Agentic [_Interview] Flow
Human-like conversations with Mara, our synthetic interviewer.
The Agentic Interview Flow is a Labs experiment for teams that depend on high-quality conversations and want to turn them into compelling written work: customer stories, expert interviews, thought-leadership articles, internal discovery, stakeholder sessions, and strategic intake.
At the center is Mara, our synthetic interviewer. Powered by gpt-realtime-2, Mara can conduct natural, human-like conversations while your team defines the objective, reviews the output, and shapes the final article.
Most interview workflows are split across too many tools. Notes live in one place, transcripts in another, follow-up questions stay in someone's head, and the article still depends on who has time to read everything carefully and find the story.
The Agentic Interview Flow explores a more fluid workflow: a synthetic conversation layer that can run structured interviews through Mara, capture the nuance of what was said, and transform the material into article-ready drafts without turning the interview into a rigid script.
Inside the [_experiment]
The tool can help shape an interview plan around a clear objective, audience, and set of hypotheses, so the conversation starts with intention rather than a blank page.
Mara uses gpt-realtime-2 to conduct human-like conversations that feel responsive rather than scripted, allowing participants to answer naturally while the system keeps the interview focused.
During an interview, Mara can probe unclear answers, return to important themes, and ask follow-up questions while there is still time to go deeper.
After the conversation, the tool can turn messy dialogue into well-structured article drafts, with usable themes, quotes, narrative flow, and supporting takeaways ready for editorial review.
Good interviews are not only about asking questions. They are about attention, timing, context, and judgment. Good articles require the same care: finding the thread, preserving the speaker's nuance, and shaping the material into something worth reading.
This experiment looks at how synthetic interviewers can extend the human craft of interviewing and editorial writing: more consistent conversations, less admin, and better articles based on what was actually said.
The Agentic Interview Flow is an active Labs prototype. We are exploring Mara's interview style, the realtime interaction model, the article-drafting workflow, and the right boundaries between synthetic conversation, editorial assistance, and human judgment.
Interested in using AI to turn interviews into stronger articles, research, or discovery outputs? Let's explore where this could fit in your workflow.