Civico
Research · Knowledge graph

How does the evidence connect?

The short answer

Every pattern, framework, technology, industry and sourced fact in the library, linked. Start with a guided tour below: pick a pattern and the graph walks you from the claim to each source behind it.

548items
1,530connections
15patterns to tour
How to read this
  • Each dot is one item; colour shows its type (key above the graph). Lines show what supports or informs what.
  • A solid line was linked by a researcher; a dashed line was matched by keywords and is shown separately until someone confirms it.
  • Click a dot to see what it is and what it connects to. Drag to move, scroll or pinch to zoom, or search by any word.
Updated 9 Oct 2026Download the data ↓
Guided tours
curated link matched by keywords

What the evidence says, by theme

Each theme is a judgement written by the research run. Facts marked curated were picked by a researcher; facts “matched by keywords” were linked automatically and are shown separately until a person confirms them (513 automatic links in the graph).

Self-reported benefit outruns measured P&L impact · 12 curated · 109 matched · 3 frameworks

Most firms say AI helps; few measure it against a baseline. Headline productivity claims (tripled output, time saved) rarely come with a method, while the studies that do measure find little or no P&L effect yet.

Frameworks: Future-Built and the 10-20-70 rule; AI Maturity Model (five levels) and AI-ready data; Accountable Acceleration (annual enterprise adoption study). See the Frameworks view.

Also matched by keywords (109)
The pilot-to-production funnel narrows hard · 8 curated · 55 matched · 7 frameworks

Evaluation is common; production and scale are rare. Studies from MIT NANDA, BCG, Deloitte, IBM, McKinsey and Gartner put scaled, value-producing deployment between roughly 5% and 25% of firms.

Frameworks: State of AI / AI high-performer practices; Future-Built and the 10-20-70 rule; AI Maturity Model (five levels) and AI-ready data; Enterprise AI Maturity Model; GenAI Divide: State of AI in Business 2025; Root Causes of Failure for AI Projects; AI Governance Alliance playbooks. See the Frameworks view.

Also matched by keywords (55)
Governance and liability catch up with deployment · 7 curated · 48 matched · 5 frameworks

Courts, regulators and insurers are now acting on AI use: fake citations, chatbot liability, insurer exclusions and board oversight lag. Frameworks such as NIST and ISO help size controls, but none is shown to move the P&L.

Frameworks: Cloud Adoption Framework for AI and agents; Cloud Adoption Framework for AI, ML and generative AI; AI Risk Management Framework 1.0; ISO/IEC 42001 AI management system; AI Governance Alliance playbooks. See the Frameworks view.

Also matched by keywords (48)
Southeast Asia: adoption is rising faster than ROI measurement · 6 curated · 43 matched

In Vietnam and ASEAN, adoption and self-reported gains are high, but most adopters are still experimenting and few have a consistent ROI framework.

Also matched by keywords (43)
Agents can do more than firms have deployed · 6 curated · 41 matched · 2 frameworks

Benchmark task success has risen sharply while enterprise deployment of agents stays in single digits by function, and governance and data readiness lag.

Frameworks: State of AI in the Enterprise (2026 edition); AI Maturity Model (five levels) and AI-ready data. See the Frameworks view.

Also matched by keywords (41)
Headcount cuts attributed to AI are being reversed · 7 curated · 37 matched

Klarna, Commonwealth Bank and employer surveys show firms rehiring after AI-attributed cuts, and Gartner expects half of such firms to rehire by 2027.

Also matched by keywords (37)
Small and founder-led firms adopt later and see less · 6 curated · 33 matched · 3 frameworks

Adoption rises with firm size in the US, UK and OECD data; micro firms start high but grow slowly, and most small-firm users report no revenue change.

Frameworks: State of AI / AI high-performer practices; Accountable Acceleration (annual enterprise adoption study); AI adoption by small and medium-sized enterprises. See the Frameworks view.

Also matched by keywords (33)
Skills and training are the binding constraint · 6 curated · 27 matched · 3 frameworks

Skills are the top cited barrier, executives themselves use AI little, and where employers pair tools with training use rises sharply.

Frameworks: State of AI in the Enterprise (2026 edition); AI adoption by small and medium-sized enterprises; ADKAR applied to AI adoption. See the Frameworks view.

Also matched by keywords (27)
Cheaper tokens, bigger bills · 2 curated · 31 matched

Price per unit of capability keeps falling while enterprise AI budgets overrun and software and services pricing is repriced.

Also matched by keywords (31)
Healthcare: modest measured gains, large claims · 6 curated · 24 matched

Measured clinical-documentation savings are small and financial payoff is unproven, while vendor and payer claims are large and regulators and recall data show real risk.

Also matched by keywords (24)
IT services and outsourcing are repricing · 10 curated · 19 matched

IT services, BPO and contact-centre firms report revenue pressure and repricing attributed to AI, while some regional players report growth in AI work.

Also matched by keywords (19)
Manufacturing: measured value lags, early adopters can lose productivity first · 6 curated · 20 matched

Plant-level data show a productivity J-curve for early AI adopters, and most manufacturers report value without scaling it across sites.

Also matched by keywords (20)
Time saved is not money saved · 7 curated · 14 matched

Controlled trials and government pilots show real but modest time savings, mostly self-reported, and an experienced-developer trial found a slowdown. Savings seldom show up as revenue or cost lines.

Also matched by keywords (14)
Winners redesign the work; they do not just add a tool · 6 curated · 10 matched · 4 frameworks

High performers are more likely to have redesigned workflows and jobs; most firms have not. BCG's 10-20-70 rule puts about 70% of AI value in people and process.

Frameworks: State of AI / AI high-performer practices; Future-Built and the 10-20-70 rule; State of AI in the Enterprise (2026 edition); Enterprise AI Maturity Model. See the Frameworks view.

Also matched by keywords (10)
Evidence on training leaders in AI is thin · 4 curated · 2 matched

No AI executive programme has independently measured P&L outcomes. The best causal evidence concerns AI access, not courses, and shows AI amplifies existing managerial judgement.

Also matched by keywords (2)
Keep exploring
Frameworks

Which playbook can we trust?

15 frameworks, evidence checked →Use cases

Who is already doing it?

452 real deployments, studies and filings →Workflow Map

Which work changes first?

Most disrupted: Customer support →
Have a specific challenge?Ask the library in your own words, or talk it through with Civico faculty.
Ask the library →Book a call →

Thank you.

Follow Civico on LinkedIn for the weekly Insight and news from the residentials.