How Do You Set Up an Innovation Lab in the Era of AI?

A function, not a place. An innovation lab is the visible part of what Innovation Mode 2.0 calls the innovation function: “an always-on system of processes, services, technology tools, knowledge, and talent, aiming to maximize the ability of the organization to discover and pursue opportunities.” The room and the demo wall are optional. What is not optional: a mandate, a small team that enables others, the services that take an idea from a problem to a validated opportunity, and a digital home where the whole company can reach them. In the era of AI, much of the service layer is software, and the people who remain are the ones who judge.

• What the function provides: capabilities “for capturing and evaluating ideas, sensing the market, rapidly developing prototypes, and running in-market experiments, as well as setting up and facilitating innovation workshops”
• Services, not a building: “In some cases, these services may be offered by specialized teams serving well-defined objectives,” such as market intelligence or large innovation workshops. A lab is such a team, with a name and a door
• The test of a lab: does it make the rest of the company more able to innovate, or does it innovate instead of the company? The first survives; the second is the lab that gets closed
• Four things every lab needs: a mandate, a team, the services and a home. Everything else, including the space, follows from those
• What AI changes: the services that once needed a room full of analysts and facilitators run as agents; see what changes when AI does the work
• Whether you need one at all is a separate question, answered in do you need an innovation lab

The names overlap; the designs do not. Companies use lab, center, hub, studio and team for the same thing and for five different things. What separates them is the job: a team enables, a lab builds and tests, a center houses several services, a hub connects groups that innovate separately, and a studio builds ventures. Innovation Mode 2.0 uses only one of the names, hub, and only for a pattern. It describes the innovation function and a core team whose mission is to enable; a Portal that is the function's digital home; and the Innovation Hub, the network pattern for organizations with several innovation groups. Pick the design by the job, then call it what your company will adopt.

• The team: the Dream Team of the book, a small cross-disciplinary group whose “fundamental mission is to open up innovation to everyone.” It does not build products
• The lab: usually a group that prototypes and runs experiments, often in its own space. Its risk is the one named in the book: “impressive prototypes but minimal business impact”
• The center: a larger unit that offers several innovation services under one roof, often for a region or a division. In the book's terms, an instance of the innovation function
• The hub: not a unit but a pattern: several labs, R&D groups and innovation centers run the same methods and share through one portal, “independently but in synchronization”
• The studio, the incubator and the accelerator: they build or back companies, not prototypes; the venture building guide compares them. R&D looks further out, at technologies rather than opportunities. A center of excellence builds a skill, such as AI, not opportunities
• Not the campus kind: universities, schools and libraries run innovation labs too, and they dominate the search results. This guide is about the corporate innovation lab: a unit inside a company, with a mandate from its leadership and a budget it has to justify
• What to call yours: the word your organization will use without irony. The design decisions in this guide apply whatever the name

They produce prototypes, not business. The labs that fail were set up to innovate in a corner, and the corner is where the work stayed. As I wrote in Innovation Mode 2.0, many organizations have witnessed “the failure of stand-alone innovation labs that produced impressive prototypes but minimal business impact,” which feeds the fear of an “innovation bubble” disconnected from the core business. The lab demonstrates, the business units watch, and nothing crosses the gap. The fix is in the design, not in the people: a mandate to enable, a link to the agenda, a path from a validated opportunity to a product team, and measurement that counts outcomes.

• The bubble: a team seen as “a benefit for a select few who enjoy exploring ideas and using the latest technologies in safe conditions without the pressure of production and operations-related tasks.” Nobody outside it wants its output
• No agenda: the lab chooses its own problems, so its results answer questions the business did not ask; the six deficits behind corporate innovation failure start with leadership for this reason
• No path to market: without a venture-building capability, “great opportunities end up waiting for implementation or are outsourced to vendors or assigned to other teams with limited capacity or skill gaps.” A lab without MVP and venture leads ends at the demo
• The wrong scoreboard: “Many organizations evaluate long-term innovation using short-term KPIs, while others rely on vanity metrics of innovation activity, such as the number of ideas generated, patents filed, and workshops conducted.” A lab measured by demos optimizes for demos
• Silos are made by the program, not the team: “it is the strategy and the overall innovation program that makes or breaks silos.” A lab designed as a hub removes silos; one designed as a destination creates one
• Theater: the lab becomes the company's proof that it innovates, which is the opposite of innovating; see how to spot innovation theater
• How many fail: nobody knows. The figure that circulates, “up to 90%”, is attributed to a 2015 Capgemini report that, as the trade press reported it, measured adoption, not failure: 38% of the world's 200 largest companies were experimenting with such centers. A 2017 Capgemini study found that 87% of respondents had an innovation center while 17% thought they had a company-wide culture of innovation. That gap, not a failure rate, is the problem this guide is about

