Skip to main content
In-depth interviews (IDIs) give you a one-on-one conversation with a single persona. Where a focus group reveals group dynamics, an IDI reveals the individual: unarticulated motivations, emotional responses, personal context, and the reasoning behind a stated preference. Boses supports two IDI modes:
  • AI-automated (idi_ai) — You upload an interview script and Boses conducts the full multi-turn interview with each persona automatically, then generates a structured analysis.
  • Manual (idi_manual) — You drive the conversation yourself, chatting directly with a specific persona one message at a time. You end the session when you’re done and Boses generates the report.

​
AI-automated IDI

In an AI-automated IDI, you provide a script of interview questions. Boses conducts a multi-turn conversation with each persona using that script, probing naturally for depth. When all personas are complete, you receive a per-persona transcript and structured analysis.Use idi_ai when you want to run the same interview across your whole persona group without moderating each conversation manually.

​
How it works

  1. You create a simulation with simulation_type: "idi_ai".
  2. You upload your interview script as a .txt or .docx file.
  3. Boses parses the script into individual questions, then conducts a multi-turn interview with each persona — asking follow-up questions to probe their answers.
  4. For each persona, Boses generates a full transcript and a structured analysis report.
  5. An aggregate report summarises themes and findings across all personas.

​
Running an AI IDI

1

Create the simulation

Create the simulation. Do not include a prompt_question — the script you upload in the next step drives the conversation.
curl -X POST https://api.temujintechnologies.com/api/v1/projects/<PROJECT_ID>/simulations \
  -H "Authorization: Bearer <TOKEN>" \
  -H "Content-Type: application/json" \
  -d '{
    "persona_group_id": "<GROUP_ID>",
    "simulation_type": "idi_ai"
  }'
Save the simulation id as SIM_ID.
2

Upload your interview script

Upload your script as a .txt or .docx file. Boses extracts the questions from the file and automatically starts the simulation — there is no separate start step.
curl -X POST https://api.temujintechnologies.com/api/v1/projects/<PROJECT_ID>/simulations/<SIM_ID>/script \
  -H "Authorization: Bearer <TOKEN>" \
  -F "file=@interview_script.docx"
Supported file types are .txt and .docx. Write your script as a numbered list of questions. Sub-questions or probes are automatically incorporated — the AI moderator uses them as follow-up prompts when a persona gives a surface-level answer.
3

Poll for completion

The simulation runs in the background. Poll until status is "complete". The progress field shows how many persona interviews are done.
curl https://api.temujintechnologies.com/api/v1/projects/<PROJECT_ID>/simulations/<SIM_ID> \
  -H "Authorization: Bearer <TOKEN>"
Status progression: pending → running → generating_report → complete
4

Fetch the results

curl https://api.temujintechnologies.com/api/v1/projects/<PROJECT_ID>/simulations/<SIM_ID>/results \
  -H "Authorization: Bearer <TOKEN>"
Each persona’s result includes a full conversation transcript and a structured analysis: key themes, emotional tone, notable quotes, and a per-persona summary. The aggregate result synthesises findings across all interviews.

​
Results

Both IDI types produce the same result structure. Per-persona transcript — The complete turn-by-turn conversation, formatted and ready to reference or share. Structured analysis — For each persona (or for your single manual interview), Boses extracts:
  • Key themes from the conversation
  • Emotional tone and intensity
  • Notable quotes for presentations
  • A summary of this persona’s core positions
Aggregate report — For idi_ai runs across multiple personas, the aggregate synthesises the themes and patterns that appeared most consistently, along with strategic recommendations.