployment on each within two quarters. Have department heads deliver a one-page readiness assessment for each workflow, covering data access, approval chains, and rollback procedures.
2 signals underneath · momentum
6 / 10
Freshness
+0
Acceleration
3 sources
Corroboration
Low
Novelty
Week 8
Tracking
Sep 13
Model Context Protocol hits 120M cumulative downloads, outpacing Kubernetes adoption trajectoryinteroperability_standard
A September 13, 2026 report noted that the Model Context Protocol (MCP) — a standard for connecting AI models to external systems, now managed by the Agentic AI Foundation under the Linux Foundation — reached 120 million cumulative downloads in just 20 months. This vastly outpaces historical adoption of Kubernetes. Major vendors including OpenAI, Google, and Microsoft have integrated MCP, signaling platform interoperability has become an immediate enterprise priority.
PlatformsT2High
Sep 10
40% of large enterprises now scaling agentic AI, up from 27% a year agoagent_deployment
McKinsey's State of AI 2026 report shows that 40% of large enterprises are now scaling agentic AI deployments, up from 27% the prior year. Meanwhile, 31% of large enterprises are specifically scaling autonomous coding agents. The shift from pilot to production-scale agentic systems is accelerating across industries, though most organizations remain gated by process documentation and data readiness gaps.
ProcessesT1High
4
Rank
Your Best People Are Hiding the Time AI Saves Them
Established↓ down 3 to #4High impactT1 confirmed→ Steady
PillarPeopleLeverTalent
What's happeningGartner reported on September 10, 2026 that "layoff remorse" is now measurable. Companies that cut headcount mainly because of AI are losing institutional knowledge and talent pipelines. Gartner predicts at least 30% of those roles will need to be restored by 2029 — at higher cost. McKinsey data backs this up: 32% of firms planned AI-driven cuts, but only 14% followed through. This follows Forrester's July finding that 55% of employers regret AI-driven layoffs.
Why it mattersCutting roles to fund AI spending creates a hidden liability. Rehiring those same roles later costs more than keeping them and retraining now.
One questionWhich roles have you eliminated in the past year specifically because of AI, and what institutional knowledge walked out with them?
What to doStop approving any headcount reduction justified solely by AI capability until HR delivers a skills-gap analysis for each affected role. Have HR list every AI-justified role cut in the last 12 months, the knowledge lost, and the estimated cost to rehire.
1 signal underneath · momentum
4 / 10
Freshness
+3
Acceleration
3 sources
Corroboration
Low
Novelty
Week 2
Tracking
Sep 10
Gartner predicts 30% of AI-eliminated roles will need to be restored at premium cost by 2029workforce_restructuring
A Gartner report published September 10, 2026 documents 'layoff remorse' as an emerging phenomenon: organizations that used AI primarily for headcount reduction are suffering depleted talent pipelines and eroded institutional knowledge. Gartner predicts at least 30% of prematurely eliminated positions will need to be restored by 2029 at significantly higher cost. McKinsey data corroborates the gap: while 32% of firms expected AI-driven job cuts last year, only 14% actually executed them.
PeopleT1High
5
Rank
93% of Audit Teams Use AI — 54% Have Never Measured the Return
Established↓ down 3 to #5High impactT1 confirmed→ Steady
PillarLeadership2ndDataLeverCost
What's happeningA Gartner survey published September 10, 2026 found 93% of audit functions use AI. Yet 60% have no formal AI strategy and 54% have never measured ROI. Only 30% use AI for core audit testing; the rest draft reports. McKinsey's broader data is consistent: only 37% of leaders can link AI usage to EBIT impact. That number is flat year-over-year.
Why it mattersActivity without measurement is cost without accountability. Every dollar spent on AI tools no one tracks against earnings is indistinguishable from waste.
One questionFor each AI tool your company pays for, can you name the P&L line it is supposed to move and by how much?
What to doRefuse to renew any AI tool subscription that lacks a written, measurable business case tied to a specific financial metric. Have finance build a single ledger of every AI expenditure, the business case behind it, and whether anyone has measured the result.
2 signals underneath · momentum
4 / 10
Freshness
-1
Acceleration
3 sources
Corroboration
Low
Novelty
Week 8
Tracking
Sep 10
Gartner: 93% of audit functions use AI, but 60% lack any formal AI strategyvalue_evidence
A Gartner survey published September 10, 2026 found that 93% of audit leaders are using AI — yet 60% have no formal AI strategy, 54% have not begun measuring ROI, and only 30% use AI for core audit testing (versus 60% for drafting reports). The finding exemplifies the systemic gap between activity metrics and strategic alignment across enterprise functions.
