Pisces AI Manager clarifies requirements, breaks down work, selects suitable models, dispatches Executors in parallel, supervises execution, reviews results, and preserves the project’s memory. You no longer have to manage every AI instruction one by one.
The complete AI Manager software, the same as the paid tiers; AI usage is billed separately.
30-day free trial
Complete AI Manager software and ongoing human support; AI usage is billed separately.
Do not want to wire up your own AI? Pick a Managed AI tier and receive Monthly Credit every month; the total below follows your choice.
Managed AI adds to your plan, it does not replace it: the add-on allowance is added to the allowance your subscription already includes, and each is released and cleared on its own billing period.
TotalUSD 7.99/month
30-day free trial
For teams and companies that need a larger monthly AI allowance; it includes more Monthly Credit than Personal.
Do not want to wire up your own AI? Pick a Managed AI tier and receive Monthly Credit every month; the total below follows your choice.
Managed AI adds to your plan, it does not replace it: the add-on allowance is added to the allowance your subscription already includes, and each is released and cleared on its own billing period.
TotalUSD 14.99/month
When you use Codex, Claude, or another AI on its own, you still have to split the work, repeat context, track progress, inspect results, and steer it back when it drifts.
No independent Manager keeps checking the objective, constraints, and definition of done.
History keeps growing, costs rise, and the important details become easier to bury.
Without difficulty-based routing, every task may consume the highest-tier model.
A chat alone does not show tasks, Executors, diffs, tests, or review evidence.
Explain the outcome as if you were briefing a project manager.
The Manager fills information gaps, creates a plan, and defines tasks and acceptance criteria.
The Manager assigns work to 1–6 Executors and selects models by difficulty.
The Manager checks results, diffs, tests, errors, and tokens before completing, correcting, or continuing.
Clarifies requirements, plans, breaks down tasks, dispatches work, supervises, and performs final acceptance.
Configure 1–6 Executors. The interface recommends 3–6 to keep collaboration clear and controlled.
Use lower-cost models for simple work and stronger models for complex work, with manual override available.
Decisions, constraints, completion records, and key context continue across chats, models, and computer restarts.
Seal older conversations into cumulative memory to reduce repeated context in long-running projects.
Bring results, tokens, diffs, tests, failure reasons, and next steps back into the same project conversation.
Pisces AI Manager is built so you can check what actually happened, instead of trusting that it happened.
Switch between the two at any time, or configure a fallback source in the same project.
Pisces AI Manager seals earlier conversation into cumulative project memory while retaining the complete original text. Later execution sends only the context needed for the current task, reducing repeated history in long-running projects.
In the illustrated long-conversation example, repeated context is reduced by up to about 90%. This is not a fixed savings guarantee. Actual results depend on the model, context window, task complexity, attachments, and project length.Subscribers can use a growing library of preset project structures, Skills, and Workflows so the Manager begins with a clear idea of what to create, how to split the work, and how to verify the result.
Move from requirement to implementation, tests, and release checks.
Plan pages, build a responsive interface, and verify delivery.
Align audience, message, channels, assets, and measurement.
Go from script and storyboard through production and final review.
Research, draft, edit, and perform a pre-publication review.
Start with your own goal, rules, and acceptance criteria.
Models such as Codex and Claude perform specific work. Pisces AI Manager clarifies the goal, breaks down tasks, coordinates multiple Executors, preserves project memory, and reviews results using diffs, tests, and usage evidence. It is the management layer, not a replacement model.
No. Software and AI usage are billed separately. You can use only your own CLI or API without incurring platform-credit charges.
Yes. Configure multiple connections, use different providers for the Manager and Executors, or keep Managed AI as a fallback.
No. No software tier — Starter, Personal or Professional — includes platform credits; software and AI usage are separate.
Add a Managed AI subscription for a recurring Monthly Credit allowance, or buy Credit as a one-time top-up. Both are deducted only when you use platform-hosted AI.
No. You settle charges directly with your own AI provider and the platform does not deduct credits.
Yes. Switch at any time or configure a fallback source inside the same project.
Starter, Personal and Professional all carry the complete Manager and support multithreaded Executors; no software feature is held back from a tier. The tiers differ only in subscription terms such as devices, trial and human support.
No. Planning, scope controls, and evidence review can reduce drift and omissions, but AI and automation can still make mistakes. Review important changes and keep backups.
Local project files remain on your computer. Cloud text retention and deletion follow the Privacy Policy and account controls; purchased Credit remains subject to the applicable terms.
No. Credit is a one-time top-up: it is not issued monthly and does not reset or clear each month, and it stays until it is used. It is separate from the Monthly Credit released by a Managed AI subscription, which is cleared at the end of every month and never accumulates.
One Manager, up to six Executors, and every result backed by evidence you can review.
Start for free