Why teams use it
Move one session into the next workflow
It keeps the current tool state useful after the calculation by shaping it for docs, agents, tickets, implementation planning, and future connected AI systems.
Please rotate your phone.
This experience is designed for portrait mode.
AI Product Tools Platform Layer
AI Handoff
AI Handoff turns a live AI Product Tool session into Prompt, Agent, JSON, Markdown, and MCP-ready exports. It helps teams move research, planning, and evaluation work out of the browser and into real delivery workflows without rebuilding the same context by hand.
Built from live state Prompt + Agent JSON + Markdown MCP-ready
Tool Output to Delivery Path
One live session can be shaped for fast prompting, readable docs, exact transport, or future connected workflows.
01
Step 1
Live Tool State
Current inputs, outputs, scenario, and mode from the active tool session.
02
Step 2
AI Handoff
The shared layer that reshapes the live session into usable export formats.
03
Step 3
Choose Export Path
Shape the same session for prompting, documentation, exact transport, or protocol-ready use.
04
Step 4
Use in Delivery
Move the output into the next downstream workflow with the right level of structure.
Export paths
Prompt / Agent
Markdown
JSON
MCP-ready
Delivery targets
Docs
Agents
Dev Workflow
Connected Systems
Why it matters
PDF and screenshots stop too early
Static exports can show a result, but they are weak formats for prompting, automation, implementation, or exact structured reuse. AI Handoff keeps the live state usable after the tool itself is done.
Who it helps
One handoff layer, different downstream needs
- UX leads: stronger framing for strategy, prioritization, and delivery conversations.
- PMs: planning-ready outputs that keep the current tool context intact.
- Developers: exact payloads, readable briefs, and better agent wrappers.
- Stakeholders: summaries that are easier to review than raw tool state.
Overview
What AI Handoff is, and why it exists
AI Handoff is the shared export layer across the AI Product Tools suite. It keeps the current session usable when the next step is documentation, planning, implementation, or AI-assisted follow-through.
What it is
A translation layer for live UX tool state
It reads the current inputs, outputs, scenario, and mode in the active tool, then reshapes that state into formats that are easier to act on outside the browser.
What changes
The result becomes usable after the calculation
The value is not just the number or recommendation on-screen. Teams still need to explain it, share it, document it, or hand it into the next workflow.
Why PDF alone is not enough
A PDF can show the result, but it does not preserve stable structure, frame work for an agent, or make the exact current session easy to reuse in docs, tickets, prompts, or connected workflows.
Step 01
Complete a live tool
Use an AI Product Tool to model a decision, define a workflow, score risk, or estimate impact from the current session.
Step 02
Read the live state
AI Handoff reads the current inputs, outputs, scenario, mode, and the context already visible in the tool.
Step 03
Shape it for use
That same state is reshaped into prompt, agent, JSON, Markdown, and MCP-ready exports.
Step 04
Move work forward
A human can review, copy, or download the export and move it into planning, documentation, coding, or follow-up analysis.
Step 05
Connect future workflows
As the shared handoff layer matures, the same structure can support more direct flows into connected AI systems and delivery tooling.
Prompt Export
Best for fast copy/paste into AI tools
Prompt export packages the current tool state into a ready-to-run instruction set for Claude, ChatGPT, Codex, Cursor, or any other prompt-driven AI client.
What it does
Turns the session into a ready prompt
It pulls the most relevant inputs and outputs into a clear structure with objectives, constraints, and trust markers so the first prompt starts from stronger context.
When to use it
Use it when speed matters most
It is the fastest route to quick synthesis, idea generation, planning briefs, or a first pass on a deliverable without setting up an integration.
Compared with JSON
Readable first, exact second
Prompt export is optimized for human copy/paste. JSON is still the better choice when stable keys and machine-readable structure matter more than speed.
Compared with Markdown
More directive, less doc-shaped
Markdown is better when the output needs to live in docs. Prompt export is stronger when the next move is telling an AI model what to produce.
Compared with MCP
Immediate use, not protocol structure
Prompt export is for immediate use. MCP-ready payloads are about future connection patterns and structured tool or client relationships.
Example
Design System ROI Calculator → executive business case prompt
Use the live ROI model to draft an executive business case, board summary, or investment memo from the exact current numbers.
Example
AI UX Methods Finder → interview guide prompt
Turn a method recommendation into a ready-to-run interview guide, research plan, or usability test plan without rebuilding the context manually.
