Rakazo vs win.sh: Persistent Bots or Business Loops?
Compare Rakazo and win.sh on task ownership, recurring processes and approvals to decide how you want to direct agent work.
Founder, Task Machine
Rakazo and win.sh give you different places to direct recurring work: a persistent bot with its own conversation, or a company operator that turns business goals into projects. For a small software team, the choice starts with whether you want to manage individual jobs or ask an operator to decide what deserves attention.
Both publish ways to schedule work, retain context and request approval. The deciding difference is how they connect those capabilities: Rakazo gives bots routines and an optional ticket board, while win.sh describes loop contracts that specify triggers, authority and verification.
Rakazo starts with a bot you keep working with
Rakazo's bots and chats documentation describes continuous conversations with persistent bots. That suits a research or inbox-triage job where you keep refining the same bot's instructions. Its memory includes editable notes and shared space context, so the conversation has a place to retain lessons.
The optional Labs ticket board adds owners, priorities, comments and status changes. Assigning a ticket to a bot starts work. For a team that needs to point to the current owner of a specific job, that is a concrete addition to the chat model.
A bot can also organize other bots and schedule their work. Rakazo's docs do not describe a project, goal or team lead responsible for reviewing progress, so evaluate fleet coordination separately from accountable ownership of a business outcome.
win.sh starts with the business outcome
The win.sh how-it-works page describes an operator that monitors metrics and turns a goal into projects with deadlines and priorities. Its authority matrix governs what kinds of work proceed independently and what comes back for approval. A founder can direct the operator around an outcome and review the work it proposes.
Its local harness also records briefs, approvals and outcomes. The public pages do not describe a shared task record with named human or agent ownership and complete history. If your team needs to hand a job from a developer to an agent and back, inspect that handoff before choosing the operator model.
These are different coordination questions. A weekly revenue target needs someone to notice the number and propose work, while a particular customer investigation needs an owner, comments and a visible status. Pick the record your team will actually inspect when work stalls.
Repeating instructions and enforcing a process
Rakazo's routines documentation covers schedules, webhooks, GitHub events, email and Slack triggers, with a history for each routine. Skills store written instructions for repeated use. Its docs do not describe an enforced multi-step process with branches and conditions.
The win.sh local harness documentation describes editable loop contracts. Those contracts name triggers, authority and verification, and a loop can choose its next useful run time. The published bug-fix loop includes ordered executor steps, conditions and follow-up.
Consider a recurring maintenance investigation. If you want a persistent bot to follow saved instructions whenever an event arrives, Rakazo's routines give you that starting point. If you want to inspect a contract that defines the next action and follow-up conditions, examine a win.sh loop. The win.sh documentation does not describe independent verifiers for individual steps, so an ordered contract should not be read as that separate checking system.
Approval settings and evidence of completion
Rakazo's approval guide describes Ask, Auto and Never ask modes, editable Allow and Block lists, and approval cards in chat. Auto evaluates an action against your request. Rakazo warns that broad Always allow permissions for tools such as the shell or browser deserve particular care in event-started routines.
win.sh applies authority rules to types of work. Its pages do not establish separate action settings for each agent. The distinction matters when your team wants different permissions for individual bots doing similar work, or wants one business-wide rule governing a category of action.
Completion needs a separate discussion from permission. Rakazo documents a demonstration-taught skill with checks and a Test action, but its docs do not describe a general system that scores delivered work against acceptance criteria. The win.sh Bug Autofix contract calls for regression tests and checks error rates against success and failure criteria after deployment. That is a concrete software loop, not published evidence of independent requirement checks for every task.
Where execution and payment meet
Rakazo's pricing page lists free self-hosting, hosted own-model plans starting at $15 a month, or $12 a month billed yearly, and Rakazo credits starting at $20 a month. Credits cover model usage at provider prices and computer hours. Its machine-readable overview calls Cloud beta even though the pricing page sells hosted plans.
win.sh publishes hosted execution and an open-source local harness using your own coding-agent subscription. Its pricing page lists self-serve monthly budgets from $50 to $10,000, a hard limit and optional top-ups. Compare the work each payment funds without treating a budget and a hosted own-model plan as interchangeable units.
Rakazo and win.sh, point by point
The table presents both products' answers to the same questions, with each answer's source and check date. Read the explanations alongside the labels to see the published scope.
