An AI agent platform from Creao AI Limited, based in Hong Kong. Describe what you want in plain language in the chat, and it calls the tools it needs and carries the work through to completion. Its distinguishing feature is that a session that went well can be converted into a reusable agent with a single click. Once created, an agent can be run again and again by swapping out its input fields, and higher plans support scheduled runs. It can also generate full-stack applications, including the backend, database, and hosting, and no programming knowledge is required. It launched in July 2025 and can be tried on a free plan.
Key Features
- Carries work through to completion from chat: Ask in plain language ─ “put together this week’s competitor report,” for instance ─ and it selects the tools it needs on its own, moving from research to generation to output in one pass. Generated files can be previewed on the spot and downloaded.
- Turns successful sessions into agents: Work that went well once can be saved as an agent with structured input fields in a single click. From then on there is no need to rewrite the same instructions, and the agent can be shared so the whole team runs the same procedure. Version control is supported.
- Scheduled runs and memory: Set an agent to run on a schedule and it keeps running at fixed times without anyone operating it (Pro Plus and above). A memory feature retains facts and preferences across conversations, so background context does not have to be restated every time.
- Automatic full-stack app generation: It generates not just the screen but the backend, database, and hosting as well, so a small internal tool can be published as is.
- 27 connectors and MCP support: It connects to 27 services, including Gmail, Google Calendar / Sheets / Drive, Slack, Notion, GitHub, Discord, and Shopify. It also supports the Model Context Protocol (MCP), so you can register your own MCP server to reach in-house tools.
- Browser automation and an API: It can perform operations on websites on your behalf while keeping login state. A developer API is also offered, so agents can be invoked from your own systems.
Pricing
Billing is credit-based: each plan grants a monthly credit allowance that usage draws down. Every plan also includes a daily bonus of 5 credits.
| Plan | Monthly (monthly billing) | Monthly (billed annually) | Monthly credits | Highlights |
|---|---|---|---|---|
| Free | $0 | $0 | 30 (one time) | 1 connector, no scheduled runs, restricted access to premium models |
| Pro | $20 | $16 | 200 | Unlimited connectors, core access to all features |
| Pro Plus | $50 | $40 | 600 | Unlocks scheduled runs and long-form video / audio rendering |
| Max | $150 | $120 | 2,000 | More simultaneous scheduled runs |
| Ultra | $1,000 | $800 | 20,000 | For large-scale operations |
If you run out of credits, additional credit packs can be purchased at $0.15 per credit.
Pricing is current as of August 2026. Please check the official site for the latest details.
Pros & Cons
✅ Pros
- Work that went well once can be turned into an agent as is, removing the chore of rewriting the same prompt every time
- Scheduled runs are supported, so routine tasks keep going without a person starting them
- With 27 connectors and MCP support, automations can be wired into the tools a team actually uses
- Because the backend and hosting are generated too, non-engineers can publish something that works
- A free plan is available, so the experience can be tested at no cost
⚠️ Cons
- Credit-based billing makes consumption hard to predict depending on how heavy the work is, so monthly cost is difficult to estimate in advance
- The free plan is tightly limited ─ one connector and no scheduled runs ─ and is not well suited to evaluating serious automation
- Scheduled runs require Pro Plus ($50 per month) or above, which raises the cost if recurring execution is the main goal
- Launched in July 2025, it is a relatively new service, and Japanese-language information and case studies are still scarce
- Because automation reaches into business systems, permission management for connected accounts has to be designed by the user
Comparison with Similar Services
| Criteria | Creao AI | Manus | Genspark | Lovable |
|---|---|---|---|---|
| Main use | Task execution + agent reuse | General-purpose agent doing work for you | General-purpose agent + search and document creation | Web app generation |
| Reuse of output | Sessions become shareable agents | Mainly per-session | Mainly per-session | Generated apps can be edited continuously |
| Scheduled runs | Supported (Pro Plus and above) | Supported | Supported | Not supported |
| External tool integration | 27 connectors + MCP | Broad support | Broad support | Mainly development services |
| Billing | Credit-based (free plan available) | Credit-based (free tier available) | Credit-based (free tier available) | Credit-based (free tier available) |
Who Is It For
- People who want weekly or monthly routine reports and data collection to keep running without manual effort
- Small teams that want to share the same procedure across everyone and reduce dependence on one person
- Non-engineers who want to build and publish a small internal tool themselves without asking an engineer
- Anyone who wants to try a custom automation wired to in-house tools through MCP
- Anyone who has used a general-purpose AI agent once and found rewriting the instructions every time tiresome
Summary
Creao AI is a platform built around the idea that agent-driven work should not end as a one-off run: a procedure that succeeded can be kept as an asset. Combine scheduled runs, memory, 27 connectors, and MCP support, and work first tried out in chat becomes part of how the business operates. On the other hand, the difficulty of estimating cost under credit-based billing and the requirement of Pro Plus or above for scheduled runs are real constraints. Start on the free plan to confirm that your own work actually runs through to completion, and consider a paid plan once you have found tasks you repeat.