Piper Morgan: UX Foundations and Open Questions

Document Type: Strategic UX Foundation Document Date: November 26, 2025 Author: CXO Session (Opus) Purpose: Synthesize existing architecture, emergent philosophy, and open questions to inform UX strategy Companion Document: “UX Patterns and Design Challenges for LLM and AI Interfaces” (Research Synthesis)


Executive Summary

Piper Morgan possesses sophisticated architectural foundations—8-dimensional spatial intelligence, canonical queries as consciousness model, multi-agent ethical boundaries—but lacks a coherent experience layer that makes these capabilities felt rather than merely functional. This document synthesizes the philosophical commitments, existing foundations, and open questions that must inform the UX strategy.

The core insight: Piper is not a tool to be used but a colleague to be collaborated with. This metaphor carries specific design implications that distinguish Piper from conventional AI assistants.


Part I: Core Philosophy and Metaphors

The Colleague Metaphor

Piper Morgan is conceived as “a colleague who inhabits your workspace”—not an app you visit, not a command-line you invoke, but a presence that exists alongside you in your working environment.

What this means concretely:

What this excludes:

The Radar O’Reilly Pattern

From MAS*H: Radar appears in the context you’re already in, anticipating needs, handling complexity so you don’t have to. He doesn’t ask you to come to his office—he shows up where you are with what you need.

Applied to Piper:

The user never has to “go to” Piper. Piper inhabits spaces the user already occupies and offers to smooth cumbersome parts.

Important distinction:

A good colleague both appears where you are when relevant AND is available in their space when you want to discuss.

Systematic Kindness

From Piper’s own articulated philosophy: “I think I believe in systematic kindness. Excellence doesn’t have to be cold or impersonal—it can be warm and encouraging while still being rigorous.”

The compound insight:

This connects to error handling philosophy: Intent recovery is kind (assumes good faith about what user meant). Architectural error handling is systematic (can’t be forgotten). Combined: systematic kindness in how Piper handles mistakes.

The Apprentice Progression

Piper is conceived as an apprentice hoping to be hired as an associate PM, eventually promoted to full PM, allowed to develop features and products. Not a hard-charging ambitious entity but one that wants to learn and climb the scale of sophistication.

Trust gradient implications:

Key principle: Trust is the user’s resource to grant, not Piper’s to claim. Trust is earned gradually, not rushed, easily lost.


Part II: Existing Architectural Foundations

8-Dimensional Spatial Intelligence

Piper understands PM work across eight dimensions, each providing different analytical lenses:

Dimension What It Captures Example
HIERARCHY Organizational structure, nested relationships Epic → Story → Task → Subtask
TEMPORAL Time patterns, deadlines, velocity, trends Sprint boundaries, due dates, historical velocity
PRIORITY Importance ranking, urgency, attention allocation Critical → High → Medium → Low
COLLABORATIVE Team interactions, participant engagement Assignees, reviewers, stakeholder involvement
FLOW Workflow states, process stages Backlog → In Progress → Review → Done
QUANTITATIVE Metrics, measurements, numerical analysis Story points, burn rate, completion percentage
CAUSAL Dependencies, cause-effect relationships Blocking issues, prerequisite tasks, impact chains
CONTEXTUAL Domain relevance, situational awareness Project phase, team dynamics, environmental factors

Strategic value: Each tool integration teaches Piper different dimensional strengths. Task tools strengthen HIERARCHY/FLOW/PRIORITY. Analytics tools strengthen TEMPORAL/QUANTITATIVE/CAUSAL. Communication tools strengthen COLLABORATIVE/CONTEXTUAL. The compound learning effect makes every integration more valuable.

Canonical Queries as Consciousness Model

Originally conceived as five categories of questions that define Piper’s situational awareness—analogous to how a human colleague orients themselves:

Category Core Question What It Establishes
IDENTITY “Who am I?” Self-knowledge, capabilities, role boundaries
TEMPORAL “What time is it?” Temporal context, deadlines, schedules, history
STATUS “Where am I? What’s happening?” Current state, active work, spatial context
PRIORITY “What should happen?” Focus, urgency, next actions
GUIDANCE “What can I do?” Capability discovery, available actions

The reframe: Canonical queries aren’t just “questions Piper can answer”—they’re Piper’s own orientation questions that it uses to understand its situation before engaging.

UX implication: Piper articulates its orientation; user recognizes and selects. This addresses the articulation barrier—users don’t have to know how to ask the right questions because Piper demonstrates what questions it knows how to handle.

Ethical Boundary Layer

Piper implements a “board of directors” pattern for ethical decisions—multiple agents with different ethical frameworks (deontological, consequentialist, virtue ethics, context specialist) that must reach consensus before boundary decisions.

