North Star first
One sentence naming the business need, the challenge and the one metric AI must move, before any tool is named. AI’s ability is uneven, so each effort is tied to the work it can actually do.
Innovation NavigatorStrategic Game
AI Strategy
from AI as a tool to acquire→to AI as a capability the strategy directs
The Goal
Point AI at what the business needs before scaling the tools: name the North Star and the one metric AI must move, check every AI effort already under way against it, decide KEEP, STOP or TEST for each, and set the trigger that would tell you the bet is wrong.
Ideal For
Leadership teams whose AI activity is ahead of their AI direction: AI strategy offsites, board prep, post-pilot resets, M&A integration and technology pivots.
The Shift
From AI as a tool to acquire to AI as a capability the strategy directs.
The Science
AI’s ability is uneven: in a field experiment, consultants using AI did better on tasks within its reach and worse on a task just outside it, so each AI effort is tied to the work it can actually do. In a large field study, an AI assistant raised productivity on average, most for less-experienced staff, so where it goes matters. Imagining the strategy has already failed tends to surface more reasons than asking whether it might, and deciding in advance what will stop an initiative tends to guard against staying the course because of what’s already invested.
An AI portfolio check, step by step
- Name the business unitThe unit, function or P&L in scope, and the forum the check goes to next.
- Write the North StarOne sentence: the business need, the challenge in the way, and the one metric AI must move. No metric means a wish, not a strategy.
- List what’s already runningEvery AI effort in scope, sanctioned or not: pilots, tools bought, copilots switched on, unsanctioned use. Surfaced, not punished.
- Score and call eachWhich metric it moves, who owns it, what would stop it; then KEEP (with a kill criterion), STOP or TEST (with a P4.7 test card). No owner means STOP until someone takes it.
- Say one no, then set the triggerOne thing AI could do here that you choose not to fund. Then a pre-mortem (“It’s eighteen months on and our AI strategy failed. Why?”) and the biggest risk as a trigger: “If [signal] crosses [level], [owner] will [move].”
- Commit“By [date], [owner] brings the portfolio check to [forum], with each item marked KEEP, STOP or TEST.” The full canvas is the P4.11 HD half day with P4.3.
The techniques, and why they work
Check the portfolio
List every AI effort already under way, sanctioned or not; score each on the metric it moves, its owner and what would stop it; call it KEEP (with a kill criterion), STOP or TEST; and say one explicit no. Deciding in advance what will stop an initiative tends to guard against staying the course because of what’s already invested.
Fail it first, then set the trigger
Imagine the strategy has failed eighteen months on and ask why; write the biggest reason as “If [signal] crosses [level], [owner] will [move].” Imagining a failure has already happened tends to surface more reasons for it.
Selected research
- Dell’Acqua, F., McFowland, E., III, Mollick, E., Lifshitz, H., Kellogg, K. C., Rajendran, S., Krayer, L., Candelon, F., & Lakhani, K. R. (2026). Navigating the jagged technological frontier: Field experimental evidence of the effects of artificial intelligence on knowledge worker productivity and quality. Organization Science, 37(2), 403–423.
- Brynjolfsson, E., Li, D., & Raymond, L. (2025). Generative AI at work. The Quarterly Journal of Economics, 140(2), 889–942.
- Mitchell, D. J., Russo, J. E., & Pennington, N. (1989). Back to the future: Temporal perspective in the explanation of events. Journal of Behavioral Decision Making, 2(1), 25–38.
- Sleesman, D. J., Conlon, D. E., McNamara, G., & Miles, J. E. (2012). Cleaning up the big muddy: A meta-analytic review of the determinants of escalation of commitment. Academy of Management Journal, 55(3), 541–562.
- Rumelt, R. P. (2011). Good Strategy/Bad Strategy: The Difference and Why It Matters. Crown Business. (Practitioner source.)
Also runs as a half day · P4.11 HD
AI Strategy – Half Day
Half day (3 h 30 min), “Direct AI, then draw the line”: this lab’s portfolio check and focus areas in the morning, then P4.3 Human Advantage in an AI World on the focus areas’ real work, and a plan in three stages with triggers, with a quarterly review.
from an AI roadmap of tools and dates→to a North Star, a line drawn on real work, and stages that end in decisions
The Goal
Set the direction for AI and draw the line in the work it touches: check what’s already running against one North Star, choose the focus areas, map their real activities into what AI replaces, what it augments and what stays human-led, and build a staged plan with triggers to watch.
Ideal For
Leadership teams setting the AI strategy for a business unit or function, together with the people who run its workflows: AI strategy offsites, annual planning, post-pilot resets, and organizations where AI use has spread faster than anyone decided.
The Science
AI’s ability is uneven: in a field experiment, consultants using AI did better on tasks within its reach and worse on a task just outside it, so each placement is tied to the work, with evidence. People who rely on automated output tend to miss its errors, and skills that aren’t practised tend to decay, so the checks and what stays human are decided on purpose. Deciding in advance what will stop an initiative tends to guard against staying the course because of what’s already invested.
The techniques, and why they work
Direct AI from what’s already running
One North Star with the metric AI must move; every AI effort in scope called KEEP, STOP or TEST; one or two focus areas and one explicit no. AI’s ability is uneven, so each effort is tied to the work it can actually do.
Draw the line on real work
The focus areas’ actual activities sorted into AI Replaces, AI Augments and Human Leads, each AI placement with an evidence mark and a named check, each human one with a principle and practice plan. People who rely on automated output tend to miss its errors, and unpractised skills tend to decay.
Plan in stages, then watch
The capabilities the map requires; three stages, each ending in a decision with a kill criterion; up to three triggers from a pre-mortem. Deciding in advance what will stop an initiative tends to guard against staying the course because of what’s already invested.
Selected research
- Dell’Acqua, F., McFowland, E., III, Mollick, E., Lifshitz, H., Kellogg, K. C., Rajendran, S., Krayer, L., Candelon, F., & Lakhani, K. R. (2026). Navigating the jagged technological frontier: Field experimental evidence of the effects of artificial intelligence on knowledge worker productivity and quality. Organization Science, 37(2), 403–423.
- Brynjolfsson, E., Li, D., & Raymond, L. (2025). Generative AI at work. The Quarterly Journal of Economics, 140(2), 889–942.
- Parasuraman, R., & Manzey, D. H. (2010). Complacency and bias in human use of automation: An attentional integration. Human Factors, 52(3), 381–410.
- Arthur, W., Jr., Bennett, W., Jr., Stanush, P. L., & McNelly, T. L. (1998). Factors that influence skill decay and retention: A quantitative review and analysis. Human Performance, 11(1), 57–101.
- Mitchell, D. J., Russo, J. E., & Pennington, N. (1989). Back to the future: Temporal perspective in the explanation of events. Journal of Behavioral Decision Making, 2(1), 25–38.
- Sleesman, D. J., Conlon, D. E., McNamara, G., & Miles, J. E. (2012). Cleaning up the big muddy: A meta-analytic review of the determinants of escalation of commitment. Academy of Management Journal, 55(3), 541–562.
- Gollwitzer, P. M., & Sheeran, P. (2006). Implementation intentions and goal achievement: A meta-analysis of effects and processes. Advances in Experimental Social Psychology, 38, 69–119.
- Rumelt, R. P. (2011). Good Strategy/Bad Strategy: The Difference and Why It Matters. Crown Business. (Practitioner source.)
From the research
The thinking behind this lab, published in full: