Comprehensive guide to briefing your AI interviewer using text snippets.
Context & Training allows you to provide background information to the AI interviewer. This section is optional but improves interview quality by helping the AI understand your domain and terminology.
Without context, the AI interviewer might:
Misunderstand industry-specific terminology
Miss important nuances in participant responses
Ask generic follow-up questions
Fail to recognise relevant concepts
With context, the AI interviewer will:
Understand your domain and terminology
Ask informed follow-up questions
Recognise and probe relevant topics
Generate more accurate insights
Text fields where you write background information, define terms, and provide context.
Good to know:
1,400 characters per snippet
You can use multiple snippets to organise your context by topic
A live character counter shows remaining space
The first snippet cannot be removed
Define specialised vocabulary so the AI understands participant language.
In our industry, 'churn' means customers cancelling their subscriptions. 'MRR' stands for Monthly Recurring Revenue. A 'power user' is someone who uses more than 10 features regularly. 'Freemium' refers to our free tier with limited features. 'Onboarding' is the first-week experience after signup.
Help the AI understand what you offer and how it works.
TaskFlow is a project management platform for small teams. Key features include: kanban boards for visualizing work, time tracking for billing, team chat for collaboration, and AI-powered task suggestions. We launched 2 years ago and have 50,000 users, mainly at tech startups and creative agencies.
Include correct spellings of your company, competitors, products, features, and people. The AI will then spell these correctly in analysis and recognise them when participants mention them.
Our company is called 'BrightBasket' (not Bright Basket or Brightbasket). Our CEO is Sam Rivers. Our main competitors are FreshCart and GroceryGo. Our flagship feature is called 'Smart Lists' (two words, capital S and L). Our pricing tiers are: Free, Professional, and Enterprise.
Provide context about why you're conducting this research.
This research follows a study we conducted earlier this year about mobile shopping cart abandonment. We found that 60% of users abandon carts on mobile devices, with slow checkout being the top reason. This study focuses on understanding specific pain points in our new streamlined checkout process launched recently.
Share relevant background about who you're interviewing.
All participants in this study have purchased from our platform in the last 30 days. They represent our 'active customer' segment. About 60% are repeat buyers, 40% are first-time customers. They've consented to product research and understand this is about improving the shopping experience.
The Research Objective field on the Instructions tab caps out at 700 characters, which is too short for detailed legal disclaimers, compliance rules, or a list of topics the AI interviewer should avoid. Put that longer guardrail language in a context snippet instead.
The AI treats every snippet as an instruction it follows for the whole interview, the same way it follows your terminology and product background. A snippet works for restrictions like:
Do not discuss competitor pricing, unreleased features, or pending litigation. If a participant asks for medical, legal, or financial advice, decline and redirect them to the next discussion guide question. This study must comply with [your applicable regulation], so do not collect full dates of birth or national ID numbers.
Because a single snippet holds up to 1,400 characters and you can add more than one, you have room for guardrails that would never fit in the Research Objective or Probing Guide fields. Give the restriction its own snippet rather than folding it into a product or terminology snippet, so it stays easy to find and update.
Use one snippet per topic or category.
Example Structure:
Snippet 1: Product overview and key features
Snippet 2: Industry terminology and definitions
Snippet 3: Competitor landscape and differentiation
Snippet 4: Current user segments and characteristics
Snippet 5: Research background and objectives
Snippet 6: Legal guardrails and prohibited topics
When a participant responds, the AI references your context to understand terminology and concepts.
Participant: "The onboarding flow was confusing."
Without context: AI doesn't know what "onboarding" means in your product.
With context: AI knows onboarding = first-week experience after signup, and asks: "What specific part of the first-week experience was confusing?"
Context enables the AI to ask smarter follow-ups.
Participant: "The pricing was the deciding factor."
Generic follow-up: "Tell me more about that."
Context-informed follow-up: "Was it the annual contract requirement or the per-seat pricing that was most important?" — because the AI knows from context that you offer annual contracts with per-seat pricing.
Context improves AI-generated insights and themes. Without it, the AI groups all features as generic "product features" and treats all mentioned tools equally. With it, the AI categorises by your feature structure (Core, Premium, Enterprise) and flags mentions of your top competitors as competitive insights.
Even if terms seem obvious to you, define them. The AI doesn't know your industry jargon.
Include correct spellings of:
Your company and products
Competitors
Features and terminology
People (if relevant)
Don't just define terms - give examples of how they're used.
Vague: "Power user means experienced user."
Better: "Power user means someone who uses 10+ features regularly, creates custom workflows, and uses keyboard shortcuts."
Focus on information that directly helps the AI understand participant responses, ask informed follow-ups, and generate accurate insights.
Break information into logical chunks using multiple snippets (see "Organising Multiple Snippets" above) — clearer than one large snippet.
Owner or Can Edit: Full editing access — add, edit, and delete snippets.
Can View: Read-only; can view context but not change it.
Draft or Ready: Fully editable — add as much context as needed, then test and refine.
Live: Locked to maintain consistency; cannot edit.
Completed: Cannot edit (historical record).
Changing context mid-study could cause the AI to interpret responses differently, compromising consistency and data quality.
Is context required?
No, it's optional. But it significantly improves interview quality.
Can I add context after publishing?
No. Add all context before publishing. Run a test interview to verify the AI has enough information.
Can I add multiple snippets?
Yes. You can use multiple text snippets, each up to 1,400 characters, to organise your context by topic.
Will too much context confuse the AI?
The AI prioritises relevant information, but focus on quality over quantity. 3-5 focused snippets are better than 10 vague ones.
Can participants see my context?
No. Context is internal - it informs the AI but isn't shown to participants.
Where do I add legal guardrails or a list of prohibited topics if they don't fit in Research Objective?
Add them as a context snippet. Research Objective is limited to 700 characters and Probing Guide is meant for follow-up direction on a single question, so neither has room for full guardrail language. A context snippet gives you up to 1,400 characters per snippet, and you can add more than one, so long compliance rules or prohibited-topic lists fit there.
After adding context, proceed to:
Discussion Guide - Write your interview questions
Run a test interview to verify the AI uses your context appropriately
Screening Questions - Set up participant qualification (optional)