Hire Claudekit
Our one product is the Custom AI Agent for your business, priced by how many features it has. Our AI agents build it and run automated tests, with no human review step, and you get the code in a private GitHub repo you own.
A feature is one concrete job the agent does. For example, it reads new orders from Shopify, answers customer emails from a template, writes a sales report every morning, posts to Facebook or looks up data in Google Sheets. A feature includes the connection it needs: "reads new orders from Shopify" is one feature, and connecting to Shopify is part of it. Count your features, and the count sets the tier and the price.
The process is fully automated. AI agents build the agent and test it with automated tests; no person reviews the code before it reaches you. That is how we keep prices fixed and low. You check the delivery, and the 2 revision or addition requests cover anything that doesn't match your list. You can see our agents at work and browse sample kits.
How it works
-
Step 1
Send your feature list
List the jobs you want the agent to do, one per line, in the form below. Each job includes connecting to the system it needs.
-
Step 2
The count sets the tier and price
Your number of features picks the tier, and the tier fixes the price. We email the tier, the price and an invoice. No scoping call.
-
Step 3
Agents build and test
AI agents write the code and run automated tests for each feature on your list. There is no human review step.
-
Step 4
Private repo delivery
You get a private GitHub repo with the code, the tests and setup instructions to run it on your own infrastructure.
-
Step 5
2 revision requests
After delivery, every tier includes 2 revision or addition requests, to fix a feature or adjust it.
Delivery targets count from when we have the first payment, your feature list with examples and the test accounts we need.
Pricing by feature count
Prices in USD, checked October 25, 2026. The founding-client price applies to our first 10 clients, in exchange for a case study on claudekit.ai (it can be anonymous), a short testimonial and a 20-minute feedback call.
-
1 to 3 features
Starter
A small agent that does one to three jobs well.
$490 Founding-client price (first 10 clients)
List price $790
- Delivery
- About 5 business days
- Payment
- 100% upfront
- Revisions
- 2 requests
- Warranty
- 14 days
- Extra feature
- $400 ($250 founding)
-
4 to 7 features
Business
An agent that runs a whole workflow across a few of your tools.
$1,190 Founding-client price (first 10 clients)
List price $1,900
- Delivery
- About 7 to 10 business days
- Payment
- 50% to start, 50% on delivery
- Revisions
- 2 requests
- Warranty
- 30 days
- Extra feature
- $350 ($220 founding)
-
8 to 12 features
Pro
A larger agent that ties several systems and workflows together.
$2,900 Founding-client price (first 10 clients)
List price $4,500
- Delivery
- About 2 to 3 weeks
- Payment
- 50% to start, 50% on delivery
- Revisions
- 2 requests
- Warranty
- 30 days
- Extra feature
- $450 ($290 founding)
-
13+ features
Custom
For 13 or more features. Send your feature list and we reply with a fixed quote before any payment.
Quote
Included in every tier
- Every feature on your list, built by AI agents
- Automated tests for each feature
- Logging, error handling and retries
- Settings and keys in environment variables, never in the code
- A private GitHub repo you own, with setup instructions
- 2 revision or addition requests after delivery
Not included
- Hosting or running the agent for you
- API and hosting costs while it runs
- Dashboards or a custom UI, unless you list them as features
- Ongoing maintenance after the warranty
- Live training sessions
You provide
- Your feature list, one job per line
- An example of the input and the result you expect for each feature
- A sandbox or test account for each outside system
- Your own Anthropic API key, if the agent calls Claude
Need more features than your tier allows? Upgrade to the next tier, or add single features at the extra-feature price above. That price is the tier's price divided by the middle of its feature range, rounded to the nearest $10. Add-ons never cost more than the next tier. If your tier price plus add-ons would exceed the next tier's price, you simply pay the next tier's price instead.
Examples of agents we build
These are examples, not separate packages. Every agent is priced the same way: count the features.
