Continuous Discovery

This skill implements the Continuous Discovery Habits (CDH) methodology by Teresa Torres. It covers the full pipeline from raw interview data to tested assumptions.

Pipeline overview

Individual Interviews -> Create Snapshots -> Synthesize Patterns -> Create Opportunities -> Generate Solutions -> Identify & Test Assumptions
     |                    |                    |                    |                       |                        |
[Raw Data]        [Structured Stories]   [Shared Patterns]    [Problem Statements]    [Product Ideas]      [Risks & Tests]

Workflows

Each workflow has a detailed reference document:

1. •

   Create Interview Snapshots - references/create-interview-snapshots.md Extract structured insights from qualitative interviews or user tests.

2. •

   Synthesize Interview Snapshots - references/synthesize-interview-snapshots.md Analyze multiple snapshots to identify common patterns and create comprehensive insights.

3. •

   Create Opportunities - references/create-opportunities.md Extract and prioritize opportunities from snapshots and synthesis using Opportunity Solution Trees.

4. •

   Generate Solutions - references/generate-solutions.md Generate multiple potential solutions through AI-human collaborative ideation.

5. •

   Identify and Test Assumptions - references/identify-and-test-assumptions.md Extract assumptions, categorize them, prioritize "leap of faith" assumptions, and design lightweight tests.

1. Create Interview Snapshots

When to use

After conducting a qualitative interview or user test session.

Input

Raw interview notes, transcripts, or recordings.

Output

• •Format: Markdown (.md)
• •Location: user-interviews/snapshots/
• •Filename: snapshot-[participant-name]-[date].md

Process

1. •Data Validation: Assess completeness and quality of interview data
2. •Context Extraction: Identify session type, research goals, and participant context
3. •Story Identification: Extract specific behavioral stories, not generalizations
4. •Experience Mapping: Create user journey maps for key stories
5. •Opportunity Analysis: Identify pain points, needs, and improvement areas
6. •Insight Synthesis: Recognize patterns and behavioral insights
7. •Snapshot Creation: Compile into structured interview snapshot
8. •Quality Review: Validate completeness and clarity

For the full output template and guidelines, see references/create-interview-snapshots.md.

2. Synthesize Interview Snapshots

When to use

• •After completing 3-5+ interviews on the same topic
• •Before creating opportunities or generating solutions
• •When sharing research findings with stakeholders

Input

• •Minimum 3-5 interview snapshots from step 1
• •All snapshots should follow consistent format and cover similar topics

Output

• •Format: Markdown (.md)
• •Location: user-interviews/synthesis/
• •Filename: synthesis-[initiative-name]-v[version].md

Key features

• •Incremental synthesis: Process only new snapshots when updating existing synthesis
• •Version management: Auto-increment version numbers, never overwrite
• •Pattern recognition: Behavioral patterns, emotional journeys, workarounds

For the full framework, see references/synthesize-interview-snapshots.md.

3. Create Opportunities

When to use

• •After completing interview snapshots or synthesis
• •When identifying customer problems worth solving

Input

• •Interview snapshots from user-interviews/snapshots/
• •Synthesis documents from user-interviews/synthesis/
• •Strategic materials from company-level-context/

Output

• •Format: Markdown (.md)
• •Location: opportunities/[topic]/
• •Filename: opportunities-[topic]-v[version].md

Key principles

• •Focus on customer needs, pain points, and desires (not feature requests)
• •Use problem-focused statement format: "I want to ~ but ~ makes it difficult"
• •Organize using Opportunity Solution Tree structure
• •Pause for user review at two checkpoints

For the full process, see references/create-opportunities.md.

4. Generate Solutions

When to use

• •After identifying a clear target opportunity
• •Before committing to a single solution approach

Input

• •Prioritized opportunities with supporting evidence
• •Direct opportunity input from user

Output

• •Format: Markdown (.md)
• •Location: solutions/[topic]/
• •Filename: solutions-[topic]-v[version].md

Critical rule

MANDATORY: The user must generate at least 3 individual ideas before the agent generates any solutions. If the user requests solutions without individual ideation, stop and explain the requirement.

Process

1. •Review target opportunity
2. •Individual ideation (human) - MANDATORY
3. •AI-human collaborative ideation
4. •Repeat and expand (target 15-20 ideas)
5. •Evaluate and select top 3

For the full process, see references/generate-solutions.md.

5. Identify and Test Assumptions

When to use

• •After creating opportunities
• •When preparing to generate or downselect solutions
• •Whenever a new idea is proposed and you need to surface risks

Input

• •Prioritized opportunities
• •Early solution sketches
• •Interview snapshots and synthesis

Output

• •Format: Markdown (.md)
• •Location: assumptions/[topic]/
• •Filename: assumptions-[opportunity-name]-v[version].md

Key concepts

• •Five categories: Desirability, Usability, Feasibility, Viability, Ethical
• •Assumption Mapping: 2D grid (Evidence Known x Importance)
• •Leap of Faith (LoFA): Maximum 3 assumptions from top-right quadrant (Weak Evidence + More Important)
• •Test Cards: Smallest viable simulation with clear success criteria

For the full process, see references/identify-and-test-assumptions.md.

Shared guidelines

File naming

All CDH outputs use semantic naming:

• •Use kebab-case topic/initiative names
• •Auto-increment version numbers (v1 -> v2 -> v3)
• •Never overwrite existing files
• •Check for existing files before creating new ones

Quality standards

• •Focus on concrete behaviors over opinions or intentions
• •Preserve customer language and exact quotes
• •Document evidence strength for each insight
• •Connect all outputs back to the Opportunity Solution Tree

Follow the writing standards in _shared/writing-standards.md for all outputs.