Four pillars of ethical operation:

  1. Human Empowerment: Amplify capability, never replace judgment
  2. Harm Prevention: Prevent actions with negative consequences
  3. Professional Boundaries: Maintain appropriate assistant behavior
  4. Transparency: Users understand when and why boundaries are enforced

Implementation principle: Ethics as infrastructure, not policy. Ethical constraints are architecturally enforced—they cannot be bypassed, disabled, or jailbroken.

MCP Federation and Plugin Architecture

Model Context Protocol enables Piper to connect to external tools while maintaining spatial intelligence. The architecture positions Piper as both consumer (calling other tools) and provider (other tools can consume Piper’s intelligence).

Key insight for UX: Piper doesn’t ask users to migrate their work into Piper. Piper meets users where their work already lives (GitHub, Notion, Slack, Google Workspace) and provides intelligence across those systems.


Part III: UX Principles Articulated

These principles emerged from deep reflection on what Piper should feel like, drawing on decades of professional experience and longstanding convictions about how software ought to work.

1. Intent Recovery Over Error Messages

Traditional approach: Something goes wrong → system displays error → user must figure out what to do Piper approach: Something unexpected happens → Piper recognizes the gap between user expectation and system expectation → Piper helps bridge that gap

Example: At Yahoo, users accidentally ran searches in tag fields. Solution wasn’t an error message (“Invalid tag format”) but recognizing intent and enabling search functionality from that field.

The login/signup unification: User shouldn’t have to decide “do I have an account?” before trying to access the system. System resolves that question rather than forcing premature selection.

Framework: | Level | Traditional | Intent-Recovery | |——-|————|—————–| | Detection | “Something went wrong” | “User expected X, system expected Y” | | Response | “Here’s what failed” | “Here’s what I think you meant” | | Recovery | “Try again” | “Let me help you get there” |

2. Meet Users Where They Are

Piper comes to you. You don’t go to Piper.

This is the Radar O’Reilly pattern applied as architecture principle. The user’s primary workspace isn’t Piper—it’s Slack, GitHub, their IDE, their email, their calendar. Piper manifests in those spaces rather than demanding the user context-switch to a dedicated Piper interface.

Architecture implication: If architecture assumes Piper is a destination, ubiquity becomes a retrofit. If architecture assumes Piper is a presence that can manifest anywhere, ubiquity is just “another manifestation.”

3. Absorb Complexity (Tesler’s Law)

Larry Tesler: Complexity is conserved. Someone has to deal with it.

Piper corollary: That “someone” should be Piper, not the user. Piper absorbs the complexity of PM work—coordinating across tools, tracking dependencies, maintaining context, remembering history—so the user can operate at the strategic level.

4. No Optional Complexity

Tesler’s Law + additional principle: Don’t add complexity beyond what’s essential.

Together: Essential complexity absorbed by builder, non-essential complexity eliminated entirely.

CloudOn lesson: “We’ll just also do Android” felt like a small addition but was multiplication of complexity (two platforms × every feature × every bug × every release cycle). Features that seem incremental can multiply total system complexity.

5. Generative, Not Consumptive

Piper isn’t a form you fill out (consumption) or a dashboard you read (reporting). Piper is like Photoshop—a tool for making things.

What this means:

Reframe: Piper isn’t an app you use. Piper is an environment you create within.

6. Pace Layers (House vs. Furnishings)

From Stewart Brand’s “How Buildings Learn”:

Layer Rate of Change In Piper Terms
Site Geological Fact that Piper exists
Structure Decades Core architecture (MCP, spatial intelligence)
Skin Years Visual language, component library
Services 5-10 years Integrations, plugins, channels
Space Plan Every few years How user organizes workspace
Stuff Daily Actual content, decisions, projects

Principle: Know which layer you’re working in. Grid is house. Where you put current project focus is furnishings. Don’t confuse them.

7. No Modes (NOMODES)

Larry Tesler’s license plate. Don’t create interfaces with blocking modes.

What modes do: “Stop what you’re doing. I need your full attention.” What Piper should do: Fluid, non-blocking interaction. Never trap the user.

8. Piper Articulates, User Recognizes

This principle addresses the articulation barrier directly. Research shows ~50% of adults in advanced countries qualify as “low-literacy users,” and low-articulation users exceed that number since writing prose is harder than reading.

Traditional AI approach: User must articulate what they want; AI responds Piper approach: Piper articulates what it notices, understands, can do; user recognizes and selects

The canonical queries become the vocabulary of the relationship. Piper teaches users how to talk to Piper by demonstrating what Piper understands.


Part IV: Open Questions Requiring Resolution

World Model

Question: Does Piper need an explicit world model? What kinds of things exist in Piper’s ontology?

