5 AI Prompts Grounded in Behavior Change Research

Most AI prompts for habits produce encouragement. Encouragement is not a behavior change mechanism.

These five prompts are designed differently. Each one is derived from a specific finding in the behavior change research literature. Each one asks the AI to help you execute a validated mechanism — not to inspire you, but to help you design and analyze your way to sustained behavior.

Use them in order for a new habit. Use them selectively as diagnostics when an existing habit is struggling.

Prompt 1: Implementation Intention Design (Gollwitzer, 1999)

The research: Implementation intentions — specific if-then plans — significantly increase follow-through compared to goal-setting alone, across more than a hundred studies.

I want to build the habit of [describe target behavior]. 
Help me write three implementation intentions in this format:
"If [specific situation or cue], then I will [specific behavior], 
and I will know I succeeded when [verifiable success criterion]."

For each plan: identify whether the cue is event-based or time-based 
(event-based is usually more reliable), and flag any behavior description 
vague enough that I could tell myself I partially did it without actually 
completing it.

What to do with the output: Choose the option with the most reliable cue. Write it somewhere you will see it before the cue window occurs.

Prompt 2: Cue Reliability Diagnostic (Wood, 2019)

The research: Wendy Wood’s research on habit automaticity shows that cue reliability — whether the cue is consistently present — is one of the strongest predictors of habit formation success. Habits anchored to variable cues take longer to form and are more fragile.

My implementation intention is: [paste your if-then plan].
My success rate over the past two weeks has been [X out of Y attempts].
On the days I did not complete the behavior, the cue was [describe what happened 
with the cue on failed days].

Is my cue event-based or time-based? Is it reliably present in my daily 
routine, or does it vary? What is a more reliable alternative cue I could 
anchor this behavior to — one that is already part of my automatic daily sequence?

What to do with the output: If the AI identifies a more reliable cue, revise your implementation intention. Do not adjust the behavior itself — adjust the trigger.

Prompt 3: Self-Monitoring Review (Burke et al., 2011)

The research: Systematic self-monitoring is one of the most robust predictors of behavior change success. The feedback loop between tracking and action is the mechanism — not the tracking itself.

Here is my behavior log for the past seven days:
[Day 1: did / did not. Context: ___]
[Day 2: did / did not. Context: ___]
[... repeat for all seven days]

My target was [X occurrences per week]. 
Identify: (1) any pattern in when I succeeded vs. failed, 
(2) the most likely structural reason for the failures — 
cue problem, motivation problem, time problem, or context problem,
(3) one specific change to my implementation intention based on this data.
Do not give me general encouragement. Give me a diagnostic and a revision.

What to do with the output: Implement the single recommended change. Run another week of logging. Review again.

Prompt 4: Autonomy-Supportive Reflection (Deci and Ryan, SDT)

The research: Self-determination theory research, including a 2023 study in Motivation and Emotion, found that autonomy-supportive coaching (asking questions, surfacing options, reflecting) produces more durable behavior change than directive coaching (telling people what to do). AI that asks you to articulate your own reasons is more effective than AI that provides reasons for you.

I am trying to build [habit]. I have been working on it for [X weeks]. 
My current success rate is [X%].

Instead of telling me what I should do, ask me four questions that would 
help me articulate: why this behavior matters to me in my own words, 
what specific obstacles I keep encountering, what I have tried that has 
not worked and why, and what the smallest version of success would look like.

After I answer, reflect back what you heard without evaluation. 
Do not add advice unless I ask for it.

What to do with the output: Answer the four questions before asking for any advice. The act of articulating your own reasoning is the intervention — not the AI’s response.

Prompt 5: Automaticity Test and Fade-Out Design (Graybiel, Wood)

The research: Ann Graybiel’s work on basal ganglia-encoded habits and Wendy Wood’s automaticity research both indicate that the goal of behavior change support is for behaviors to become context-triggered and no longer require deliberate activation. AI coaching should work toward its own irrelevance.

I have been tracking [habit] for [X weeks]. My average completion rate 
has been [X%]. 

Here is my honest assessment of how automatic it feels: 
[describe — does it feel like a decision each time, or does it just happen?]

Based on this, help me design a four-week fade-out plan with the following 
structure: gradually reduce the frequency of AI check-ins (from weekly to 
biweekly to none), include a one-week no-AI experiment at week three, 
and define a clear criteria for declaring the habit successfully encoded 
versus still needing support.

If the behavior is not yet automatic, identify what is missing: 
is it cue strength, reward clarity, or routine length?

What to do with the output: Follow the fade-out plan. If the habit holds without AI prompting, it is encoded. If it falls apart, return to Prompt 2 and re-examine cue reliability.

These five prompts cover the complete behavior change arc: design (Prompt 1), diagnose (Prompt 2), review (Prompt 3), internalize (Prompt 4), and graduate (Prompt 5). Start at the beginning with any new habit and work through the sequence.

The single best prompt to start with today is Prompt 1. Pick one behavior. Write one implementation intention. Everything else follows from having that committed to paper.

Related:

• The Complete Guide to Research on AI and Behavior Change
• How to Apply AI Behavior Change Research to Your Own Habits
• The TRACE Framework for AI Behavior Change

Tags: AI prompts behavior change, implementation intention prompt, self-monitoring AI, behavior change research prompts, autonomy-supportive coaching