your-ai-app/output
AI output
Generated work
Needs expert judgment
Deploy expert
Developer preview
·
MCP + REST
Configure the human layer inside your AI product
Use HumanDeploy to trigger expert outcomes, package context, route work to your team or our network, and return structured results inside your app.
<hd-action
surface="before_publish"
outcome="launch_check"
route="hybrid"
/>
humandeploy
Active
Context packed
Expert deployed
Review active
Memory saved
Expert-backed work returns inside your app.
01
Worth finishing
AI products are creating work worth finishing
Better AI means users ship more ambitious work — drafts, designs, plans, code, recommendations, brand assets, launch checks. The output is valuable. The completion is human.
Most products stop at “generate.” Users still need someone to design it, debug it, validate it, brand it, finish it, or ship it. HumanDeploy lets you ship the outcome.
01Is this brand-ready?
02Is this launch-ready?
03Can someone finish this?
04Can someone ship it with me?
your-ai-app
user hesitates
opens Slack
searches forums
DMs a contact
asks a teammate
From AI answer to expert-backed outcome
AI-generated
Human-reviewed
AI-drafted
Expert-improved
AI-scoped
Expert-delivered
AI-suggested
Accountable recommendation
Embedded human capability, on demand.
03
Lifecycle
How a human outcome runs
Six steps. Configurable per surface.
01
surface
<hd-action surface="before_publish" />
Triggered
Surface triggered
An hd-action element fires, an API call comes in, or an agent decides.
02
context-packet
goal
captured
files
3 packed
brand
loaded
Context packed
Goal, AI output, files, constraints, brand context — packaged.
03
outcome
type
launch_check
scope
confirmed
eta
~ 1h
Outcome scoped
Mapped to a typed, productized outcome with confirmed scope.
04
routing
policy
hybrid
route
hd_network
Matched
Work routed
Your team, our experts, or both — by rule.
05
returned-object
launch-check.md
Delivered
Result returned
Typed deliverable lands inside your product.
06
memory
decision saved
Compounded
Memory saved
Future requests inherit the decision.
04
Outcomes
Productized outcomes, not open-ended services
Each outcome is a typed contract — known scope, deliverable shape, expected timeline, routing logic.
Outcome
Design Pass
A senior designer review on AI-generated visuals or copy.
Outcome
Bug Fix
A diagnosed and resolved bug, with the fix returned.
Outcome
Brand Kit
On-brand assets, voice, and treatments based on your context.
Outcome
Launch Check
A ship-ready audit for a campaign, page, or product change.
Outcome
Finish This App
Take an AI-generated app to a deployable state.
Outcome
Partner Support
A defined customer-success touch handled by your team or ours.
05
Routing
Route to your team, our experts, or both
Every outcome can route differently. Send product support to your customer team. Send brand reviews to a HumanDeploy expert. Mix both with rules your product owns.
Partner-fulfilled
Your team handles it.
We package the context. Your team works it inside your tools — Slack, Linear, your inbox — with full state.
HumanDeploy-fulfilled
Our experts handle it.
Vetted experts in our network. You stay out of recruiting, vetting, scheduling, payments, and quality control.
Hybrid
Both, by rule.
Per outcome type. Per skill. Per overflow threshold. Rules live in your product.
routing.policy
Active
product_support → partner_team design_pass → humandeploy_network enterprise_onboarding → partner_success overflow → humandeploy_experts
06
Primitives
Built as product primitives, not a services workaround
HumanDeploy turns expert work into typed objects your product can trigger, display, route, and remember. Compose them. Render them. Store them in your data model.
tag<hd-action>
surfacebefore_publish
triggerlive
Object
Surface
A trigger point where outcomes are requested.
typelaunch_check
scope✓
eta~ 1h
Object
Outcome
A typed, productized request.
brief.md
brand.md
prior.md
Object
Context Packet
Brief, files, brand memory, prior decisions.
design_pass→network
support→partner
overflow→network
Rule
Routing Policy
Who fulfills — your team, ours, or both.
brief
skillscopybrand
urgencypriority
Object
Expert Request
The atomic unit routed to a human.
scoped
routed
in_progress
returned
State
Fulfillment State
The lifecycle of every request.
typelaunch check
scopeready
eta~ 1h
Object
Price Card
Outcome cost + timeline. Surfaced before commit.
launch-check.md
Object
Returned Object
The typed deliverable that lands in your product.
approve
revise
State
Approval Status
Accepted, revised, or escalated.
decision saved
2026·04·28
Memory
Memory Event
Saved decision improving future requests.
07
What you don’t build
Don’t build the expert operation
Embedding human outcomes sounds simple until you have to recruit, vet, match, scope, price, route, run quality control, manage payments, transfer context, and handle every revision.
That is an operations company. It is not the product you set out to build. HumanDeploy runs the operations layer — and can route requests to your own team with full context when that is the right fit.
You ship the integration. We ship the operation.
07
Integration options
Start simple. Go native when it matters
Drop in a widget today. Graduate to native primitives when the integration earns it.
01 · Drop-in
Widget
One HTML element. No backend code.
<hd-button />
02 · API
REST API
Programmatic deployments from your agent.
POST /v1/deployments
03 · Tool
Agent Tool / MCP
Plug into Claude, Cursor, Windsurf, or your own agent.
deploy_expert
04 · Native
Native workflow
Branded escalation in your product.
hd.requests.create()
09
Use cases
Use it wherever AI reaches the edge of trust
your-design-app/canvas
AI design tools
Have a senior designer review this before I send it to the client.
your-marketing-app/campaign
AI marketing tools
Turn this campaign draft into a launch-ready campaign.
your-sales-app/sequence
AI sales tools
Have a GTM expert improve this outbound sequence.
your-builder/landing-page
AI website builders
Get a conversion expert to review this landing page.
your-strategy-app/recommendation
AI strategy tools
Validate this recommendation before the team acts on it.
your-workflow/task-graph
AI workflow tools
Escalate complex tasks to a vetted human expert.
10
Commercial modes
Bill it the way your product bills
Outcomes can be priced any way that fits your model. Mix per outcome, per surface, per plan.
Mode
User pays
End user pays per outcome at the point of escalation. Price card surfaced before commit.
Mode
Included credits
Outcomes are bundled into the user’s plan. They draw down a balance the product manages.
Mode
Partner billed
You pay HumanDeploy at wholesale. Pass through, mark up, or absorb.
Mode
Internal routing
Free to the user. Fulfilled by your own team. We provide the operating layer.
11
Context
The human doesn’t arrive cold.
Every request can carry the user’s task, the AI output, source material, files, brand context, prior decisions, constraints, and the desired outcome. The expert inherits the state of the work before they ever touch it.
humandeploy/context-packet
Inherited
brief
"Review the Q-launch hero against brand voice…"
files
draft.md
brand.md
audience.md
brand memory
positioning
tone
prior decisions
3 events recalled
constraints
no exclamations · no SaaS jargon
goal
ready to ship
handed to
Brand Strategy
Travels with the request. Inherited by the expert.
Context travels with the work
Experts inherit the task state
Every expert decision becomes memory
The product gets smarter each deployment
Build with HumanDeploy
Give your AI product a human outcome layer.
Your product shouldn’t end at “generate.” Configure surfaces, productize outcomes, route to your team or ours, and return the finished work inside your app.