LeadershipT1High
Sep 10
Only 37% of leaders can link AI usage to EBIT impact despite 80% reporting personal productivity gainsvalue_evidence
McKinsey and Deloitte data referenced in September 2026 reporting shows 80% of workers report AI has improved their individual productivity, yet only ~37% of leaders can map those gains to tangible EBIT impact — a figure that has remained flat year-over-year. The persistent disconnect between individual adoption and measurable financial outcomes continues to define the enterprise AI maturity gap.
LeadershipT1High
6
Rank
A New Bill Would Force Companies to Log Every AI Agent Action
Established– holds #6High impactT1 confirmed→ Steady
PillarProcesses2ndLeadershipLeverRevenue
What's happeningOn September 9, 2026, Representatives Gottheimer and Lawler introduced the bipartisan "Stop Rogue AI Act." It requires tamper-evident logging (unalterable records) of all material AI agent actions. It also mandates continuous inventories of active agents and cryptographic identity checks. Compliance is tied to federal procurement. This follows Senator Warner's AI AGENT Act in July.
Why it mattersIf this bill passes, any company without a real-time inventory of its AI agents loses access to federal contracts. That is revenue at risk, not a compliance checkbox.
One questionCould you produce a complete, accurate list of every AI agent running in your organization by the end of this week?
What to doMandate a company-wide AI agent registry — every autonomous tool, who owns it, what it accesses — updated continuously, not quarterly. Have engineering and IT deliver a first draft of that registry within 30 days, covering every department, with legal reviewing logging requirements.
2 signals underneath · momentum
3 / 10
Freshness
-3
Acceleration
4 sources
Corroboration
Low
Novelty
Week 8
Tracking
Sep 9
Stop Rogue AI Act introduced in Congress, mandates tamper-proof agent logging and inventoriesregulation_policy
On September 9, 2026, U.S. Representatives Gottheimer and Lawler introduced the bipartisan 'Stop Rogue AI Act,' requiring NIST to develop standards for agentic AI deployment. The bill mandates tamper-evident logging of all material agent actions, continuous machine-readable inventories of active AI agents, and cryptographic identity verification. Compliance is tied to federal procurement, effectively forcing private-sector adoption.
LeadershipT1High
Sep 9
OpenAI advocates for mandatory capability-based AI safety regulation, backs four California billsregulation_policy
On September 9, 2026, OpenAI publicly endorsed mandatory, capability-based national AI safety regulations in the U.S. and explicitly backed four California state bills (SB 813, AB 1405, SB 1119, AB 1864) in what it calls a 'reverse federalism' strategy. The move signals that Tier 1 platform vendors now view democratically accountable standards as essential to maintaining public trust and enterprise adoption.
LeadershipT2Med
7
Rank
AI Governance Goes Global: Institutions Step In
EstablishedMedium impactT1 confirmed• Emerging
PillarLeadershipLeverRisk & liability
What's happeningOn September 11, 2026, the UN Secretary-General proposed a $3 billion Global Fund for AI Capacity-Building. It targets gaps in compute and data infrastructure across developing nations. On September 8, the OECD updated its Responsible AI Policy overview. It reinforced international consensus around risk-proportionate governance aligned with ISO/IEC 42001 and the NIST AI Risk Management Framework. This follows the UN Global Dialogue on AI Governance in Geneva in July.
Why it mattersInternational AI standards are converging. Companies operating across borders face a narrowing window to align practices before compliance splits into costly, country-by-country patchwork.
One questionIn how many countries does your company deploy AI, and which local AI regulations apply in each?
What to doName one person accountable for tracking international AI regulation across every market you operate in. Have legal produce a country-by-country matrix of current and pending AI rules affecting your products or operations, updated monthly.
2 signals underneath · momentum
4 / 10
Freshness
n/a
Acceleration
3 sources
Corroboration
High
Novelty
Week 1
Tracking
Sep 11
UN Secretary-General proposes $3B Global Fund for AI Capacity-Buildingregulation_policy
On September 11, 2026, the UN Secretary-General recommended establishing a Global Fund for AI Capacity-Building with an initial $3 billion to address geographic gaps in compute, high-quality data, and digital public infrastructure for developing nations. The proposal signals that AI readiness is now treated as a matter of geopolitical stability and global economic equity, not just corporate competitiveness.
LeadershipT1Med
Sep 8
OECD updates Responsible AI Policy overview, emphasizing balanced governancestandards_framework
On September 8, 2026, the OECD published an updated Responsible AI Policy overview emphasizing the need for governance that mitigates harms like bias and privacy violations while supporting innovation. The update reinforces an emerging international framework consensus around risk-proportionate AI regulation aligned with standards like ISO/IEC 42001 and the NIST AI RMF.