Enable innovation, do not own it. The one design decision that decides the rest is the mandate. In Innovation Mode 2.0, the core team “is not supposed to produce innovation. Instead, their mission is to enable people, teams, and the entire organization to do so.” A lab with this mandate provides the methods, the services and the tools for everyone; it hosts the workshops, runs the assessments and connects innovators to sponsors. A lab mandated to produce innovation competes with the business units for ideas, people and credit, and it loses that competition the day budgets tighten.

• Four responsibilities: lead the innovation program, provision the capabilities, operate the function, and drive the culture. The book details each in section 3.4.1
• What the mandate excludes: “there are no pure implementation roles such as software product development and related technical profiles.” Building belongs to product teams; the lab makes sure they get validated opportunities
• The definition of done: “The wide adoption of the innovation capabilities is a critical element of the definition of done for the innovation team and a leading indicator of success.” Adoption, not output
• Autonomy with a line to the top: the team “needs increased autonomy to steer the efforts as needed,” informed by metrics, and reports its insights and proposals to the C-suite
• Handle the noise: as I wrote in my post on the innovation team, innovation can disrupt the company before it disrupts any market; the mandate includes protecting the rhythm of the business
• Write it down: the mandate is written into the Innovation Masterplan, where the leadership signs it

Twelve capabilities, start with three. The lab runs the innovation framework: in the book, twelve capabilities that “work together as a single system to streamline the journey of an idea, from draft thoughts to a validated proposal and potentially to commercial paths.” Few labs need all twelve on day one, and the book prescribes no universal order: the Capability Index in how to start decides it for your company. My default is three: the innovation agenda, which sets the problems worth solving; a way to capture and assess ideas against one model; and a simple way to test the best of them with real customers, by hand, before any validation service exists. The framework “is presented independently of specific technologies,” so each can start as a manual service and grow with adoption.

• Setting the focus: the Innovation Agenda (the problems worth solving, from strategy), innovation intelligence (sensing the market) and the customer experience map
• The pipeline: opportunity discovery (ideas captured, framed and assessed), opportunity validation (experiments and prototypes) and opportunity realization (MVPs and ventures); see how ideas are validated
• Steering: the opportunity portfolio, the performance framework and the control panel, where leaders adjust the agenda, the assessment model and the pace
• The connective tissue: knowledge management, recognition and rewards, open innovation, and the Portal as the home of all of it
• Manual first: “some of the above could be organized as “internal services” powered by simple methods and operated manually by small, effective teams without sophisticated technologies.” Sticky notes count
• Who uses them: business leaders set the context, innovators submit and frame, evaluators flag opportunities, market analysts feed insight, product and venture leaders take opportunities to market

Six roles at the core, satellites later. In Innovation Mode 2.0, “Along with the innovation director, the following five roles define an effective core for the innovation team”: innovation architects, innovation orchestrators, innovation advocates and connectors, program managers and workshop facilitators. None of them builds products. The extended team adds the roles that make innovation actionable, such as product architects, validation leads, MVP leads and venture leads, and the whole thing can grow into satellite teams. Start smaller than six: “the team should have an innovation architect and a program manager working closely with the program director to kick off the program.”

