squad.so vs win.sh: Two Operating Models
Compare squad.so’s chief of staff and teammates with win.sh’s company operator on task ownership, authority and recurring work.
Founder, Task Machine
squad.so and win.sh both organize recurring work for a founder, but they make different commitments about control. squad.so gives you a chief of staff, teammates and a Needs you queue, while win.sh describes an authority matrix that governs which kinds of work can proceed alone.
The distinction matters before you delegate outward-facing work. A queue tells you what an agent has brought back for attention, while an authority rule determines which actions must wait. Compare those published boundaries alongside the way each product records ownership and repeated jobs.
Who does squad.so fit?
squad.so, formerly MissionControlHQ, suits a founder who wants a hosted chief of staff to hire and coordinate teammates. Its agent documentation describes approving a plan before creating a teammate. A ready-made job brief also asks the chief of staff to plan with you and begin in draft-for-review mode.
Its task documentation describes steps, comments and ownership by a teammate or a person. People can steer a task through comments, and teammates can route work to one another. The documentation says teammates cannot currently assign a task to you, although older assignments remain visible.
The task record is useful when a founder wants to inspect who is working on a particular job. Its documentation does not describe a repeatable process with a fixed route through every step. Written skills and steps within a task need to be evaluated separately from an enforced recurring workflow.
Who does win.sh fit?
win.sh suits a founder who wants one operator to turn business signals into work. Its how-it-works page describes monitoring metrics, creating projects from goals, and setting deadlines and priorities. The operator routes the work under authority rules you configure.
The local harness provides another entry point for technical founders. It runs editable loops with coding agents inside your repository, using your own agent subscription. Those loops retain briefs, approvals and outcomes.
win.sh's public pages do not describe a shared task record with named human or agent ownership and complete history. A founder who needs colleagues to exchange ownership of a task should check that requirement directly. The company operator and the task record solve related but different coordination problems.
Where does each product put human control?
squad.so's control documentation describes app permissions and the teammate's judgment about when to ask. Needs you gathers questions, approvals and blocked work. Its documentation does not describe a configurable requirement to request approval before selected actions.
The same documentation says draft-first email depends on instructions and the teammate following them. A teammate can send an email without a draft. If an outgoing message must receive approval, evaluate the exact mechanism that keeps it from being sent, not only the queue where drafts appear.
win.sh publishes an authority matrix by type of work. Its operator follows those rules and sends work for approval when required, though its pages do not establish separate action settings for each agent. Decide whether your policy is about a category such as outreach or a particular teammate's permissions.
Both systems retain lessons. squad.so's memory guide describes shared memories you can review, correct or clear, plus teammates' own notes. win.sh keeps editable company knowledge and local journals. Neither set of published descriptions should be read as a promise that a human approval rule will be inferred correctly without review.
What makes a recurring run predictable?
squad.so routines start teammates on schedules, and triggers start runs from webhooks or app events. Skills provide written instructions, while tasks can contain steps. Its documentation does not describe a repeatable multi-step process that runs the same way each time.
win.sh's local harness documentation describes loop contracts with triggers, authority and verification. The loop can select its next useful run time. Its bug-fix contract includes ordered executor steps, conditions and follow-up, though the documentation does not describe independent verifiers for individual steps.
Context is also part of recurrence. squad.so warns that a routine begins with less context than a chat and may not load teammate Notes, depending on the model. Its guide recommends placing required context in the routine instructions or naming a document to read. A successful chat interaction is therefore insufficient evidence for the scheduled version of the same job.
For software work, win.sh's Bug Autofix asks for regression tests and checks the error rate against success and failure criteria after deployment. squad.so lets you describe done and correct work in task comments, but its documentation does not describe results being checked against defined criteria. Inspect the completion evidence separately from the status update.
Which execution costs are included?
squad.so's cost guide lists $99 per month plus your AI provider's charges. Optional teammate inboxes cost $10 per month each and human seats $9 each. Research gets 1,000 free credits a month, then costs $5 per 5,000 credits. The required squad plan is flat-priced.
Its teammates run on a cloud computer. Squad for Mac adds approved access to local files, browser-tab details and screen pictures, but does not move execution to the Mac. Bringing your AI account and running the agent locally are different choices.
win.sh's pricing page lists self-serve monthly budgets from $50 to $10,000, with a hard limit and optional top-ups. It publishes managed execution and a local harness. Before paying, confirm required integrations because its catalog and other pages disagree about GitHub availability.
squad.so and win.sh, point by point
The table shows the answers to the same questions with the source and check date of each. Its explanations distinguish a published capability from a narrower mechanism.