Both columns answer the same questions from each company's public pages. Yes means the company publishes the capability the question asks about, Partly means something narrower, No means it says it does not offer it, and Not published means its public pages do not describe it. Every answer links to its source and shows when we last checked it. The oldest check on this page was November 13, 2026.
| What you compare | Rakazo | win.sh |
|---|---|---|
| Goals |
Rakazo
Partly
You give each persistent bot its job in chat and can schedule recurring work. Its docs do not describe an owned goal with measurable targets and progress tracking. |
win.sh
Yes
win.sh monitors business metrics, takes a goal and creates a plan with projects, deadlines and priorities. Its company operator routes the work under your authority rules. |
| Leads |
Rakazo
Partly
A bot can organize other bots, rewrite their instructions and schedule their work. Its docs do not describe a project, goal or team lead responsible for reviewing progress. |
win.sh
Yes
The company operator reviews metrics, routes work and creates projects from goals. Its pages do not describe distinct goal, project and team leads with their own coordination responsibilities. |
| Tasks |
Rakazo
Yes
The Labs ticket board is shared by you and your bots, with owners, priorities, comments and status changes. Assigning a ticket to a bot starts its work. |
win.sh
Partly
win.sh lists task delegation, and the local harness records briefs, approvals and outcomes. Its public pages do not describe a shared task record with named human or agent ownership and complete history. |
| Recurring work |
Rakazo
Yes
Routines run on schedules, webhooks, GitHub events, email or Slack messages, and can watch for changes. Each routine keeps a run history. |
win.sh
Yes
Schedules, events and signals can start a run, and local loops choose their next useful run time. The documentation does not describe a stable root task with a separate task for every occurrence. |
| Workflows |
Rakazo
Partly
Skills save written steps for bots to repeat, and routines run prompts on triggers. Its docs do not describe an enforced multi-step process with branches and conditions. |
win.sh
Yes
Editable loop contracts specify triggers, authority and verification, and the bug-fix loop defines ordered executor steps, conditions and follow-up. The documentation does not describe independent verifiers for individual steps. |
| Autonomy |
Rakazo
Yes
Ask, Auto and Never ask modes control actions, with editable Allow and Block lists. Approval cards collect requests in chat, and Auto checks actions against what you requested. |
win.sh
Yes
An authority matrix sets what each type of work can do alone or send for approval, and the company operator follows those rules. Current pages do not establish separate action settings for each agent. |
| Learned autonomy |
Rakazo
Not published
Its docs do not describe proposing permission changes from your approval history. Always allow is a choice you make on an approval card, while Auto judges actions against your request. |
win.sh
Partly
Approvals, edits and rejections become operating rules, and Bug Autofix can propose more authority after at least 10 successful fixes and other checks. The documented upgrade proposal is not tied to approval history. |
| Memory |
Rakazo
Yes
Bots keep persistent notes and shared space memory. You can read, edit, download or delete notes, and optionally connect services for recall by meaning. |
win.sh
Yes
win.sh saves decisions, rules and learnings as editable company knowledge, and local loops keep journals. Its pages do not document restoring notes or separate agent, project, goal, team and workspace memory scopes. |
| Planning |
Rakazo
Partly
Each bot has a continuous chat and an editable Open work list. Its docs do not describe turning a planning conversation into a reviewed execution plan before work starts. |
win.sh
Yes
Conversational onboarding analyzes a website and creates a 30-day plan, and a goal becomes projects. The pages do not describe reviewing acceptance criteria before each agent task starts. |
| Intake |
Rakazo
Yes
A webhook URL and key route an outside POST request to a bot's routine. Email and Slack messages can also start routines, subject to connection and sender restrictions. |
win.sh
Partly
win.sh describes support issue triage, and its hosted API documents snapshot reads. The API says hosted write endpoints are not public yet, and the cited pages do not establish routing incoming requests to an owner or agent. |
| Connectors |
Rakazo
Yes
The app catalog supports custom remote MCP servers, OpenAPI and GraphQL connections. Routines accept webhooks and GitHub events, while self-hosted catalog availability depends on server configuration. |
win.sh
Partly
The catalog marks Stripe, Plausible and Google Search Console live, but marks GitHub, custom API and custom MCP soon. Other win.sh pages say GitHub is available and enterprise customers can add custom integrations, so availability is unclear. |
| Models |
Rakazo
Partly
You choose a main model, per-bot or per-routine overrides, and ordered fallbacks for failures or limits. Its docs do not describe automatic model selection by task. |