• Architects design the function and the program; orchestrators run it and watch its health; advocates and connectors tell the story and bring people together; program managers deliver the capabilities; facilitators run the workshops
• The extended team, just in time: technology advisors, product architects, product managers, opportunity validation leads, MVP leads, venture leads, legal advisors, intrapreneurs, business strategists, commercialization advisors, a content lead and a market intelligence lead; see how the team grows beyond its core
• Six satellite services as it scales: communication and content; market intelligence; opportunity discovery and concept testing; execution and product development; go-to-market and commercialization; venture building, incubation and product-market fit
• Elastic by design: the Guest Innovator program, “a rotation scheme that allows employees from across the organization to join for either a project assignment or on a time-based arrangement,” adds capacity and breaks the isolation
• Permanent over consulting: “the core innovation team benefits significantly from permanent professionals who are deeply integrated within the organization.” Consultants can fill roles early; the rotation covers peaks
• Who leads: the innovation director, whatever the title; see how to hire a chief innovation officer

Whoever owns strategy, and never one business unit alone. In the book, the team's structure is flat and “all members report directly to the innovation executive, who in turn reports to the organization's leadership, either as the chief innovation officer under the CEO or as a lower-ranking executive, such as head or VP of innovation, under another member of the C-suite.” A lab that reports into one business unit inherits that unit's horizon and budget cycle, and the other units stop seeing it as theirs. More important than the box on the chart: a direct line to where the agenda is set, and a steering rhythm where the lab brings its metrics and the C-suite resets the agenda and the pace.

• Three workable models: a chief innovation officer under the CEO; a head of innovation under a C-suite member; or innovation embedded in the CTO or chief strategy officer role. The book uses “Innovation Director” for all three
• Governance, in the book's design: the orchestrators “drive broader ‘innovation steering’ discussions with key stakeholders,” and the leadership uses the control panel to reset priorities: the agenda, the assessment model, the calendar and the rewards. Decision rights sit with the units for their products, with the lab for the method
• What a business-unit line takes away: the company-wide mandate. The lab becomes that unit's prototyping shop, and its budget disappears with the unit's next bad quarter
• Co-funding without capture: units can fund specific validation work or MVPs through the lab's services; the core team and the agenda stay centrally funded
• The leader matters more than the line: “a leader who is educated in innovation methodologies and organizational design” unites the organization rather than centralizing innovation; see who should lead innovation

A digital home first. In Innovation Mode 2.0, “Innovation needs a digital home”: one place where every employee can see the agenda, find the tools, submit an idea, follow the projects and reach the people. The book calls it the Innovation Portal and notes that “in today's era of remote work and distributed teams, it has become an essential capability.” A physical lab serves a narrower purpose well: workshops, prototyping equipment, demonstrations. It serves it badly when it is the only place innovation is allowed to happen, and worst when it is off-site, where the business cannot wander in.

• What the home does: it is “the primary point of reference for innovation” and “the organization's most accurate and complete source of innovation assets”
• What a room does: hosts hackathons, design sprints and prototyping, and gives the community a visible place. Worth having once there is a community to fill it
• When to add it: after adoption. The book's own example: “providing the Makerspace or gateway capability from day one adds no value, as there would not be enough content, a sufficient level of adoption, or readiness to derive value from it”
• Where to put it: inside the flow of the business. An off-site lab is the physical form of the innovation bubble
• What AI did to the cost: “Artificial intelligence, APIs, reusable components, and development frameworks have dramatically reduced the technical barriers to such a platform,” so the digital home no longer needs a large project; start with what delivers value and iterate
• Distributed companies: the home is the only place the whole company shares; see the innovation tech stack for the categories of tools behind it

As a hub, not a silo. The lab's connection to the business units is designed, not hoped for. Innovation Mode 2.0 describes the pattern for large organizations: divisions innovate “independently but in synchronization,” each running its own instance of the same framework, using the same templates, the same assessment model and the same metrics, and sharing through the Portal. Smaller companies get the same effect with people: innovators embedded in the units, a rotation into the lab, and a defined path by which a validated opportunity becomes a product team's work.

• One language: units “use the same methods, processes, and tools, such as idea and problem framing templates, the idea assessment model, recognition and rewards logic,” so their ideas are comparable and exchangeable
• One network: “all the isolated innovation entities in the organization, e.g., labs, discovery teams, and innovation centers, opt into the company-wide innovation network”
• People as the bridge: the Guest Innovator rotation, innovation advocates who connect talent to initiatives, and embedded innovation teams; see the risks of embedding innovation teams in business units
• A handover path: validated opportunities go to MVP leads and venture leads with the unit that will own the product. Without this step the lab ends at the demo; an OECD OPSI analysis of 137 public innovation labs calls the assumption that “someone, anyone, will pick them up” a fallacy
• Problems in, not only ideas out: the units feed the agenda with their problems and their customer signals; sales and customer success hold market intelligence nobody else sees
• The hub's cost: some noise. The book accepts it: “a stream of problems worth solving or formulated ideas could lead to high-potential growth opportunities”

