AI as Undermind Mirror
Use a structured five-days-of-questions prompt to make an LLM reflect back patterns you already sense but can't articulate.
- Difficulty
- Starter
- Time to result
- ~days to results
- Steps
- 4
- Confidence
- 85%
Baxley uses AI as a life-coach mirror: because he'd fed it context over a long period, it can statistically reflect patterns from his 'undermind' (pre-language processing) back into conscious language he can then recognize and act on. Rather than passively asking one-off questions, he sets up a deliberate multi-day interview so the model gets better at helping with a specific self-reflection task.
Origin
Baxley's practice, drawing on the concept of the 'undermind' from Guy Claxton's book 'Hare Brain, Tortoise Mind.'
Core principles
- 01Your undermind processes information before it becomes language (like compiled vs interpreted code)
- 02An LLM can statistically reflect existing patterns back into conscious language
- 03It's a mirror, not a pusher: it clarifies what you already think rather than pushing new directions
- 04It's no substitute for a human therapist or coach who actually cares about you
- 05The more context it holds about you, the more useful the reflection
How to run it
- 1
Start a dedicated project with an explicit role
Open a fresh project and tell the model to be your life coach for a specific self-reflection task.
- 2
Ask it to interview you over multiple days
Instruct it to ask you five questions a day for five days so it can explicitly get better at helping you.
Pro tip Structuring it as a multi-day interview beats one-off prompts because the model builds up task-specific understanding of you.
- 3
Use it to surface, not to be advised
Treat outputs as reflections of patterns already in your undermind, translated into language your conscious mind can respond to.
Pro tip Try prompts like 'what's an outdated mindset I'm holding onto that's not serving me?' which surfaced Baxley's attachment to control.
Watch out Its inferences come from what you previously asked help with, so it can be biased; switching to explicit interviewing counters that.
- 4
Keep a human in the loop
Use it as a supplement to, never a replacement for, a real therapist or coach.
Watch out It is not a human being and does not care about you; don't use it instead of real human support.
In the wild
Baxley asked the model for an outdated mindset he was clinging to. It replied, politely, that given his age and profession he was wedded to control, which won't suit his next chapter of writing and speaking, and coined the phrase 'focus on choreography over control.'
→ He found the reframe genuinely useful and now uses the phrase, including elsewhere in this conversation about scaling design teams.
Common mistakes
Letting it infer you passively from past task requests
Its picture of you skews toward whatever you happened to ask help with; an explicit structured interview gives a truer reflection.
Treating it as a substitute for human support
It doesn't care about you and can't replace a therapist or real coach; it's a mirror, not a relationship.
Is it for you?
Best for
Individuals in a transition or identity shift who want to clarify half-formed thoughts they can't yet articulate.
Not ideal for
Anyone needing genuine care, accountability, or clinical support, where a human is required.
From the transcript
“I want you to ask me five questions a day for the next five days”
“it was statistically reflecting patterns back to me that already existed in my under mind”
“try to focus on choreography over control”
From the episode
35 years of product design wisdom from Apple, Disney, Pinterest, and beyond
Bob Baxley