Task Machine vs win.sh

Task Machine and win.sh agree on several choices that separate them from autonomous-company builders. Both connect to accounts the customer owns, take no revenue cut, cap spending, support recurring work, and hold risky actions for approval. The useful comparison starts after those similarities: win.sh runs a continuous business loop and reports back, while Task Machine lets a person direct explicit work and routes judgment calls into one inbox.

What does win.sh do well?

win.sh is built around a 24/7 loop that monitors connected systems, proposes the next move, acts inside rules you set, and reports each morning through Telegram and a Decisions tab. A per-work-type authority matrix gives different kinds of work different trust levels. Approvals, edits, and rejections become operating rules, so autonomy can increase without becoming one global switch.

The commercial and ownership model is unusually clear for this lane. Customers connect accounts they own, choose a monthly budget from $50 to $10,000, get dollar-based receipts and a hard cap, and pay no revenue share. The product also includes recurring workflows, categorized memory, a CLI, and an Agent Skills catalog. Its onboarding and autonomous-run polish are real strengths.

How does the operating rhythm differ?

win.sh starts from continuous monitoring. The loop runs before someone asks, handles work inside its authority, and condenses the result into a morning brief. That suits a founder who wants the business watched continuously and is comfortable governing through standing rules and review.

Task Machine starts from direction. Chat turns a discussion into work, the inbox gathers approvals, questions, failed verifiers, and exceptions, and tasks preserve the detailed state. Recurring work runs through explicit workflow graphs rather than one open-ended company loop. The operator sees where work is and can intervene before a consequential step ships.

How is control represented?

win.sh represents control through the authority matrix, approval categories, budgets, receipts, and the Decisions surface. The system learns which kinds of action may proceed as the founder accepts or corrects them.

Task Machine represents control inside each workflow. Human-question, approval, and verifier nodes sit beside agent steps and branch conditions. A failed verifier becomes inbox work attached to the task and run. This is more setup than governing a general loop, but it leaves a readable process and step-level history when a client, teammate, or operator needs to understand what happened.

What does each product ask from the operator?

win.sh asks the operator to connect accounts, define the rules, review the brief, and tune authority. It is the better fit when low-touch operation is the goal and the work can be governed by stable categories.

Task Machine asks the operator to install or define workflows, connect workers and tools, and stay available for the judgment calls they deliberately keep. It fits work where the sequence, handoffs, evidence, and approval point matter as much as the final action.

When should you choose each one?

Choose win.sh when you want a polished autonomous daily loop over accounts you own, a light morning review rhythm, and granular authority that grows from previous decisions.

Choose Task Machine when you want the three-surface workflow of chat, inbox, and tasks, explicit verifier-backed runs, and one queue for judgment across work performed by humans and agents. You keep 100% of your revenue, Task Machine takes no cut, and it never custodies your accounts.

Common questions

Does Task Machine replace the win.sh morning brief?

No. Task Machine centers the live inbox and task state rather than a morning-summary rhythm. Scheduled work still runs, but the primary control surface is the queue of decisions that need attention.

Which product requires more setup?

Task Machine usually does because the workflow graph, gates, workers, and connectors are explicit. win.sh is ahead when the goal is to connect systems and let a general loop begin monitoring quickly.

Can both products stop risky actions?

Yes. win.sh uses approval gates and an authority matrix. Task Machine uses explicit approval and verifier nodes, budgets, permissions, and autonomy levels inside the workflow.