PACE

Pattern for Agentic Conversational Experience

Abstract

PACE (Pattern for Agentic Conversational Experience) is a design framework that inverts the traditional relationship between users and interfaces. Instead of presenting catalogs, menus, or navigation hierarchies for users to browse, PACE implementations use an AI guide to lead users through conversation toward their goals.

The framework defines four behavioral principles that form a recursive acronym:

  • Proactive — The guide initiates, suggests, and anticipates
  • Adaptive — The guide matches user expertise and adjusts communication
  • Contextual — The guide remembers, references, and builds on history
  • Efficient — The guide is concise, actionable, and moves users forward

The recursive acronym encodes at two layers: Layer 1 (Framework) — Pattern for Agentic Conversational Experience; Layer 2 (Principles) — Proactive, Adaptive, Contextual, Efficient. A key property of PACE is its domain-agnostic nature: the same four behaviors that make AI guidable also make people guidable, enabling cross-domain transfer from UX to hiring and beyond.

Keywords: conversational UX, agentic interfaces, AI-guided discovery, design patterns, cross-domain transfer, natural metaphors

The Problem PACE Solves

Traditional digital interfaces operate on a browse-and-hunt paradigm that places cognitive burden on users. Users must: understand the taxonomy, navigate the hierarchy, evaluate options, and backtrack when lost. This model worked when catalogs were small and users were patient. It fails at scale.

PACE inverts this relationship. Instead of the user navigating to the answer, the guide brings the answer to the user through conversation. The user describes their goal; the guide does the heavy lifting — filtering, recommending, explaining, and adapting in real time.

Traditional: Browse & Hunt

  • User navigates menus and catalogs
  • Cognitive burden on the user
  • One-size-fits-all presentation
  • Context lost between pages

PACE: Guide-Led Discovery

  • Guide brings answers to the user
  • Cognitive burden on the guide
  • Adapts to each user's level
  • Context accumulates across conversation

The Four Principles

P

Proactive

The guide initiates, suggests, and anticipates. First message demonstrates capability — not a passive greeting. Suggests next steps before being asked. Auto-prompts after inactivity. Surfaces relevant options proactively.

  • Capability-demonstrating first message
  • Anticipatory suggestions based on context
  • Inactivity prompts (5-second default)
  • Proactive handoffs when limits are reached
A

Adaptive

The guide matches user expertise and adjusts communication. Detects expertise from language patterns. Adjusts technical depth dynamically. Progressive disclosure — starts simple, adds depth on request.

  • Signal detection (beginner vs. expert indicators)
  • Response calibration based on expertise level
  • Multiple entry points for different user types
  • Dynamic adjustment throughout conversation
C

Contextual

The guide remembers, references, and builds on history. Maintains conversation history. References previous statements explicitly. Builds recommendations from accumulated context.

  • Conversation history maintained across messages
  • Previous statements referenced explicitly
  • Executive Summary Pattern — live display of what guide has learned
  • Transparent reasoning showing how context informs recommendations
E

Efficient

The guide is concise, actionable, and moves users forward. Leads with the answer, follows with explanation. Eliminates filler. Provides clear next actions — specific, not open-ended.

  • Lead with the answer, explain after
  • No filler ("I'd be happy to help...")
  • Word budget: simple factual 1-2 sentences, recommendation 3-4, explanation 1-2 paragraphs
  • Clear, specific next actions

Cross-Domain Transfer

A key property of PACE is its domain-agnostic nature. The same four behaviors that make AI guidable also make people guidable. The framework has been transferred from conversational UX to hiring assessment:

PrinciplePACE UX (AI)PACE Hiring (People)
ProactiveAI suggests next steps, anticipates needsCandidate anticipates needs, doesn't wait for instructions
AdaptiveAI adjusts to user expertise levelCandidate adjusts to team culture and context
ContextualAI remembers conversation, builds on itCandidate absorbs tribal knowledge, builds on it
EfficientAI moves toward actionable outputCandidate ships results, no wasted effort

This transferability validates that PACE captures something fundamental about effective guided interaction — whether the guide is artificial or human.

Biological Inspiration

PACE is inspired by cormorant foraging behavior, extending the Cormorant Foraging framework into interaction design:

Proactive

Diving Foraging

Cormorants actively pursue prey rather than waiting passively.

Adaptive

Visual Hunting

Cormorants adapt dive depth based on light conditions.

Contextual

Strategy Switching

Generalist feeders that switch strategies based on prey type.

Efficient

Energy Management

Manage energy carefully with short dives and rest periods.

Metrics Framework

PACE defines measurable targets for each principle, with minimum viable and exceptional thresholds:

PrincipleKey MetricMinimumExceptional
ProactiveFirst-message engagement> 60%> 80%
AdaptiveExpertise detection accuracy> 75%> 90%
ContextualContext reference rate> 50%> 75%
EfficientTime to first action< 90s< 45s

Per-Response Scoring

Criterion012
ProactivePassive, waitsSome suggestionsAnticipates needs
AdaptiveOne-size-fits-allSome adjustmentFully calibrated
ContextualNo memorySome contextSynthesizes history
EfficientVerbose, unclearMostly conciseOptimal clarity

Conclusion

PACE provides a principled framework for designing AI-guided conversational experiences. The four principles — Proactive, Adaptive, Contextual, Efficient — are measurable, transferable across domains, and grounded in biological observation. The recursive acronym encodes the framework at two levels, making it both memorable and structurally coherent.

The cross-domain transfer from UX to hiring validates that PACE captures fundamental properties of effective guided interaction. Whether the guide is an AI assistant or a human candidate, the same behavioral principles determine success.

PACE is part of the broader Semantic Intent philosophy: clarity before code, intent before implementation.

References

  1. Shatny, M. (2024). Semantic Intent as Single Source of Truth. DOI: 10.5281/zenodo.17114972
  2. Shatny, M. (2025). Cormorant Foraging: A Sound-Space-Time Framework. DOI: 10.5281/zenodo.18904952
  3. Cormorant Hunting Intelligence Algorithm: A Bio-inspired Optimization Approach. ResearchGate, September 2025.
  4. When cormorants go fishing: the differing costs of hunting for sedentary and motile prey. PMC / Biology Letters.
  5. Prey ecology and behaviour affect foraging strategies in the Great Cormorant. Marine Biology, July 2010.

© 2025 semanticintent. Licensed under CC BY 4.0. ORCID: 0009-0006-2011-3258

Cite This Work

APA Style

Shatny, M. (2025). PACE: Pattern for Agentic Conversational Experience. semanticintent.dev. DOI: 10.5281/zenodo.18904342. ORCID: 0009-0006-2011-3258

BibTeX

@misc{shatny2025pace,
  author = {Shatny, Michael},
  title = {PACE: Pattern for Agentic Conversational Experience},
  year = {2025},
  url = {https://semanticintent.dev/papers/pace-pattern},
  doi = {10.5281/zenodo.18904342},
  note = {ORCID: 0009-0006-2011-3258}
}

Related Research