LeadershipT1Med
8
Rank
Employees Spend 20 Days a Year Fixing AI Mistakes
Established↓ down 5 to #8High impactT2 supported↓ Decelerating
PillarData2ndProcessesLeverRisk & liability
What's happeningA Syskit report from September 10, 2026 found that 76% of US and UK organizations have deployed AI tools like Microsoft 365 Copilot. But only 43% reviewed data permissions first. KPMG and Alation analyses from September confirm 63% of organizations lack confidence in their data management for AI. When AI crawls ungoverned file shares, it surfaces sensitive documents to people who should never see them.
Why it mattersDeploying AI on top of messy permissions does not just slow results. It exposes confidential data to anyone with a prompt. That is a breach waiting to happen.
One questionWhen was the last time someone audited who can access what in the file stores your AI tools read from?
What to doForbid any new AI tool deployment on internal data until a permissions audit of the target data store is complete and signed off. Have IT and security audit permissions on every data source connected to an active AI tool and deliver a risk-ranked report within 45 days.
3 signals underneath · momentum
4 / 10
Freshness
-16
Acceleration
5 sources
Corroboration
Low
Novelty
Week 8
Tracking
Sep 10
76% of organizations deployed AI on Microsoft 365 data, but only 43% reviewed permissions firstdata_governance_move
A Syskit report published September 10, 2026 found that while 76% of UK and US organizations have piloted or deployed enterprise AI tools like Microsoft 365 Copilot, only 43% conducted a thorough review of data permissions and oversharing risks beforehand. Deploying AI against ungoverned data stores effectively weaponizes internal knowledge bases against an organization's own security perimeter.
DataT2High
Sep 10
63% of organizations lack confidence in data management for AI; analysts predict majority of unsupported AI projects will be abandoneddata_readiness_evidence
September 2026 analyses from KPMG and Alation emphasize that 63% of organizations lack confidence in their data management practices for AI. Legacy data warehouses designed for human-read dashboards are fundamentally inadequate for autonomous agent consumption. The market is pivoting toward continuous, machine-readable metadata monitoring and AI-ready data pipelines as the critical enabler for scaling agentic workflows.
DataT2High
Sep 10
Caterpillar hiring 'Governance, Data Quality & AI Readiness Lead' — signaling new enterprise data rolesai_role_emergence
A job posting published in the September 7–14 window shows Caterpillar recruiting for a 'Governance, Data Quality & AI Readiness Lead,' requiring cross-functional expertise in privacy, compliance, and ML observability. The role represents the emergence of dedicated positions bridging legacy IT and modern AI requirements — a structural indicator that enterprises are formalizing data readiness as an organizational function.
PeopleT3Low
9
Rank
AI Skills Expire Faster Than Your Hiring Can Keep Up
Established↓ down 2 to #9Medium impactT2 supported↓ Decelerating
PillarPeople2ndLeadershipLeverCompetitive exposure
What's happeningA Datacom survey from September 10, 2026 reports that 42% of organizations have appointed a Chief AI Officer (CAIO) or equivalent. Another 21% plan to within a year. Wharton and MIT CSAIL now market 9-to-12-month CAIO certification programs. The role bridges the CTO, Chief Data Officer, and business units. It owns risk tolerances, investment governance, and AI compliance. This follows Forrester's late-2025 finding that only 16% of workers scored high on AI readiness.
Why it mattersAI decisions are scattered across your CTO, CDO, and line-of-business heads. Without one person accountable, investment overlaps and risk gaps multiply.
One questionIf your board asked today who owns AI investment decisions, risk limits, and compliance across the company, could you give one name?
What to doName a single executive accountable for all AI investment, risk, and compliance decisions — whether a new CAIO hire or a current leader with expanded authority. Have HR draft a role charter covering budget authority, risk thresholds, and reporting lines, with input from legal and finance, within 30 days.
1 signal underneath · momentum
4 / 10
Freshness
-16
Acceleration
3 sources
Corroboration
Low
Novelty
Week 7
Tracking
Sep 10
42% of organizations have appointed a Chief AI Officer; CAIO certification programs proliferateai_role_emergence
A Datacom survey published September 10, 2026 reports 42% of organizations have appointed a CAIO or equivalent, with 21% more planning to within 12 months. Executive education providers including Wharton and MIT CSAIL are now marketing 9-to-12-month CAIO certification programs. The CAIO is emerging as the bridge between CTO, CDO, and business units — responsible for risk tolerances, investment governance, and AI compliance.
PeopleT2Med
10
Rank
The AI Market Is Consolidating Around a Handful of Giants
Established↓ down 5 to #10Medium impactT2 supported↓ Decelerating
PillarPlatformsLeverCompetitive exposure
Builds onHarvey $200M round at $11B valuation (GIC & Sequoia)(Mar 2026)Harvey $160M round at $8B valuation (Andreessen Horowitz)(Dec 2025)
What's happeningOn September 9, 2026, Harvey — an AI legal automation company — closed a $550 million Series F at a $15.5 billion valuation. This follows Harvey's $200M round at $11B in March. Reports from September 14 indicate Anthropic expects a second straight quarter of positive adjusted operating income on an annualized run-rate near $65 billion. Capital is concentrating in a few platform vendors.
