AI Ideation Modes

Use AI ideation to generate experiment ideas tailored to your risk appetite, resources, and strategic goals.

Why AI Ideation?

Coming up with high-quality experiment ideas is one of the biggest bottlenecks for growth teams. Our AI ideation engine analyzes your business context, funnel data, and objectives to generate structured experiments with pre-written hypotheses and calibrated ICE, ROTI, and AAA scores - saving hours of brainstorming and research.

Available Ideation Styles

You generate ideas from the "Generate next batch" panel on the Experiment Board. You give the batch a goal (required, a batch with no goal is a pile, not a plan), and your company context is added automatically. Then you pick one of four Styles, each optimized for a different scenario and strategic priority.

🎯 Safe Bets Mode

Generate low-risk experiments based on proven growth tactics.

Safe Bets mode draws from a database of validated growth tactics that have worked across hundreds of companies. These experiments minimize downside risk while offering solid potential for incremental improvements.

Best for:

• Teams new to experimentation who need to build confidence and momentum
• Validating fundamental assumptions about your product and users
• Achieving quick wins that demonstrate ROI to stakeholders

🚀 Bold Bets Mode

Generate high-impact, potentially higher-risk experiments for breakthrough growth.

Bold Bets mode suggests innovative experiments that challenge conventional assumptions. These tests may have lower confidence scores but offer the potential for step-change improvements.

Best for:

• Mature experimentation teams ready to take bigger swings
• Situations where incremental improvements aren't sufficient
• Testing disruptive approaches that could differentiate your product

⚡ Copy What Works Mode

Generate experiments based on proven playbooks from similar companies and industries.

This mode analyzes your business type and context to suggest experiments that have already worked for companies like yours.

Best for:

• Teams looking to replicate proven strategies
• Quickly establishing best practices
• Reducing risk by using external evidence

📊 Evidence-First (Market Validated) Mode

Generate experiments grounded in market evidence - customer reviews, competitor analysis, and past learnings.

💡 Tip: Evidence-First mode is the most rigorous option. It requires experiments to cite at least two evidence sources and displays a Market Grounding Score (0–10).

Each generated experiment includes an Evidence Panel showing:

• Specific source quotes and citations
• Evidence type badges (customer reviews, competitor gaps, past learnings)
• Overall market grounding score
• Implications for your experiment design

Best for:

• Data-driven teams that want market-validated hypotheses
• Reducing risk by grounding experiments in real market signals
• Building a culture of evidence-based experimentation

How AI Ideation Works

1. Context Analysis - The AI reviews your business context, industry, growth stage, target audience, and objectives
2. Gap Identification - Based on your funnel data and objectives, the AI identifies high-value opportunities
3. Experiment Generation - Creates structured experiments with If/Then/Because hypotheses
4. ICE Calibration - Each experiment includes AI-calibrated ICE scores
5. Evidence Attachment (Evidence-First mode) - Links market evidence and past learnings

Using AI Ideation Effectively

Provide Detailed Context The more information you provide in Company Settings about your business, users, and constraints, the more relevant and actionable your AI-generated experiments will be.

Treat Suggestions as Starting Points AI-generated experiments are designed to spark ideas and accelerate ideation - not replace your judgment. Refine them further with the Hypothesis Refiner.

Combine Multiple Modes Run some Safe Bets to maintain momentum while testing Bold Bets for breakthrough potential.

Getting Started

1. On the Experiment Board, click "Generate next batch"
2. Enter the goal or challenge this batch should attack (required, e.g. "lift activation for new signups"). Your company context is added automatically
3. Pick a Style: Safe Bets, Bold Bets, Copy What Works, or Evidence-First
4. Optionally set a Funnel focus to aim the batch at one stage
5. Click "Generate batch". The batch lands in the Drafted column with calibrated ICE, ROTI, and AAA scores
6. Review the experiments, commit the strongest to Running, and polish with the Hypothesis Refiner and Metric Validator

All documentation