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 | squad.so | win.sh |
|---|---|---|
| Goals |
squad.so
Partly
A teammate can produce a recurring report on metrics you name. Its documentation does not describe a goal owned by an agent or named owner with a tracked metric or target. |
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 |
squad.so
Yes
The chief of staff hires and coordinates teammates, and teammates can route work to one another. The documentation does not describe separate goal and project leads that review and propose gaps. |
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 |
squad.so
Yes
Tasks hold steps and comments, and can be owned by a teammate or a person. Teammates cannot currently assign a task to you, although older assignments remain visible. |
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 |
squad.so
Yes
Routines start a teammate on a schedule, while triggers start a run from a webhook or connected app event. The documentation does not describe a stable root task with a separate task history for each occurrence. |
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 |
squad.so
Partly
Skills give teammates written instructions and tasks can be broken into steps. The documentation does not describe a repeatable multi-step process that runs the same way each time. |
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 |
squad.so
Partly
Needs you collects questions, approvals and other blocked work, and app permissions limit what teammates can do. The choice to ask a question is the teammate's. The documentation does not describe a configurable requirement for approval before selected actions. |
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 |
squad.so
Not published
Teammates draft outward-facing work for approval until you trust them. Its documentation does not describe proposals to change autonomy based on approval history. |
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 |
squad.so
Yes
Shared memories learn from conversations and can be reviewed, corrected or cleared. Teammates also have their own notes. The documentation does not describe the same separate agent, project, goal, team and workspace memory scopes with restore. |
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 |
squad.so
Yes
A ready-made job brief asks the chief of staff to plan the job with you and start in draft-for-review mode. When hiring a teammate, you answer questions and approve a plan before creation. The documentation does not describe reviewing acceptance criteria before every task 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 |
squad.so
Yes
Webhook and connected-app triggers start a chosen teammate with instructions, and tasks can be assigned to a teammate. The documentation does not describe project-token intake routed through a team lead's approval queue. |
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 |
squad.so
Yes
The app catalog, custom API-key integrations and hosted MCP servers give teammates tools, and triggers accept webhooks or connected-app events. The custom MCP setup does not accept SSE server URLs, and the documentation does not describe a general custom command-line connector. |
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 |
squad.so
Partly
You choose a squad or teammate model and an ordered fallback list. The documentation describes fallback on failure or rate limits, not automatic model choice by task or step. |
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 |
squad.so
Partly
Squad teammates run on a cloud computer. Squad for Mac adds approved access to local files, browser-tab details and screen pictures. Mac access does not move teammate execution onto your computer. |
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 |
squad.so
Yes
Squad teammates use your ChatGPT, Claude Code or GitHub Copilot subscription, or supported provider API keys, for model usage. Your provider bills you under your own plan or key. |
win.sh
Yes
win.sh's local harness lets you run loops with your own agent subscription using Codex or Claude Code. |
| Pricing |
squad.so
Partly
The squad costs $99 per month plus your AI provider's charges. Optional teammate inboxes cost $10 per month each, human seats $9 each, and research gets 1,000 free credits a month, then costs $5 per 5,000 credits. The required squad plan is flat-priced. |
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 |
squad.so
Partly
You can describe what done looks like and correct a task through comments. The documentation does not describe results being checked against defined 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 | squad.so | win.sh |
|---|---|---|
| Code review rules |
squad.so
Partly
The GitHub integration can work with code, branches, pull requests and reviews. Its documentation does not describe reviewing changes 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 |
squad.so
Not published
The GitHub integration can work with pull requests and reviews. Its documentation does not describe an independent second review of risky changes. |
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 |
squad.so
Not published
The GitHub integration can read pull requests and repository information. Its documentation does not describe a risk rating of a change before merge. |
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 |
squad.so
Partly
The GitHub integration supports code, branches, pull requests and reviews. The documentation does not establish agents addressing pull request 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. squad.so 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 squad.so if a chief of staff, hosted teammates and task comments fit how you delegate and review work. Its documentation leaves the decision to ask with the teammate, so draft-first instructions are an important limit for mandatory approval requirements.
Pick win.sh if goal-driven projects, type-of-work authority and adaptive loops fit your operating routine. Its public pages do not describe complete shared task ownership, and its integration pages need reconciliation before you rely on a disputed connector.
Where Task Machine fits
Task Machine gives people and agents shared Tasks and lets you set each agent's actions to Ask first or Automatic. Its autonomy documentation describes consequential decisions arriving in the Inbox. Workflows specify approval steps, branches and checks that can return failed work with feedback.
Task Machine is not the right choice if your priority is a hosted chief of staff or one company operator selecting the next useful loop. Compare Task Machine vs squad.so and Task Machine vs win.sh, then test one outgoing-email job by identifying exactly what prevents sending before approval.