Why it mattersWhen your AI vendors are backed by billions and approaching profitability, switching costs rise fast. The longer you wait to diversify, the harder it gets.
One questionWhat percentage of your AI spending goes to a single vendor, and what is your fallback if that vendor changes pricing or terms?
What to doSet a ceiling: no single AI vendor may account for more than 50% of your total AI spend without board-level approval. Have procurement map every AI vendor relationship by annual spend and contract terms, and flag any vendor above the threshold.
3 signals underneath · momentum
5 / 10
Freshness
-29
Acceleration
4 sources
Corroboration
Low
Novelty
Week 8
Tracking
Sep 14
Anthropic approaches $65B annualized revenue run-rate with second consecutive quarter of positive adjusted operating incomecorporate_strategy
Reports circulating September 14, 2026 indicate Anthropic expects a second consecutive quarter of positive adjusted operating income, backed by an annualized revenue run-rate approaching $65 billion. The figure underscores that the foundational AI platform market is consolidating around a few heavily capitalized entities capable of sustaining gigawatt-scale infrastructure investment.
LeadershipT2Med
Sep 9
Harvey raises $550M Series F at $15.5B valuation for AI-powered legal workflowsfunding_round
On September 9, 2026, Harvey — an AI company specializing in legal process automation — closed a $550 million Series F round at a $15.5 billion valuation. The round signals strong investor confidence in vertical-specific, context-aware agentic AI applied to highly regulated domains, moving well beyond generic chatbot capabilities.
T2Med
Sep 9
TAR raises $120M Series A for off-grid AI power infrastructurefunding_round
TAR raised a $120 million Series A on September 9, 2026, focused on off-grid power solutions for AI data centers. The investment reflects the growing recognition that physical energy infrastructure is becoming a binding constraint on AI scaling, and that purpose-built power generation is now a distinct, investable category.
T3Med
Signal ticker
All 18 findings this week
Sep 14PlatformsGPT-6 Astra launches as autonomous computer operator with aggressive pricing model_releasesource ↗T1
Sep 14LeadershipGPT-6 Astra reaches 'critical cybersecurity threshold' — can discover and exploit zero-day vulnerabilities safety_incidentsource ↗T1
Sep 14LeadershipAnthropic approaches $65B annualized revenue run-rate with second consecutive quarter of positive adjusted operating income corporate_strategysource ↗T2
Sep 13PlatformsModel Context Protocol hits 120M cumulative downloads, outpacing Kubernetes adoption trajectory interoperability_standardsource ↗T2
Sep 11LeadershipUN Secretary-General proposes $3B Global Fund for AI Capacity-Building regulation_policysource ↗T1
Sep 10LeadershipGartner: 93% of audit functions use AI, but 60% lack any formal AI strategy value_evidencesource ↗T1
Sep 10LeadershipOnly 37% of leaders can link AI usage to EBIT impact despite 80% reporting personal productivity gains value_evidencesource ↗T1
Sep 10PeopleGartner predicts 30% of AI-eliminated roles will need to be restored at premium cost by 2029 workforce_restructuringsource ↗T1
Sep 10Data76% of organizations deployed AI on Microsoft 365 data, but only 43% reviewed permissions first data_governance_movesource ↗T2
Sep 10Data63% of organizations lack confidence in data management for AI; analysts predict majority of unsupported AI projects will be abandoned data_readiness_evidencesource ↗T2
Sep 10Processes40% of large enterprises now scaling agentic AI, up from 27% a year ago agent_deploymentsource ↗T1
Sep 10People42% of organizations have appointed a Chief AI Officer; CAIO certification programs proliferate ai_role_emergencesource ↗T2
Sep 10PeopleCaterpillar hiring 'Governance, Data Quality & AI Readiness Lead' — signaling new enterprise data roles ai_role_emergencesource ↗T3
Sep 9LeadershipStop Rogue AI Act introduced in Congress, mandates tamper-proof agent logging and inventories regulation_policysource ↗T1
Sep 9LeadershipOpenAI advocates for mandatory capability-based AI safety regulation, backs four California bills regulation_policysource ↗T2
Sep 9Market movesHarvey raises $550M Series F at $15.5B valuation for AI-powered legal workflows funding_roundsource ↗T2
Sep 9Market movesTAR raises $120M Series A for off-grid AI power infrastructure funding_roundsource ↗T3
Sep 8LeadershipOECD updates Responsible AI Policy overview, emphasizing balanced governance standards_frameworksource ↗T1
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.