CatDoes
Also known as: Cat Does, catdoes.com, DogDoes
CatDoes is an autonomous cloud agent with its own computer that plans, writes and tests React Native Expo mobile apps and React websites from a description, provisions a backend with database, sign in, storage and edge functions, and prepares App Store, Google Play and web releases. It also rebuilds an existing website as a native app, works on imported GitHub codebases and fixes errors caught by built in crash monitoring.
CatDoes, from CatDoes Inc., sells an AI agent that builds mobile apps and websites to business owners, designers, founders and engineers who want a working product without building it by hand. Mobile apps use React Native with Expo, websites use React with Vite and Tailwind, and imported projects can also be Next.js, TanStack Start or Create React App.
The agent runs in the cloud on its own computer, so it plans, writes code, installs packages, runs tests and fixes its own errors while the builder is away, and it carries a release through web deployment, Android builds and App Store submission in the chat.
CatDoes Cloud supplies the database, sign in by email, phone, Google or Apple, storage and edge functions, with the agent asking before risky backend changes, and CatDoes Watch captures runtime errors from web, iOS and Android with stack traces so they can be sent to the agent to fix.
Every message saves a checkpoint to rewind to, a plan can be agreed before any code is written, an existing site can be turned into a native app from its address, and GitHub repositories can be imported or kept in sync. Four agent tiers, Low to Max, trade reasoning depth against credits.
The Free plan gives 15 credits and one public web project. Core costs $20 a month with 50 credits, Starter $50 with 200 credits and iOS and Android releases, Plus $100 with 400 credits and GitHub sync, Pro $200 with 900 credits and code export, and Max $399 with 2,000 credits; yearly billing saves 16%. Enterprise, called DogDoes, adds private deployment options, unlimited projects, a dedicated support team and custom SLAs.
CatDoes fits non technical builders and small teams who want an agent to take a mobile app all the way to the stores, with a backend and error monitoring included. It fits less well for teams that need a choice of model, an API, single sign on and security attestations, or integrations beyond what can be wired through keys and edge functions.
Vendor details
Canonical URL
https://catdoes.com
Category
Coding agent
Subcategory
Autonomous cloud agent that builds React Native Expo mobile apps and React websites with a managed backend, error monitoring and store releases
Funding status
Privately held.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
CatDoes Cloud for database, authentication (email, phone through Twilio, Google, Apple), storage and edge functions; GitHub import, push and two way sync; Expo and Apple Developer accounts for builds and App Store submission; encrypted environment variables for keys such as Stripe or OpenAI; CatDoes Watch for error monitoring across web, iOS and Android.
In practice
A cafe owner describes a booking app; CatDoes plans it, builds the React Native app with sign in and a bookings table on CatDoes Cloud, and submits it to the App Store after she enters her Apple Developer details in a secure field.
A founder pastes the address of a Lovable site, CatDoes studies its pages and builds a matching native app, and later sends a crash from CatDoes Watch to the agent, which fixes it and ships a new build.
An engineer imports a Next.js repository, asks the agent to add a feature in debug mode, and pushes the result back to a new GitHub branch.
