AI Coaching Cognitive Architecture Agentic Systems Behavioral Design

The coach
that shows up
between
sessions

Paloma is an AI coaching companion that coaches independently between sessions. It makes every live session more powerful by surfacing what shifted, what's compounding, and what's ready to move.

Project Context

This project started inside a coaching practice, not a design studio. After building Common Project, a real-time note-taking tool designed to capture what emerges between coach and client, it became clear the problem was bigger than capturing sessions. It was everything that happened between them. Paloma is the answer to that gap.

Role
Lead AI Product Designer
System architecture, UX, behavioral design, safety model
Platform
Mobile-first
Client companion app with optional session prep brief
Model
Relational Memory System
Contextual retrieval grounded in behavioral science frameworks
Domain
Coaching · Behavioral Change
AI coaching companion + human coach amplification layer
The Problem
What the industry assumed

Coaching is what happens in the session.

The traditional model treats the 60-minute session as the product. Between meetings, clients are on their own, expected to hold insights, honor commitments, and stay regulated without any infrastructure for doing so.

The result: value leakage. Coaches spend the first 20 minutes of every session re-establishing ground they already covered. Clients arrive dysregulated, or having forgotten the thing that mattered most. The work doesn't compound.

What was actually happening

Change takes flight between sessions.

The moments that matter most, the pause before reacting, the decision under pressure, the small act nobody noticed, happen in daily life, not in the coaching room.

The continuity gap isn't a coaching problem. It's a systems design problem. What was missing wasn't more sessions. It was an intelligent witness that could hold the thread, notice what's compounding, and surface the right evidence at the right moment.

“The work doesn’t happen in the room. It happens on a Tuesday, when nobody’s watching.”

The Core Loop

Behavioral change doesn’t happen through insight alone. It happens through small acts of courage, consciously noticed and accumulated as evidence against a limiting belief. Paloma was designed around this loop: not as a metaphor, but as the literal interaction model.

Behavioral Science Foundation

The science isn’t inspiration.
It’s architecture.

Every design decision, what Paloma says, stores, asks, and refuses, was grounded in a specific behavioral science framework. This isn’t a chatbot with coaching aesthetics. The frameworks are load-bearing.

Solutions-Focused Coaching
Paloma never asks what went wrong.
Forward-facing questions activate different neural pathways than retrospective ones. “What would someone who knows you well notice about you this week?” bypasses shame and accesses a more accurate self-reflection. “What went wrong?” doesn’t.
BJ Fogg · Tiny Habits
The celebration happens at the moment of the behavior.
Celebration fires dopamine at the exact moment of a new behavior, tagging it neurologically for retention. The 5-second pause in the celebration screen isn’t a UI flourish. It’s a behavioral intervention. Timing is everything.
Internal Family Systems
Resistance is never shamed. It’s asked what it’s protecting.
Users arrive with protective parts, exiled parts, and inner critics. “There’s a part of you that wants to give up. What’s it afraid will happen if you keep going?” That’s one interaction design decision shaped entirely by IFS.
Rick Hanson · Positive Neuroplasticity
Good things slide off. Paloma makes them stick.
The brain requires deliberate attention to absorb positive experience. Growth slides past unregistered unless something structures the pause. Every reflection prompt and celebration moment is designed to make the good land.
Attachment Theory · Social Witnessing
Being seen by something consistent changes things.
We learn who we are through being seen by an attuned other. Paloma provides a consistent, non-judgmental witness, not as a substitute for human connection, but as a corrective presence that reflects growth back clearly.
Cognitive Reappraisal
Step outside the story to see it clearly.
Perspective-taking activates the prefrontal cortex and quiets the shame-driven limbic response. Paloma regularly invites users to see themselves from a more compassionate vantage point, a designed interruption of the loop that keeps people stuck.
Memory Architecture — 5 Layers

Over-remembering destroys trust. Under-remembering makes the system useless. Paloma's memory system was designed to feel personally attuned without feeling surveilled: a precise calibration between depth and restraint. Each layer maps directly to a moment in the user experience.

Outcomes

Change sounds like this.

“When I told it I never follow through, it showed me seven times I had. I couldn’t argue with that.”
Client, on Paloma surfacing evidence file entries during a shame spiral
“My clients arrive knowing what they want to work on. We used to spend the first 20 minutes figuring that out.”
Coach, on the session prep brief and between-session continuity
“A client had a breakthrough on a Monday. By Wednesday Paloma had already helped them sit with it.”
Coach, on the between-session coaching layer
Design Convictions

Human-in-the-loop
means amplifying the relationship

Paloma coaches independently. But when a human coach is present, Paloma becomes the infrastructure that makes their relationship compound. The coach holds the relationship. Paloma holds the thread between sessions. Designing that boundary clearly was the most important decision in the product.

Behavioral science belongs
in the prompt architecture

General-purpose models drift. Left unguided, an LLM will advise when it should reflect, praise when it should question, and empathize when it should redirect. IFS, solutions-focused coaching, and neuroplasticity frameworks aren’t inspiration. They’re structural constraints baked into Paloma’s cognitive design.