Example
Design Debt Tracker → backlog generation prompt
Push prioritization logic into a prompt that can generate backlog-ready debt items, tiers, and action framing for the team.
Agent Export
Prompt plus a stronger wrapper for build agents
Agent export uses the same core content as Prompt, but wraps it more explicitly for systems that benefit from clearer task framing, rules, and execution context.
What changes
More explicit framing than a plain prompt
Agent wrappers make the task, constraints, and relationship to the current tool state more explicit. That helps when the next step is implementation, breakdown, or structured follow-through.
When to use it
Use it when an agent needs clearer operating instructions
If a coding or build agent will act on the output, the wrapper often performs better than a plain prompt because it establishes scope, source of truth, and behavioral rules more clearly.
Agent target
Claude
Strong for synthesis, planning, stakeholder framing, research interpretation, and longer-form reasoning.
Agent target
Codex
Best when the next step is implementation work, task breakdown, coding support, or build-ready direction.
Agent target
Cursor
Useful when the output needs to be applied directly inside a development workspace with surrounding code context.
Agent target
Generic Agent
A neutral wrapper for systems that do not need provider-specific framing but still benefit from explicit task context.
Example
AI UX Process Builder → Jira or Linear breakdown for Codex or Cursor
Use the workflow state, roles, steps, and outputs to generate implementation tickets and delivery sequencing for a coding-focused agent.
Example
Component Complexity Estimator → component PRD for a coding agent
Wrap scope, effort, and acceptance expectations so an implementation-oriented agent can turn the estimate into a build-ready artifact.
JSON Export
The stable transport layer for exact structured handoff
JSON export preserves the current tool state as machine-readable structure. It is the clearest option when exactness, integration, debugging, or future automation matters more than narrative formatting.
Why it matters
Exactness beats screenshots
Screenshots are visually helpful, but they do not preserve exact keys, values, modes, and outputs. JSON keeps the state transportable and inspectable.
When to use it
Use it for integrations, automation, and debugging
JSON is best when another system, script, or developer needs the exact payload rather than a prose summary. It is also helpful when you want to compare or inspect the state directly.
Practical use
Saving state
Preserve the precise live outputs from a tool so another teammate or system can reuse the same current model later.
Practical use
Sending exact outputs to another system
Move the current tool state into another workflow, integration, or script without manually re-entering values.
Practical use
Using JSON as the stable transport layer
Even when a team later converts the output into Markdown or prompts, JSON remains the strongest exact representation underneath.
Markdown Export
Readable handoff for docs, tickets, and async collaboration
Markdown export gives the current session a clean, readable format that works well in GitHub, Notion, Jira, Confluence, specs, planning docs, and many AI clients.
Why teams use it
Readable without losing structure
Markdown keeps the handoff easy to scan for humans while still remaining structured enough for many AI systems to interpret well.
When it wins over JSON
Use it when the audience needs to read first
If the output needs to live in documentation, planning notes, or collaborative tools, Markdown is often a better default than JSON because it is easier to absorb and discuss.
Example
Accessibility Audit Scorecard → remediation brief
Turn current issues, risk framing, and remediation logic into a readable brief that can live in Jira, GitHub, or Confluence.
Example
Design System ROI Calculator → planning memo
Give leadership and cross-functional partners a cleaner planning artifact than a raw payload when the conversation is about investment and timing.
Example
AI UX Process Builder → team playbook summary
Use the generated workflow and outputs to create a readable playbook summary that onboarding, governance, and delivery teams can use.
MCP-Ready
MCP in practical terms
MCP is a standard way for AI clients to connect to tools, prompts, and resources. In practice, it is part of how future AI systems can discover what a tool is, what data it expects, and what context it can use.
What MCP means here
Protocol-shaped output, available today
AI Handoff can already produce MCP-style payloads that users can inspect, copy, and download. The shape is ready now even though the connection is still mostly manual.
Why it matters
It points toward direct tool connectivity
Today the user can review MCP-style payloads manually. Later, the same handoff layer can support more direct connections into MCP-capable AI systems and clients.
1. AI Product Tool
Live state
The tool contains the current inputs, calculations, selected mode, scenario, and outputs.
2. Worker / handoff layer
Shared export formatting
The Cloudflare Worker and shared panel layer shape the current state into a consistent handoff structure.
3. MCP-ready payload
Schema + context
The export can describe tool definition, schema, and current session context in a protocol-friendly form.
4. AI client or coding agent
Connected use later
Future MCP-capable systems can use that structure more directly instead of relying only on pasted text.