Sources & related URLs
Related / legacy domains
Research sources
Agentic Index coverage score
7.5 / 14 capabilities · 54%
| Integrations & Tool Calling | Partial |
|---|---|
|
When an app needs a backend, the agent checks with the builder and then provisions CatDoes Cloud with a database, email, phone, Google and Apple sign in, storage and edge functions for custom logic, webhooks and third party calls. Keys for services such as Stripe or OpenAI go into encrypted environment variables that the agent asks for by name, phone sign in runs through Twilio, the agent pushes commits to a linked GitHub branch and handles pulls and merge conflicts on Plus and above, and Expo and Apple Developer credentials drive mobile builds. There is no catalog of ready connectors and no MCP support. Sourcedocs.catdoes.com/catdoes-cloudread 2026-10-18 |
|
| Workflow Orchestration | Full |
|
An autonomous agent runs in the cloud on its own computer, plans the work, writes the code, installs packages, runs the tests and fixes its own errors, so the builder can close the tab and return when it is done. Bigger tasks get a planning step on their own, backend functions are checked before they deploy and verified afterward, and the agent carries a release through web deployment, Android builds or App Store submission in the chat. One project can hold a website and a mobile app side by side, each in its own folder. Sourcedocs.catdoes.com/features/plan-moderead 2026-10-18 |
|
| Knowledge Grounding & RAG | Partial |
|
An existing GitHub repository, including one exported from Lovable, Bolt or Base44, can be imported with its git history so the agent works from the real codebase, and typing @ points the agent at any file or folder. Marked up screenshots and pasted error messages add context for a fix. There is no knowledge base, document upload library or retrieval over a team's own material beyond the project's code. Sourcedocs.catdoes.com/features/import-from-githubread 2026-10-18 |
|
| Human Oversight & Guardrails | Full |
|
Asked to plan first, the agent shares the plan and waits for a go ahead before writing code. Before a risky backend change, such as changing data in the database, deleting a function or disconnecting the Cloud, it asks for Allow or Deny, it asks before a GitHub pull would overwrite work, and it never removes or relaxes an API key without asking. Every message saves a checkpoint the builder can rewind to, saved versions can be restored from History, and the live site only changes when the builder asks for another deploy. Sourcedocs.catdoes.com/features/checkpointsread 2026-10-18 |
|
| Security, Identity & Governance | Partial |
|
Every environment variable is encrypted before it is stored, Expo tokens, Apple passwords and verification codes go into secure fields that are encrypted end to end and never visible to the AI, and Watch redacts emails, passwords, API keys and card numbers before storing errors. Private projects visible only to the builder's team need a paid plan, and Enterprise offers private deployment and custom SLAs. Single sign on, roles and security attestations are not offered. Sourcedocs.catdoes.com/features/environment-variablesread 2026-10-18 |
|
| Observability & Auditability | Partial |
|
CatDoes Watch captures runtime errors from web, iOS and Android apps with stack traces, file locations and device details, groups them into issues with event counts, and filters by app, status and development or production. The Cloud tab shows the app's database, users and backend, and History keeps every saved version. The agent's own steps show in the chat as it works, but there is no exportable record of its commands and no feed to outside monitoring. Sourcedocs.catdoes.com/features/watchread 2026-10-18 |
|
| Memory & State Persistence | Partial |
|
Compact and save stores the current chat as a version in History and starts a fresh conversation while the project and codebase carry over, and long chats are best compacted because the agent can lose track of earlier context. Projects keep their code, backend and versions. No memory of preferences or decisions follows the builder from one chat or project to another apart from the code itself. Sourcedocs.catdoes.com/features/checkpointsread 2026-10-18 |
|
| Deployment & Data Residency | Partial |
|
Websites deploy to a catdoes.app address or a custom domain, Android builds come back as files to publish on Google Play, iOS builds are submitted to the App Store, and APK, AAB and IPA exports are available from Starter up. Apps are standard React Native Expo or React with Vite and Tailwind projects, downloadable from Plus up. The service runs on CatDoes's cloud, Enterprise lists private deployment options without further detail, and no region choice is offered. Sourcedocs.catdoes.com/features/deploymentread 2026-10-18 |
|
| Prebuilt Agents / Templates / Packs | Partial |
|
Showcase apps such as a furniture and decor shop, a matcha shop and a short video feed show what a finished build looks like, and guided paths turn an existing Lovable or Base44 site into a native iOS and Android app. Prompt guides cover planning, debugging and saving credits. There is no library of templates or packaged agents to start a project from. Sourcecatdoes.comread 2026-10-18 |