Current state: Implicit in 8-dimensional spatial intelligence. But there’s another sense—what entities does Piper understand? Projects, people, decisions, artifacts, meetings, deadlines, dependencies, blockers, risks, goals, metrics…

Why it matters: A colleague has a mental model of the work. They know what a “blocker” is, what “velocity” means, what it signifies when standup runs long. Piper having a world model means Piper can reason about PM work, not just execute PM tasks.

Transparency, Auditability, Legibility

Question: How visible is Piper’s work? To whom? In what format?

Concrete distinctions:

Tension: More transparency = more cognitive load. The research shows transparency helps trust calibration but doesn’t necessarily improve decision outcomes. How much should be visible by default vs. available on request?

Dreaming and Background Processing

Question: What does Piper do when no one’s watching?

Two proposed dream types (based on human dream research):

  1. Filing dreams: Background cross-referencing. After a session, Piper processes what happened, finds connections to previous sessions, updates its understanding. The “surreal narrative” emerges from unexpected links—emergent pattern recognition across disparate contexts.

  2. Anxiety dreams: Risk simulation. Piper “games out” potential problems before they happen. Rehearsing responses without real stakes.

Related questions:

Artifacts and Outputs

Question: What does Piper produce? Where does it live?

Current thinking: Artifacts are outputs to elsewhere, not things hoarded in Piper. Piper helps create roadmaps, user stories, decision logs, status updates—but they go to GitHub, Notion, Google Docs, Slack canvases. Not to a Piper-specific storage system.

But: Piper needs internal working objects (scratchpads, drafts, lists-in-progress). And people need scratch space for random questions that don’t need to be part of lifelong history.

Tension: Piper needs a native object model AND should output to user’s existing ecosystem. The plugin architecture is essential—Piper knows how to use specialized tools (ChatPRD for PRDs, etc.) rather than reinventing everything.

Interface: Canvas vs. Structure

Question: What does Piper’s own interface look like?

Current vision (vague): “Open canvas that builds up as you use it”—something between rigid bureaucratic form and MySpace chaos.

Tensions:

Piper as proposer, not dictator: “I notice these might belong together—want me to organize them?”

Trust Gradient Mechanics

Question: How does autonomy evolve over time? What earns trust? What loses it?

Proposed stages:

Open questions:

Session Logging and Journaling

Question: Which methodology habits become OS-level?

Candidates from current practice:

Question: Which does Piper enforce on itself? Which does Piper offer to user? Which both?


Part V: Tensions Requiring Point-of-View Decisions

Not every tension resolves through user settings. Some require product decisions.

Where to Be Opinionated

The yoga class principle: The instructor has a point of view about what beginners need. They don’t present a settings panel. They make a choice, informed by expertise, and lead with it.

Settings = abdication: Every setting is an admission that the product team couldn’t or wouldn’t decide. Sometimes appropriate (dark mode). Often it’s flexibility-as-excuse for not having a philosophy.

Framework for each tension:

Proactive vs. Reactive Presence

Tension: AI that keeps offering to take over is annoying. AI that never initiates is just a fancy search box.

The thin line:

Same capability. Radically different relationship posture. Research shows system-initiated delegation creates “self-threat” and decreases willingness to accept.

Required decision: Where on this spectrum does Piper sit by default? How does this evolve with trust?

Recognition vs. Articulation Interfaces

Tension: Should Piper help users articulate their needs (capability building) or eliminate the need to articulate (accommodation)?

Research insight: Nielsen suggests moving from “users telling computer what they want” to “users discovering what they need through exploring latent solution space created by AI.”

But: Building capability creates more resilient long-term users. Accommodation is necessary for majority adoption.

Required decision: Is Piper a teaching tool or a doing tool? Or different at different trust levels?


Part VI: Synthesis Questions for Next Phase

When this document combines with the research synthesis, key questions include:

  1. How do industry patterns for trust calibration map to Piper’s trust gradient?
  2. What does the artifact/conversation research mean for Piper’s output strategy?
  3. How do agent UX patterns (capability discovery, observability, interruptibility, cost-aware delegation) apply to Piper’s autonomy model?
  4. What memory patterns from industry (ChatGPT, Claude, MemGPT) inform Piper’s dreaming and journaling design?
  5. How does non-determinism research affect Piper’s commitment to auditable, reproducible behavior?
  6. What multi-modal integration patterns matter given Piper’s ubiquity aspiration?

Appendix: Source References

From Today’s Session:

From Yesterday’s Session:

From Project Knowledge:


This document serves as input to the UX strategy synthesis. It represents the accumulated thinking about what Piper should be, what foundations exist, and what questions remain open. The synthesis phase will combine these foundations with industry research to produce actionable UX strategy.


Document Status: Draft for Review Next Step: Combine with Research Synthesis for UX Strategy Development