Dr. John Sviokla

Ideas · Framework

The RISE framework: the four stages of AI adoption

By · Developed from my July 2025 Forbes column · Updated

RISE is a four-stage model of how a company builds the capability to use AI: Research & Education, Islands of Innovation, Scaling & Orchestration and Emergent Intelligence. Each stage has its own job, its own funding rule and its own proof of success. You cannot skip a stage, and most companies stall in the gap between the second and the third.

Why does AI adoption need a stage model?

Because AI is a capability, not a purchase. As I argued in Forbes in July 2025, technologies can be bought and implemented, but capabilities must be grown. You can license the same models as your competitors. You cannot license the judgment, workflows and data your organization builds by using them.

That makes RISE an intelligence maturity model, not a technology maturity model. The loop that matters has two sides. People learn to use the systems, and the systems get more specific to your business as people use them. Each turn of the loop raises what the next turn can do. Advantage compounds, which is why firms that wait may never catch up.

What are the four stages of RISE?

StageThe jobHow to fund itProof of successTypical horizon
R · Research & EducationLearn what is possibleLearning budget. No ROI test.Fluency and informed decisions3 to 6 months
I · Islands of InnovationBuild capability in a few functionsPortfolio funding. The portfolio must break even.Proven use cases and champions6 to 12 months
The GapBuild what scale requiresCapability investment, budgeted as suchGovernance, shared platform, trained ambassadors6 to 9 months
S · Scaling & OrchestrationPush proven capability into core processesConventional ROI casesReturn on investment and better operating metrics12 to 24 months
E · Emergent IntelligenceMake AI disappear into the operating modelOperating leverageRevenue per employee, return on assets, growthOngoing

Match the money and the metric to the stage. Demanding hard ROI from a learning budget kills learning. Accepting break-even from a scaled program wastes capital.

R · Research & Education

The job is to learn what AI can do for your business. Leaders get briefed on AI economics, risk and the adoption path so they decide on evidence rather than hype. Every employee gets safe access to approved tools, a usage policy they can follow and hands-on training in their real work.

You are here if AI was handed to IT and treated as a tool rather than a business priority, projects live in sandboxes, and a person checks every output. Treat spend as a learning budget. Measure use and confidence, not ROI. You are ready to move on when use is habitual and natural champions have surfaced.

I · Islands of Innovation

The job is to build capability in a few functions at no net cost. Give champions tools, budget and air cover. Every island needs an owner and an economic case. Review the portfolio each quarter on three measures: belief, milestones and net present value. Kill or scale accordingly, and recycle part of the surplus from the winners into next year’s portfolio.

Pick projects with PRISM, a five-rule screen. Does the project hit existing metrics? Does it teach the staff new skills? Does it tie to the firm’s identity and purpose? Does it fit the business rhythm, with short cycles rather than long ones? Does it keep cost and benefit in the same unit, so power and politics do not stall it?

You are here if a thousand flowers are blooming. Teams launch their own projects, work gets duplicated and standards vary. Horizontal tools such as copilots spread small gains thinly while the high-value use cases get stuck in pilot.

The Gap: from islands to enterprise

The hardest move in RISE is from Islands to Scaling. Most firms do not lack pilots. They lack reusable capability, funding built for the crossing and proof. In Bain’s November 2025 executive survey, 74% of respondents called AI a top-three strategic priority, yet only about 23% said generative AI had delivered more revenue or lower costs. Software development led the move to scale, with 40% of pilots scaling. Customer service followed at 32%.

Firms stall because wins stay local while costs and risks multiply, and because no shared governance, platform or evaluation discipline exists. Crossing takes four things:

S · Scaling & Orchestration

The job is to move AI from the edge of the business to its core. Projects now reinvent processes rather than assist them, and each carries a conventional ROI case with an owner accountable for the result. Strategy is set centrally and executed locally. Governance covers privacy, ethics, explainability and risk, with C-suite sponsorship.

