Design the incentives. Change the behavior.
Every outcome you dislike is somebody responding rationally to an incentive you set. The Incentive Lab is where you redesign that system — model the choice, find the incentives that backfire, and deploy the fix.
The Incentive Lab is an interactive workspace for designing incentive systems. You describe the decision you want to change, the lab maps the choices and incentives around it using the Incentives LLM's behavioral-economics corpus, diagnoses which incentives backfire, and produces a 30/60/90-day deployment plan. Signed-in users save projects and revisit them; the same engine is available programmatically through the MCP server.
Three stages, one system
Design
Describe the decision you want to change — who chooses, what they choose between, and what outcome you actually want. The lab structures it as a choice architecture problem, not a motivation problem.
Diagnose
The engine maps the incentives currently in play, flags the ones that backfire — perverse incentives, Goodhart effects, misaligned rewards — and scores the system against the behavioral corpus.
Deploy
You get a 30/60/90-day plan: which incentive to change first, what to measure, and what second-order effects to watch for. Saved projects let you re-diagnose as the system evolves.
Frequently asked questions
What is the Incentive Lab?
The Incentive Lab is a workspace where you design, diagnose, and deploy an incentive system for a real decision — a pricing page, a sales comp plan, a policy, a product flow. It applies choice architecture and behavioral economics from the Incentives LLM's corpus to your specific situation.
How does the Incentive Lab work?
In three stages. Design: you describe the decision, the actors, and the outcome you want. Diagnose: the engine maps the current incentives, flags perverse incentives and misaligned rewards, and scores the system. Deploy: you get a 30/60/90-day plan with the specific incentive changes to make first.
Do I need an account to use the Incentive Lab?
Yes. Signing in lets the lab save your projects so you can return, refine the design, and track how the diagnosis changes over time. Your projects are private to your account.
What powers the Incentive Lab?
The Incentives LLM — a purpose-built engine trained on the lab's corpus of behavioral models, biases, fallacies, and incentive case studies. The same engine is exposed through the MCP server, so your own applications and AI agents can run the same design and diagnosis tools.
Can I use the Incentive Lab from my own application?
Yes. The MCP server exposes the lab's tools — search_incentives, answer_incentives, list_dimensions, and run_incentive_scan — over a standard protocol that AI clients and agents can connect to. See the MCP server page for connection details and pricing.
What kinds of problems is the Incentive Lab for?
Any situation where people respond to incentives: pricing and packaging, sales compensation, employee performance systems, public policy, product onboarding, retention loops. If a KPI keeps backfiring or a well-meant reward produces the opposite behavior, that is an incentive design problem.
Keep going
Sign in to design, diagnose, and deploy your own incentive system.
Connect your own applications and AI agents to the same engine.
Learn the incentive-design fundamentals the lab applies.
The field guide to how incentives shape behavior.