Use cases/Travel & experiences

Travel agency website

Explore how an approved content contract, parallel design and validation work, and early journey review could reduce rework on a travel website.

Cost · elapsed time · output quality
Trace the workWorking methods, tradeoffs and evidence

Different tools.
Shared project context.

Model, harness and platform choices used in the examples.

ClaudeClaude Code
OpenAICodex CLI
GeminiGemini CLI
AndroidApp modernization

Explore the collection

Where the work changes.
How to judge the difference.

Start with the process: cost, elapsed time and output quality.
Inspect the method, tradeoffs and evidence for each case.

Now exploring: Agree the journey before multiplying the pages.

Process study · proposed method

Agree the journey before multiplying the pages.

Explore how an approved content contract, parallel design and validation work, and early journey review could reduce rework on a travel website.

Establish the baseline

  • Agency brand and approved imagery
  • Destinations and sample itineraries
  • Enquiry fields and delivery destination

Evidence to retain

  • Approved journey and content contract
  • Reusable components and validation checks
  • Review findings and accepted revisions

Why work this way?

The process, and what it could improve.

Proposed method · no measured savings claim

Cost

Resolve the journey before multiplying pages.

Review one destination-to-enquiry slice with approved content. Reuse the accepted components and field definitions across the remaining pages.

What to compare
Design, model, asset and review costs per accepted journey, including discarded concepts and content corrections.
The tradeoff
Reusable components only save work after the underlying journey is agreed. Repeating an assumption multiplies rework.

Resolve the journey before multiplying pages.

Compare the same scope and acceptance criteria. Include setup, coordination, failed attempts and human review alongside the successful model call.

Inspect the agents, competing approaches, context and review gates.

Design the work

One problem. A coordinated agent team.

Proposed task plan

A model reasons. A harness runs its agent session and tool loop. Orchard coordinates the work, context and handoffs across them.

Orchestration phase 1: Split the work. Preserve the question.

Orchard orchestratorScopes work · delegates · synthesizes evidence
CONTROL

Implementation lead

01
Model familyClaudeHarnessClaude Code

Design destination and itinerary pages around the agency’s expertise and visual identity.

Bounded task assigned

Counterproposal agent

02
Model familyGPTHarnessCodex CLI

Propose a search/enquiry-led journey and identify where visual storytelling hides the next action.

Opposing assumption assigned

Independent reviewer

03
Model familyGeminiHarnessGemini CLI

Test the selected journey on mobile, with keyboard navigation and synthetic enquiry delivery.

Evidence criteria assigned
Shared project contract

Rules · skills · hooks · MCP configuration

Adapted to each harness
01 / 04 · Dispatch

Split the work. Preserve the question.

One objective, distinct responsibilities

Build a travel-agency website by testing an editorial design against a task-focused enquiry journey.

Approved destination content and enquiry scope. No invented inventory, prices, or live booking claims.
Implementation leadeditorial prototype + content model
Counterproposal agentalternative journey + usability tasks
Independent reviewerPlaywright results + accessibility findings
Illustrative execution planAssignment Evidence Review

Model families and pairings are illustrative. Assignments here are manual examples, not measured model comparisons or automatic cost optimization.

Trellis / shared knowledge

Brand and content

Approved imagery, agency specialties, itineraries and tone.

Design decisions

Competing prototypes, task observations and accepted component patterns.

BEAM / coordination

Bounded tasks, ownership notices, findings and handoffs between sessions. A queued message and a processed receipt are distinct states.

Recall → work → retain decisions

Tools / execution

Browser + design previewRepository + static buildForm / CRM sandbox

Access follows the configured harness, credentials and approved scope.

Inspect the example work orderJSON

An illustrative work-order format for this scenario. Changing an example session updates this document; no agents run from this page.

{
  "example": "travel-agency-website",
  "objective": "Build a travel-agency website by testing an editorial design against a task-focused enquiry journey.",
  "boundary": "Approved destination content and enquiry scope. No invented inventory, prices, or live booking claims.",
  "sessions": [
    {
      "role": "proposer",
      "model_family": "Claude",
      "harness": "Claude Code",
      "task": "Design destination and itinerary pages around the agency’s expertise and visual identity.",
      "expected_artifact": "editorial prototype + content model"
    },
    {
      "role": "challenger",
      "model_family": "GPT",
      "harness": "Codex CLI",
      "task": "Propose a search/enquiry-led journey and identify where visual storytelling hides the next action.",
      "expected_artifact": "alternative journey + usability tasks"
    },
    {
      "role": "reviewer",
      "model_family": "Gemini",
      "harness": "Gemini CLI",
      "task": "Test the selected journey on mobile, with keyboard navigation and synthetic enquiry delivery.",
      "expected_artifact": "Playwright results + accessibility findings"
    }
  ],
  "project_contract": [
    "Rules",
    "Skills",
    "Hooks",
    "Tool configuration"
  ],
  "shared_context": [
    "Brand and content",
    "Design decisions"
  ],
  "evidence_required": [
    {
      "label": "The next step is clear",
      "evidence": "Complete a destination-to-enquiry task on each proposed layout."
    },
    {
      "label": "The interface is usable",
      "evidence": "Check keyboard access, mobile overflow, labels and error states."
    },
    {
      "label": "The enquiry arrives correctly",
      "evidence": "Send a synthetic enquiry through the approved sandbox routing and inspect the record."
    }
  ],
  "release_owner_gate": "The agency owner accepts the content, journey and enquiry destination before the tested build is published."
}

The process principle

Reuse accepted decisions, not untested assumptions.

Scope of this exampleThis is an enquiry-site concept. Live availability, prices, payments, and bookings require approved providers and separate integrations.

A fictional agency concept. No live trips, prices, availability, or bookings are offered in this preview. Accessible forms guidance

Reuse accepted context.

Share source references, constraints and decisions so each worker can start from the same reviewed material. Verify it is still current.

Parallelize independent work.

Split bounded tasks only after their interfaces are clear. Include coordination and integration overhead when evaluating elapsed time.

Make review change the output.

Use a second perspective to challenge a specific risk. Preserve the finding, the correction and the evidence that the correction holds.

Your next chapter

Where does your
work repeat itself?

Bring a real task, a baseline and acceptance criteria. Evaluate the process through the work it produces.

Explore Orchard
Photography and technology credits

Photography illustrates each project context. Interface concepts use fictional brands and sample records. Technology names and marks identify the tools discussed.