When AI Doesn't Know You: 20-Minute Onboarding Interview Prompt and Memory Tips

A practical onboarding prompt that guides Claude and ChatGPT through a 20-minute deep interview about your thinking, values, and goals, saving the insights dire

tau · September 12, 2026

#AITips #PromptEngineering #Claude #ChatGPT #Memory #Onboarding #Productivity

When AI Doesn't Know You: 20-Minute Onboarding Interview Prompt and Memory Tips

When using artificial intelligence as a daily collaborator for work or creative projects, a common frustration is that models frequently default to generic responses because they lack insight into how you think, what you prioritize, and what you are currently building. Creator lucas (@lucas_flatwhite) shared a practical onboarding interview prompt and memory workflow on X (formerly Twitter) designed to dramatically improve the everyday usefulness of assistants like Claude and ChatGPT.

User deep interview prompt and onboarding question interface preview

Image source: @lucas_flatwhite via X

The Need for a 20-Minute Onboarding Interview and Core Problem

When starting out with an AI assistant, most people either paste in unwieldy, static system prompts or provide almost no personal background, expecting instant contextual awareness. However, one-way prompt dumps rarely reflect the nuances of how a creator actually approaches decisions or manages active projects.

lucas highlighted this issue by building on an observation originally shared by Blake Robbins (@blakeir). Robbins noted that tools like ChatGPT and Claude would become ten times more useful if they simply spent twenty minutes conducting an in-depth interview with the user. Rather than expecting the user to anticipate every piece of relevant context upfront, the assistant takes the initiative to ask probing questions about how the individual thinks, what principles matter most to them, and what goals they are trying to achieve right now. This initial onboarding exchange establishes a vital shared foundation for both the person and the machine.

The Hands-On Onboarding Prompt and Questioning Structure

The core prompt provided by lucas flips the traditional interaction model, instructing the model to step into the shoes of an active interviewer rather than waiting passively for instructions.

Interview me deeply about the relevant parts of my life that you don't know yet. When descriptive questions work best, ask freely; when multiple-choice is better, use the AskUserQuestion tool. Save all interview insights into memory.

This prompt is effective because it gives the model discretion over how it structures inquiries depending on the subject matter:

  • Balancing Open-Ended and Structured Inquiries: When exploring thinking styles or overarching philosophies, the assistant uses open-ended questions that encourage detailed reflection. Conversely, for operational habits and clear-cut workflows, multiple-choice formats streamline responses and reduce cognitive friction.
  • Tool Utilization and Graceful Adaptation: In environments supporting interactive tools such as AskUserQuestion, the model leverages structured interactive widgets. In interfaces without native tooling support, such as standard Claude web chats, the model naturally adapts by presenting clear numbered choices and guided options directly within the message stream.

Persisting Context to Memory for Long-Term Collaboration

Conducting an interview is only half the battle; ensuring that the extracted insights endure beyond a single conversation session is what creates lasting leverage.

The directive to save all interview findings into persistent memory transforms ephemeral chat exchanges into an enduring personal profile. Whether leveraging ChatGPT's native personalization memory or Claude's project artifacts and project-level context, capturing this background yields immediate dividends across subsequent work sessions:

  • Eliminating Repetitive Context Setup: Users no longer need to re-explain their technical background, preferred tone, domain constraints, or communication style at the beginning of every prompt.
  • Aligned Decision-Making: When evaluating tradeoffs or suggesting action plans, the model weighs recommendations against the user's articulated values and active strategic priorities.
  • Continuous Partnership: The assistant functions less like a disconnected query engine and more like a dedicated thought partner that continues to refine its understanding through ongoing interactions over time.

Original source

This practical guide to AI onboarding interview prompts and persistent memory workflows is based on public posts shared by lucas (@lucas_flatwhite) and related reference discussions.