What “MCP-ready” means today vs later
Today, MCP-ready means the tools can already expose protocol-shaped payloads for inspection and download. Later, the same approach can support stronger direct connections into AI clients, coding agents, and shared tool ecosystems. The Cloudflare Worker layer matters because it gives the site a shared place to shape that output beyond the browser UI.
Use Cases
How each live tool moves work forward
The export formats stay consistent, but the practical value changes by tool. These examples show how the same handoff layer supports different jobs across strategy, planning, assessment, and delivery.
Active live tool
Design System ROI Calculator
Who uses it: Design leaders, PMs, engineering leaders, executives
How it moves work forward: Turns a business-case model into concrete leadership artifacts and downstream planning inputs.
- Executive business case and board-ready summary prompts
- Investment memo and product-engineering alignment framing
- JSON for exact ROI transport and Markdown for planning docs
Active live tool
AI UX Process Builder
Who uses it: UX leads, PMs, design ops, cross-functional leaders
How it moves work forward: Takes workflow structure and roles from the tool into implementation, governance, and execution planning.
- Team operating model and workflow implementation prompts
- Jira/Linear breakdowns for Codex or Cursor-style agent workflows
- Markdown playbook summaries for onboarding and team docs
Active live tool
AI UX Methods Finder
Who uses it: Researchers, PMs, founders, product teams
How it moves work forward: Converts a method recommendation into ready-to-run research and planning outputs.
- Research plan, interview guide, and usability test prompts
- Readable Markdown briefs for alignment with stakeholders
- JSON for preserving the selected method and decision context
Active live tool
Accessibility Audit Scorecard
Who uses it: Accessibility leads, QA, product, engineering, leadership
How it moves work forward: Carries accessibility posture into remediation planning, delivery, and risk communication.
- Remediation plan, QA checklist, and leadership risk summary prompts
- Markdown remediation briefs for tickets and docs
- JSON and MCP-style payloads for structured follow-up workflows
Active live tool
Design Debt Tracker
Who uses it: Design leads, PMs, engineering managers, operational stakeholders
How it moves work forward: Pushes debt analysis into prioritization, backlog creation, and executive framing.
- Prioritization roadmap, backlog generation, and executive justification prompts
- Markdown handoff for planning, reviews, and budget conversations
- JSON for stable transport of the exact debt model and totals
Active live tool
Component Complexity Estimator
Who uses it: Design systems teams, engineering leads, PMs, architects
How it moves work forward: Transforms scope estimates into delivery artifacts that teams can act on immediately.
- Implementation briefs, PRDs, and acceptance-criteria prompts
- Agent wrappers for coding and implementation planning
- Markdown summaries for specs and structured JSON for integrations
Best Practices
Use the right format, keep a human in the loop
AI Handoff makes UX outputs more usable, but it does not replace judgment. The strongest workflow is still structured export, human review, and the right format for the task.
Human review still matters
AI Handoff is a translation layer, not an approval layer. Review outputs before implementation, executive sharing, or governance use.
Use the right format for the job
Prompt is fastest, Agent is stronger for build systems, JSON is the stable transport layer, Markdown is best for readable docs, and MCP is for future-connected workflows.
Treat the live tool state as the source of truth
The export reflects the current inputs and outputs in the tool. If the model changes, regenerate the handoff instead of relying on stale copies.
Be thoughtful with sensitive data
Do not paste confidential, regulated, or personal information into agents or systems that are not approved for it. Review payloads before sharing.
When should I use which format?
Use Prompt for speed, Agent for implementation work, JSON for exact structure, Markdown for readable docs, and MCP for future-connected workflows.
Fastest copy/paste
Prompt
Use when you want a human-readable first pass in Claude, ChatGPT, Codex, Cursor, or another AI client right away.
Coding / build workflows
Agent
Use when the same prompt needs a stronger wrapper for implementation, planning, task breakdown, or agent-specific execution context.
Exact structure
JSON
Use when exact keys, values, and stable machine-readable transport matter more than narrative readability.
Readable handoff
Markdown
Use when the output needs to live in GitHub, Notion, Jira, Confluence, specs, meeting notes, or async documentation.
Future connected workflows
MCP
Use when you want to inspect or preserve protocol-ready structure for connected AI clients, tools, and resource workflows.
AI Handoff is designed to carry current tool state into the next step more cleanly. It does not replace strategic judgment, accessibility review, financial validation, delivery planning, or governance. Use the exports as strong working inputs, then review them in the real context where decisions are made.