Getting started

What you are building

An Insight interview is a conversational study — not a long static form. You define what you want to learn; the AI agent conducts the interview end-to-end, adapts follow-ups, can show chips/ratings mid-chat, and fills blueprint fields for charts. You are not babysitting every turn.

Core concepts

• Workspace — your team account. Studies, CRM contacts, projects, and settings live here.
• Insight interview — one conversational study with its own link, branding, voice/text mode, and analysis.
• Objectives — primary goal, persona, tone, information goals, session shape, and prohibited topics that steer the agent.
• Session shape — pulse vs deep dive pacing: how many follow-ups after each answer without rewriting the whole study.
• Structured form (blueprint) — optional fields the agent collects during chat (text, choices, rating 1–5 or 1–10, file upload, etc.).
• When to ask / Follow-up details — per-field settings for timing and how deep to probe after an answer.
• Projects — product context so the agent stays on-domain.
• Session — one participant conversation. Post-interview analysis runs after they finish.

Recommended flow (in order)

• Create an Insight interview (untitled or from a template) — new signups land here automatically.
• Set objectives and session shape in the builder (or chat with the AI co-pilot).
• Add structured form fields; configure When to ask and Follow-up details per field.
• Optionally link a project and add context assets (links/images the agent can show).
• Brand the participant experience (welcome, logo, theme, voice chat mode).
• Test with draft preview, then publish.
• Share via public link, CRM invite, or website embed.
• Review analysis: notes, editable auto-metrics, form statistics, and transcripts.

Who needs what

Researchers and product teams work in the admin dashboard. Participants only need your link — no account. Engineering can automate invites via the REST API if needed (see API docs).