The services become agents; the people become judges. Most of what a lab's staff did ten years ago, finding related work, routing ideas to the right people, drafting workshop materials, summarizing the market, scoring ideas against a model, can now be done by AI agents inside the Portal. Innovation Mode 2.0 describes them: the AI Innovation Advisor, the team builder, the organization navigator, the Innovation Broadcaster, the Workshop Designer. What stays human is the agenda, the judgment on what to pursue, the relationships with the units and the risk of the decision. The lab gets smaller in coordination roles and stronger in judgment roles.

• The advisor: an “Ask Me Anything” experience across the company's innovation assets: who works on what, to whom to pitch an idea, which similar initiatives exist; the book's examples are questions like “Who is working on project x?”
• The broadcaster: “identifies the right audience before triggering any communication,” so a flagged opportunity reaches the experts, sponsors and teams that can act on it, and nobody else
• The Workshop Designer: plans events into the calendar and generates pre-reads, problem and solution spaces and the communication plan; see AI in innovation events
• AI evaluators: ideas assessed against the company's model consistently and at once, with humans reviewing the flagged ones; see how ideas are assessed
• Everyone can prototype now: when a product team can build a working prototype in days with AI tools, the lab's monopoly on experimentation is gone, and a lab whose value was the prototype has no value left. What it keeps is what the units cannot do alone: the agenda, the assessment model every idea is scored against, the portfolio view and the handover path
• What AI cannot do: set the agenda, decide what the company will bet on, carry the risk, or make a business unit adopt a service it does not trust
• When to add it: not first. “a highly sophisticated, AI-powered ideation system would be of limited benefit for the first few iterations.” Manual services, then agents, as adoption grows

No. AI is a capability of the lab, not its subject. Many companies now set up a lab whose mandate is AI itself: pilots, use cases, a center of excellence for the technology. That is a technology program, and it is useful, but it is not the innovation function. The innovation lab starts from problems worth solving and the company's strategy; AI is one of the tools it uses to discover, validate and realize opportunities, and one of the subjects those opportunities may involve. A lab that is only about AI inherits the technology's hype cycle and dies with it. A lab that uses AI to innovate on the company's problems outlives any single technology.

• Problem first: the Innovation Agenda names problems worth solving; whether AI solves them is a finding, not a premise
• Two jobs, two teams: an AI center of excellence builds skills and platforms; the innovation lab finds and validates opportunities. One can serve the other; see how to prioritize AI projects for the first job
• Where AI belongs in the lab: in the services (the agents of the Portal), in the workshops (AI-amplified brainstorming, design sprints and hackathons) and in the evaluation of ideas
• Where the AI opportunities belong: in the same pipeline as every other idea, assessed on the same dimensions, with the data, feasibility and risk questions AI adds
• The hype cost: a lab branded on one technology must justify itself in that technology's terms; a lab branded on the company's problems is judged by outcomes
• The exception: a company whose core product is AI, where the technology and the strategy are the same thing

Activity and engagement first, outcomes later, and never demos. In Innovation Mode 2.0, the performance framework's metrics “reflect the size and quality of the corpus of ideas, the level of employee engagement and collaboration, and the efficiency of essential innovation services such as opportunity discovery, validation, and realization.” It shows the discovery funnel, how ideas flow toward the market, and scorecards for experiments and MVPs. Outcomes arrive quarters after the activity that produced them, so a lab judged on outcomes in its first year is judged on nothing. A lab judged on demos forever is theater.