win.sh
Partly
The public model router recommends a model and fallback using task, quality, speed and price. Its documentation does not establish that hosted workflows automatically choose a model at each step. |
| Cloud or your own computer |
Rakazo
Yes
Rakazo offers hosted computers and self-hosting on your server or computer. The desktop app can run Rakazo locally or connect bots to your own computer for commands and files. |
win.sh
Yes
win.sh offers hosted managed execution or an open-source local harness that runs Codex or Claude Code in your own repository. |
| Your own subscription |
Rakazo
Yes
You can connect provider API keys or sign in with an eligible ChatGPT plan's Codex allowance. Rakazo recommends Anthropic API keys for Claude and warns against using Claude consumer subscription credentials for its model connection. |
win.sh
Yes
win.sh's local harness lets you run loops with your own agent subscription using Codex or Claude Code. |
| Pricing |
Rakazo
Yes
Self-hosting is free. Hosted own-model plans start at $15 a month, or $12 a month billed yearly. Rakazo credits start at $20 a month for model usage at provider prices and computer hours. |
win.sh
Yes
Self-serve monthly budgets range from $50 to $10,000, with a hard limit, optional top-ups and a seven-day money-back guarantee. |
| Correctness |
Rakazo
Partly
A skill taught by demonstration includes what to check, and Test runs it once. Its docs do not describe a general system that scores delivered results against defined acceptance criteria. |
win.sh
Yes
Bug Autofix asks for regression tests and checks the error rate against defined success and failure criteria after deployment. It does not document independent pass, fail or uncertain checks for every task requirement. |
For software work
| What you compare | Rakazo | win.sh |
|---|---|---|
| Code review rules |
Rakazo
Partly
Rakazo publishes a pull-request review bot template and lets you edit bot instructions. Its docs do not establish code review against repository rules. |
win.sh
Not published
Bug Autofix directs the agent to add a regression test and run checks. The public pages do not describe reviewing changes against discovered repository rules or screening proposed rules against examples. |
| Adversarial review |
Rakazo
Not published
Its docs do not describe an independent second review of risky code changes. The README says bots can check each other, and groups can use parallel helpers, without specifying that review process. |
win.sh
Not published
Bug Autofix asks for risk notes and tests. The public docs describe no independent two-reviewer adversarial review for high-risk changes. |
| Risk rating |
Rakazo
Not published
Its public pages do not describe rating a change's risk before merge. The GitHub catalog lists pull-request inspection and merge actions, rather than a risk assessment. |
win.sh
Partly
The bug-fix executor reports remaining risk and test output with a pull request or patch. It does not document a merge-risk rating for the exact revision. |
| Pull requests |
Rakazo
Partly
The GitHub catalog includes creating pull requests, reading reviews and replying to review comments. Its docs do not establish an agent delivery loop that revises code in response to review feedback. |
win.sh
Partly
Bug Autofix instructs its local executor to open a pull request or provide a patch, and requires approval before merge. Its public contract does not describe exact-commit approval or addressing review feedback. |
A fact we cannot find or cannot check is left out, and prices can change after the date shown. Task Machine publishes this comparison and is an interested party. Rakazo and win.sh are named only to identify them, and there is no relationship between the companies. Has something changed? Tell us.
Who each one suits
Pick Rakazo if persistent bot conversations, editable memory and shared tickets fit how you delegate research or maintenance. Its docs do not describe an enforced branching process, so check that written skills are sufficient for your recurring job.
Pick win.sh if a company operator should monitor metrics, create projects and run editable loops under authority rules. Its public pages do not describe a complete shared task record with named human or agent ownership, which matters when colleagues exchange responsibility.
Where Task Machine fits
Task Machine answers the shared-record and process questions with Tasks owned by people or agents and Workflows with branches, approvals and verifier checks. Its workflow documentation describes returning failed work with feedback, while the Inbox collects consequential decisions. It fits a maintenance process whose handoffs and acceptance checks you need to inspect.
Task Machine's published agent execution choices are Managed Cloud and Local Workers. Task Machine is the wrong pick if you require free whole-application self-hosting, since its docs do not describe a free, open-source self-hosted application. See Switch from Rakazo to Task Machine. Use Task Machine vs Rakazo and Task Machine vs win.sh to examine that distinction, then map one recurring maintenance job to its owner, approval boundary and completion check.