-
Agent for your dev team (Claude Code kit)
A Claude Code plugin tuned to your stack: a CLAUDE.md, Skills, subagents and hooks, delivered as a private plugin marketplace.
Example feature list
- Reviews code against your style guide
- Writes release notes from your git log
- Blocks commits when the tests fail
3 features: Starter, $490 founding or $790 list.
-
Connect Claude to your own software (MCP)
An MCP server that lets Claude read from your internal API or database, read-only by default.
Example feature list
- Looks up an order by number
- Lists today's unshipped orders
- Finds a customer by email
3 features: Starter, $490 founding or $790 list.
-
Business automation agent
An agent that runs a business workflow on a schedule, on your own infrastructure.
Example feature list
- Reads new orders from Shopify
- Looks up stock in Google Sheets
- Answers customer emails from a template
- Writes a sales report every morning
- Posts new products to Facebook
5 features: Business, $1,190 founding or $1,900 list.
Policies
Payment
- Starter: 100% upfront.
- Business: 50% to start, 50% on delivery.
- Pro: 50% to start, 50% on delivery.
- We send an invoice for the tier your feature count sets. This site never takes payment.
- Extra features are an upgrade or a per-feature add-on, and only with your approval.
- Add-ons never cost more than the next tier. If your tier price plus add-ons would exceed the next tier's price, you simply pay the next tier's price instead.
Revisions, acceptance and warranty
- Every tier includes 2 revision or addition requests after delivery.
- A request can fix a feature, change how it behaves, or add features up to your tier's limit.
- You have 5 business days to check the delivery. If we don't hear back, it counts as accepted.
- Starter: 14-day warranty on bugs against your feature list.
- Business: 30-day warranty on bugs against your feature list.
- Pro: 30-day warranty on bugs against your feature list.
Refunds
- Cancel before the agents start building: 100% refund.
- Cancel during the build: 50% of what you paid is refunded, and the code is not delivered.
- A feature on your list still doesn't work after both revision requests: 100% refund.
- No refund for a change of mind after you accept the delivery.
Ownership and handover
- You own the delivered code.
- We transfer the private GitHub repo to you, or give you admin access, after full payment.
- You pay the API and hosting costs of running what we deliver.
Runs on your own Anthropic API key. Agents we deliver call Claude with your API key, on your infrastructure. We don't resell or pay for Claude usage on your behalf.
What we don't do
- Regulated data, such as health records (HIPAA) or card data (PCI)
- Jobs that need production credentials during the build
- Writes to production systems that hold sensitive data combined with custom auth (SSO or multi-tenant)
- Anything that breaks Anthropic's usage policy or another service's terms, such as unauthorized scraping or spam
- Deadlines shorter than the tier's delivery target
- Hourly work: we sell agents, not hours
Request an agent
No payment now. We reply with the tier, the price and an invoice. See our privacy policy for how we handle what you send.
More than 12 features? Send the list anyway and we reply with a custom quote.
Frequently asked questions
What counts as a feature?
One concrete job the agent does, such as reading new orders from Shopify, answering customer emails from a template or writing a sales report every morning. A feature includes the connection it needs, so reading orders from Shopify is one feature, Shopify connection included. Your count sets the tier and the price.
What if I need more features later?
Upgrade to the next tier, or add single features at the per-feature add-on price shown for your tier. The add-on price is the tier's price divided by the middle of its feature range, rounded to the nearest $10. If add-ons would make your total exceed the next tier's price, you simply pay the next tier's price. Past 12 features, we quote.
What do the 2 revision requests cover?
Every tier includes 2 requests after delivery. Use them to fix a feature that doesn't do what your list said, to change how an existing feature behaves, or to add features up to your tier's limit. Features beyond the limit are an add-on or an upgrade.
Whose API key does the agent use?
Yours. The agent runs on your own Anthropic API key and your own infrastructure, and you pay the API and hosting costs of running it. We never resell Claude usage.
Who owns the code?
You do. We transfer the private GitHub repository to you, or give you admin access, once the job is paid in full.