• Three layers: activity (ideas, problems, workshops, experiments), engagement (who participates, from where, how often they return), outcomes (validated opportunities, MVPs launched, revenue or savings they produced)
• The funnel: ideas captured, assessed, flagged as opportunities, validated, realized; the ratios between stages tell you where the lab is weak
• The baseline: the Innovation Pulse Survey and a snapshot of recent activity in the first months, repeated on a cadence, so progress is measured against something
• Adoption as the leading indicator: how many teams use the lab's services without being asked; the book makes adoption part of the team's definition of done
• What not to count alone: “the number of ideas generated, patents filed, and workshops conducted,” the vanity metrics; see the measurement mistakes that feed innovation theater
• Culture as a metric: the lab also moves the culture, measured separately; see how to measure innovation culture

People and capabilities, not square meters. The Innovation Masterplan in the book asks for “initial estimates of the resources needed for the innovation program, e.g., hiring, acquiring capabilities such as innovation management software, initiating educational programs, and hosting events,” and for “a mechanism for requesting additional funding” as the program grows. I give no figure here, because the right number follows from the capabilities you choose and the level at which you start them. Manual services are “inexpensive (to set up) and easy to start with, but they are slow and expensive (to operate and scale),” which is the trade-off the budget has to plan for.

• Five lines: the core team; capabilities and tools; events (workshops, hackathons, expos); education (the Innovation Academy); and a validation fund for experiments and prototypes that the units cannot pay for from their own budgets
• What to defer: the space, the sophisticated platform and anything that only pays off at scale; the capability matrix in how to start decides what waits
• What AI changed: the tooling line shrinks, because the digital home and its agents no longer need a large build; the people line shifts toward judgment roles rather than shrinking
• Central and co-funded: the core team and the agenda are funded centrally; validation and MVP work for a unit can be co-funded by that unit, which also tests whether the unit wants it
• The ask to leadership: a budget with a mechanism for more, tied to success criteria the leadership approved; see the chief innovation officer guide for the leadership side
• The cost nobody budgets: the time of people in the units who join workshops and rotations. Middle managers have to be brought on board for it; see how to win middle managers over

A director, two people and a Masterplan. Chapter 10 of Innovation Mode 2.0 opens the program in five steps: appoint the innovation program director; set up the core team, starting with an innovation architect and a program manager; perform a preliminary assessment of the company's innovation state; articulate the innovation vision; and draft a roadmap. The output is the Innovation Masterplan, “the foundational document for the innovation transformation initiative,” which fixes the vision and goals, the agenda, the current state, the team, the budget, the success criteria, the stakeholder map and the approach and style of innovation. The lab's charter is this document.

• The assessment: surveys and interviews across the hierarchy, a gap analysis against the ideal innovation architecture, and the company's position on the Innovation Maturity Index
• Choosing what to build first: the Innovation Capability Index: each capability scored for urgency (“it can wait” to “it is needed now”) and for tech dependency (“it can work manually” to “it requires advanced technology”)
• The four groups: urgent and manual, implement now; urgent and technology-bound, start exploring; not urgent and manual, do nothing for now; not urgent and technology-bound, it can wait. The groups change as the program advances
• Start small on purpose: “Starting small is a wise approach as long as there is a strategy for quick iterations and continual improvement”
• Sponsorship: leaders approve the Masterplan and the budget, and show up: the program “needs executive sponsorship from the top level of the organization”
• Announce it properly: a leadership member at an all-hands with “a bold innovation narrative that goes beyond buzzwords,” then the director explains the program; see how to build a culture of innovation

Awareness first, then engagement, then the first validated opportunities. The book's program takes an innovation-inactive company to the aware level in a first phase that, with a well-sourced core team, typically takes three to six months; it spreads hackathons, rotations, validation and handover over later phases that can take years. The plan below is my compressed version for a lab that starts with a sponsor and a team in place. By the end of the year the lab should have a community that returns, a handful of problems from the agenda in validation, and one or two opportunities a business unit wants to own.

• Months one to three: assess the state of innovation, package the program, connect with middle management, approve the Masterplan
• Months three to six: announce the program, start the Innovation Talks and the Academy, open the idea channel, run the first brainstorming and design sprints on problems from the agenda
• Months six to twelve: the first hackathon or validation wave, the first experiments with real customers, the Guest Innovator rotation, and the first handover to a product team
• The cadence: events “must not become a routine.” Too many “could disrupt the rhythm of business and diminish their own purpose. Too few, and the spirit of innovation will suffer from gaps and start to decline”
• What to show the C-suite at twelve months: the baseline against today on activity and engagement, the funnel, and the opportunities in validation; outcomes come in year two
• The warning sign: no business unit has asked for a service. If that is true at month twelve, the mandate, the agenda or the reporting line is wrong, and the lab is becoming a bubble