AUX Examples: Structured Agent Flows
A set of real-world AUX examples — covering app developers, product teams, and business teams. Each example follows a consistent structure designed for fast comprehension: set the stage, identify the problem and opportunity, define the AUX system, map actors and context, design the flow with AX patterns, model trust, and measure outcomes.
Sales Research Agent
Persona: SDR (Sales Development Rep)
Job to be done: "Prepare high-quality outreach for a new lead in under 5 minutes."
Problem: Manual research takes 20–40 minutes, personalisation is low, and SDRs repeat the same work for every lead.
AUX System: The agent gathers lead intelligence, surfaces company data, and drafts tailored outreach. The SDR reviews, edits, and sends. The Intent Handshake is the critical AX pattern — the agent surfaces "here's what I found and how I'll position outreach" before proceeding.
Trust Design: Sending is human-only, messaging requires confirmation, research is autonomous. Progressive autonomy — trust is earned step by step.
Outcome: 5x faster prep, higher personalisation, and trust that increases over time.
Customer Support Agent
Persona: Support agent + end user
Job to be done: "Resolve user issues quickly without escalation."
AUX System: The agent handles the full ticket lifecycle — drafting responses, suggesting solutions from the knowledge base, and routing edge cases to human agents. The Intent Handshake surfaces: "Here's what I think the issue is + solution."
Trust Design: Drafting is autonomous, sending requires confirmation, escalation is an agent decision.
Outcome: 70% of tickets auto-handled, faster resolution, human agents focus on edge cases only.
Mobile Fitness Coach
Persona: Busy professional
Job to be done: "Stay consistent with workouts without planning everything."
AUX System: The fitness agent plans personalised weekly workouts, adjusts daily based on real-time context (calendar, energy levels), and nudges the user with motivational prompts. The Intent Handshake surfaces: "Here's your plan based on your schedule."
Trust Design: Planning requires confirmation, adjustments are semi-autonomous, changes to goals require confirmation.
Outcome: Personalised experience, higher retention, feels like a coach rather than an app.
API Integration Agent
Persona: Backend engineer
Job to be done: "Integrate external API quickly and correctly."
AUX System: The agent reads API docs, suggests integration architecture, and generates production-ready code. The Intent Handshake surfaces: "Here's how I'll implement it" — ensuring alignment on architecture before any code is written.
Trust Design: Code generation requires confirmation, architecture requires confirmation, refactoring is supervised.
Outcome: Faster dev cycles, fewer errors, knowledge that scales across all integrations.
GTM Pipeline Monitor
Persona: Head of Sales
Job to be done: "Know which deals are at risk and act early."
AUX System: The agent monitors the pipeline continuously in the background, surfacing risk signals and recommending next steps before deals slip. A key AX pattern is the Invisible Pattern — the agent operates silently, only surfacing when it detects something actionable.
Trust Design: Monitoring is autonomous, recommendations are visible, actions require confirmation.
Outcome: Better forecasting, early intervention on at-risk deals, less manual reporting.
Internal Ops Automation
Persona: Finance ops
Job to be done: "Reconcile invoices quickly and accurately."
AUX System: The agent handles bulk reconciliation — matching invoices automatically and flagging inconsistencies for human review. It handles the high-volume, low-risk work autonomously and routes only genuinely complex cases to the human operator.
Trust Design: Safe matches are autonomous, edge cases require confirmation.
Outcome: 80% automation, fewer errors, consistent processing.
Product Onboarding Agent
Persona: New user
Job to be done: "Get value from product fast."
AUX System: The agent acts as a personalised guide, adapting the onboarding journey based on what the user actually needs. The Intent Handshake aligns on the user's goal before building a personalised path.
Outcome: Faster activation, every user gets a journey tailored to their specific goal.
Content Creation Agent
Persona: Content marketer
Job to be done: "Produce high-quality content consistently."
AUX System: The agent handles the full production pipeline — from research to structured drafts to refined content. The handshake happens at the outline stage — the marketer aligns on structure and angle before the agent invests in a full draft.
Outcome: Faster content production, consistent quality, brand voice maintained.
These 8 AUX examples demonstrate how intent handshakes, progressive autonomy, and trust design combine to create agent experiences that feel natural, safe, and genuinely useful — across every team and use case.