Best Practices for AI-Powered Discovery

The quality of your discovery insights depends heavily on how well you set up your project. Clarity First uses your project context to shape the interview questions and focus the analysis.

Writing Effective Project Objectives

Your project objectives tell the AI what you're trying to learn. Our Bacon agent parses these objectives to create a synthesis brief that guides the entire analysis. The more specific you are, the more targeted your insights will be.

Defining Scope

Clear scope boundaries help participants stay focused and help the AI filter relevant insights. Define what's explicitly in scope and what's out of scope.

• In scope: Topics, departments, processes, or time periods you want to explore
• Out of scope: Areas to explicitly avoid (e.g., "Not covering the US subsidiary" or "Excluding budget discussions")

Choosing the Right Framework

Frameworks shape the interview structure and the topics covered. Choose based on your engagement type:

• Data Strategy: For data maturity assessments, analytics capabilities, data governance
• AI Readiness: For AI adoption assessments, use case identification, capability gaps
• Customer Experience: For journey mapping, pain points, service improvement
• Digital Transformation: For technology modernisation, process digitisation
• Custom: Build your own framework for specialist engagements

Client Context Matters

The background information you provide about the client shapes how the AI interprets responses. Include industry, company size, and the engagement trigger (why now?).

Participant preparation is the single biggest factor in interview quality. A well-briefed participant gives richer, more useful responses. An anxious or confused participant gives guarded, surface-level answers.

The Invitation

Your invitation email sets expectations. Be clear about:

• What it is: A voice conversation with an AI interviewer
• Why them: Their perspective is valuable because...
• How long: Typically 20-30 minutes
• What topics: A brief overview (not the specific questions)
• Confidentiality: How their responses will be used and who sees them

Setting Expectations

Participants who know what to expect perform better. Key points to communicate:

• It's a conversation, not a test - there are no wrong answers
• They can speak naturally; the AI will follow up on interesting points
• They can ask for clarification if a question isn't clear
• They can skip any question they're not comfortable answering
• The interview is recorded and transcribed for analysis

Addressing Common Concerns

Anticipate and address these common participant worries:

Tips for Different Stakeholder Types

Not every stakeholder is suited to AI interviews. Some prefer human interviewers. Some have accessibility needs. Some topics are too sensitive for an AI conversation. The hybrid approach lets you combine AI and human-conducted interviews in the same project.

Why Hybrid Works

Offering choice is a strength, not a compromise. A hybrid approach lets you:

• Accommodate stakeholder preferences without losing analytical consistency
• Handle sensitive topics with human interviewers where appropriate
• Include participants who may struggle with voice-based AI interaction
• Position your consultancy as thoughtful and inclusive

When to Import Transcripts

Consider human interviews for:

• VIPs and board members: Where personal rapport matters
• Sensitive topics: Redundancy, personal grievances, whistleblowing
• Accessibility needs: Hearing impairments, speech difficulties, strong accents
• Cultural considerations: Where AI interaction may be inappropriate
• Relationship building: When the interview is also a sales or relationship touchpoint

Consistency Across Methods

Whether an interview is conducted by AI or human, the same analysis process applies:

• Same framework and topic structure
• Same AI agents (Humboldt, Bayes, Darwin, etc.) process the content
• Same synthesis output and evidence linking
• No difference in the final deliverable

The Advocacy Angle

Hybrid positioning is a selling point: "We use AI to handle volume and consistency, while reserving human interviews for situations that need a personal touch." This shows thoughtfulness and positions AI as a tool in your toolkit, not a replacement for judgment.

Understanding what happens during an interview helps you prepare participants and interpret the results.

What Participants Experience

The interview is voice-first - participants speak naturally, and the AI responds conversationally. The experience is designed to feel like a structured conversation, not an interrogation.

• A brief introduction explaining the process
• Open-ended questions that invite elaboration
• Follow-up questions based on what they've said
• Natural topic transitions
• A clear ending with thanks

How the AI Adapts

Socrates (our interviewer agent) doesn't just read from a script. It:

• Asks follow-up questions when a response suggests there's more to explore
• Skips topics that have already been covered naturally
• Adjusts language and pace based on the participant's style
• Tracks which topics have been covered to ensure completeness

What You See as Consultant

While interviews are in progress, you can monitor:

• Real-time transcription of the conversation
• Topic coverage indicators showing what's been discussed
• Session status (active, completed, abandoned)
• Key insights captured during the interview

The synthesis is where Clarity First delivers its core value. Eleven AI agents work together to extract, analyse, and synthesise findings from your interviews. But the AI is a tool, not the final word - you stay in control.

Understanding the Agents

Each agent brings a different analytical perspective:

Evidence Lenses

The Evidence tab provides multiple lenses to explore your findings. Each lens answers a different executive question:

• Is this really happening? - Validate findings against operational reality
• What are people experiencing? - Day-to-day impact on staff
• Why hasn't this been fixed? - Systemic blockers and root causes
• Where do groups see this differently? - Perception gaps between levels
• How confident are we? - Evidence strength and coverage

Nightingale's visualisations - Energy Maps, Tension Maps, and Decision Trees - make patterns immediately visible, helping executives grasp complex dynamics at a glance.

You Stay in Control

The synthesis output is a starting point, not a final deliverable. You can:

• Edit findings: Adjust wording, change severity, add nuance
• Refine the executive summary: Send it back with specific guidance ("focus more on data governance", "tone down the technology recommendations")
• Review evidence: Click through to the original quotes supporting each finding
• Add your own insights: Combine AI analysis with your consultant expertise

From Analysis to Insight

The AI does the heavy lifting. This frees you up for the higher-value work:

• Strategic interpretation - what do these findings mean for this client?
• Prioritisation - what should they tackle first?
• Recommendations - specific, actionable next steps
• Client conversations - discussing findings, answering questions, building relationships

Instead of spending days coding transcripts, you spend hours refining insights and preparing recommendations. The deliverable is better, and it's ready faster.

Great analysis is wasted if it's not communicated well. Clarity First provides multiple ways to share findings with stakeholders.

Export Options

Choose the format that fits your client's needs:

• PDF Report: Comprehensive written report with executive summary, findings, evidence, and recommendations. Best for formal deliverables and documentation.
• PowerPoint: Presentation-ready slides with key findings and visuals. Best for steering committees and workshop sessions.

The Query Portal (Ask AI)

After synthesis, you can give stakeholders access to the Query Portal. This lets them ask questions about the findings directly - and Galilei orchestrates the response, routing queries to the right specialist agents and synthesising evidence-backed answers.

• "What did people say about the CRM system?"
• "How do operations and sales differ in their views on data quality?"
• "What were the main concerns about the proposed timeline?"

This extends the value of your discovery work - stakeholders can explore specific topics without scheduling follow-up meetings with you.

Handling Questions

When presenting findings, stakeholders will have questions. Clarity First helps you answer with confidence:

• Every finding links to supporting evidence (direct quotes)
• You can show how many people raised a particular issue
• Contradictory views are captured, not smoothed over
• The Query Portal can answer ad-hoc questions in real time

Learn from common mistakes that reduce the value of discovery projects.