The advantage at this stage is the system around the model, not the model itself. I call it the AI operating system. It has four parts: a model layer that lets you swap providers by workload and cost, a router that sends each request to the cheapest model that clears the quality bar, a harness that plans work, calls tools, checks results and escalates to people, and a control plane that applies rules, testing and audit to every request. A small Center of Excellence keeps it current and turns one team’s win into everyone’s template. Manage cost per successful outcome, not cost per token.

E · Emergent Intelligence

The job is to make AI disappear into the operating model. Agentic systems coordinate work across the enterprise with limited direction, and models improve as they are used. Human and machine judgment combine in everyday decisions. The proof shows up in the operating statistics, not in demos: quality, cycle time, return on assets and, above all, revenue per employee. Few firms outside the largest technology companies operate here today.

What moves an organization from one stage to the next?

People, more than technology. Two engines do most of the work.

Ambassadors. Modeled on the quality movement, ambassadors learn, do and teach. White Belts apply AI to their own work and explain it to peers. Green Belts run multiple projects to completion. Black Belts lead complex programs, teach the teachers and shape policy. Enthusiasts are usually about one employee in ten. Research on the diffusion of innovations suggests that 10% to 20% of a population acting as active champions is enough to tip adoption. Shift the belt mix as you climb: mostly White at R, White and Green at I, Green and Black at S, Black at E.

Commitment, not compliance. Mandates can start adoption. Only commitment sustains it. AI threatens four human drives at once: to earn, to learn, to bond with colleagues and to defend what people have built. In my August 2026 Forbes column I put it this way: “Organizations that satisfy all four get commitment. Organizations that frustrate any one of them get resistance.” Be candid about which roles will change, and give each function a voice in redesigning its work.

What do companies at different stages look like?

JPMorgan Chase built a learning base fast. The bank released its internal LLM Suite to eligible employees in 2024. It went from zero to 200,000 onboarded users within eight months, according to the bank. Broad, governed access is a strong Research & Education and Islands move. The advantage arrives later, when the platform is connected to internal data and workflows.

Klarna shows that efficiency alone is not Emergent Intelligence. In February 2024 Klarna reported that its AI assistant handled 2.3 million conversations in its first month, did the work of about 700 full-time agents and cut resolution time from 11 minutes to under 2. By May 2025 its CEO said cost had carried too much weight in the decision and that customers should always be able to reach a person. The lesson is calibration. Automate the work, keep the customer promise and keep the human path real.

How do you start?

  1. Locate yourself. Name the RISE stage for the enterprise and for each major function. They will differ.
  2. Set the goal. Pick a target stage and date for each function, and adopt the funding rule that goes with it.
  3. Run the plan for your stage, and seed the ambassador program on day one. It powers every later stage.
  4. Use it yourself. Leaders who use AI every week move their organizations faster than leaders who delegate it.

Questions about RISE

What does RISE stand for?
RISE stands for Research & Education, Islands of Innovation, Scaling & Orchestration and Emergent Intelligence. They are the four stages a company passes through as it builds the capability to use AI.
Can a company skip a RISE stage?
No. In my experience no organization has jumped from the first stage to the last. Each stage builds the skills, governance and evidence the next one needs. Companies that skip the Islands stage tend to fail when they try to scale.
Is RISE a technology maturity model?
No. RISE measures intelligence maturity. The question is not which tools you have bought but how well your people and your systems learn together. A license creates access. A learning loop creates advantage.
How long does it take to move through RISE?
Typical horizons are 3 to 6 months for Research & Education, 6 to 12 months for Islands of Innovation, 6 to 9 months to cross the Gap and 12 to 24 months for Scaling & Orchestration. Emergent Intelligence has no end date. Different functions in the same company often sit at different stages.
What should you measure at each stage?
Measure use and confidence at Research & Education. Hold the Islands portfolio to break even. Demand conventional ROI at Scaling & Orchestration. At Emergent Intelligence, watch the operating statistics: revenue per employee, return on assets, cycle time and quality.
Where did RISE come from?
RISE grew out of the four-stage AI maturity model in my July 2025 Forbes column, “You Can Buy AI, But You Can’t Fake Capability.” I refined it with clients at GAI Insights into the stage, funding and measurement rules on this page.

Sources