|
| Triggers & Channel Coverage | Partial |
|
Work starts from a message in the CatDoes dashboard on the web or on a phone, where files can also be browsed and edited, and an error captured by CatDoes Watch can be sent to the agent with one click. The agent keeps going after the tab closes and support replies arrive by email. Nothing starts the agent on a schedule, from a chat channel or from an outside event; each task begins with a message from the builder. Sourcecatdoes.comread 2026-10-18 |
|
| Model Flexibility & Routing | Not documented |
|
The builder picks one of four agent tiers, Low, Medium, High and Max, that trade reasoning depth against credits per step, with Medium as the default for new projects and High or Max suggested for planning before dropping down to build. The tiers are not tied to named models or providers, and there is no provider choice or bring your own key. High needs a balance above 100 credits and Max above 230, and the tier drops on its own when the balance falls below that. Sourcedocs.catdoes.com/features/saving-creditsread 2026-10-18 |
|
| APIs / SDKs / MCP Extensibility | Not documented |
|
No API, SDK, command line tool or MCP server lets an outside tool create projects, prompt the agent or deploy an app, so every change goes through the CatDoes dashboard. Code moves out through two way GitHub sync on Pro, pushes to a linked branch on Plus and above, or a full download, environment variables can be exported from Core up, and an imported or exported project stays a standard Expo or React codebase other tools can work on. Sourcedocs.catdoes.comread 2026-10-18 |
|
| Testing, Debugging & Optimization | Full |
|
The agent runs tests and fixes its own errors as it builds, and debug mode has it place console logs through the relevant code, wait for the builder to reproduce the bug and then read the real output to find and fix the cause. Errors from development and from live users land in Watch and go to the agent with one click, backend functions are verified after deploying, and a live preview at phone, tablet and desktop sizes plus development builds on a registered iPhone or iPad let the builder check the result. Sourcedocs.catdoes.com/features/debug-moderead 2026-10-18 |
|
| Browser / Computer-use | Partial |
|
Given the address of a public website from Lovable, WordPress, Shopify or elsewhere, the agent visits and studies the public pages, extracting the website's design, branding and content to build a matching native app, though pages behind a login are out of its reach and need a GitHub import instead. That reading of public pages is the extent of it: no agent clicks through websites, fills in forms or operates a desktop. Sourcedocs.catdoes.com/features/lovableread 2026-10-18 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Pricing
Core $20; Starter $50; Plus $100; Pro $200; Max $399 a month
Monthly or yearly subscription by credit allowance and project count; credits consumed per agent step by tier
Included quota
Free: 15 credits, 1 public project. Core: 50 credits a month, 2 projects. Starter: 200, 5 projects. Plus: 400. Pro: 900, 20 projects. Max: 2,000.
What is public
Every plan price, monthly and yearly, credit allowance, project count and feature gates.
Billing mechanics
Self serve monthly or yearly subscription with credit bundles; Enterprise by contact.
Cost watchouts
iOS and Android releases need Starter; GitHub integration needs Plus; code export and two way sync need Pro; Apple Developer and Expo accounts are separate; local taxes extra.
Variable cost rationale
Each agent step draws credits at a rate set by the chosen tier and task complexity, so planning on High or Max and large builds burn the allowance faster.
Additional watchouts
Higher agent tiers lock out when the balance runs low.
Overage / add-ons
Upgrade to a higher tier for more credits or buy one time credit bundles; High and Max agent tiers need balances above 100 and 230 credits and drop automatically below them.
Sales call required
Mixed (some tiers require a call)
Free / trial
Free plan with 15 credits, one public project and web deployment
Lowest paid plan
Core at $20 a month ($17 billed yearly): 50 credits a month, up to 2 projects, private projects, file browser, GitHub import
Commercial notes
Credits can also be earned by sharing on X or Reddit and through community programs.
Key ambiguities
Credits used per step at each tier are given only as relative cost, and the free plan's credit count differs between the pricing page and the FAQ.
Support SLA / resale
Custom SLA guarantees and a dedicated support team on Enterprise.
Missing data
Credits per step, credit bundle prices and Enterprise pricing.
Related vendors
- Cognition — Maker of Devin, an autonomous AI software engineer
- 10Web — Agentic website platform whose specialized AI agents build, host,…
- AgentUI — Managed AI app builder for operations teams that generates and hosts…
- Aider — Open source, model agnostic terminal coding agent that edits your…
- Anthropic Claude Code — Anthropic's agentic coding system across terminal, desktop, IDE, web…
- Anything — Anything is an AI app builder whose agent turns a plain language…