SETU, in plain words.

1. What Setu is

1. SETU is a workspace for AI-assisted company work. You describe a job, choose its name and role, review the instructions and decide what it may access. AI-generated work needs your review; SETU must not impersonate a person.
2. AI employees are digital workers, not human staff. Each has its own role, instructions, skills and scoped memory; its permitted tools, storage and computer access depend on its configuration, which services are available and what access the owner grants. A SETU plan does not include a personal computer or code sandbox.
3. Nothing is sent, published, bought or changed outside Setu without your approval, unless you decide otherwise under Settings → Approvals or mark a tool as trusted.
4. Each employee's profile shows what it is doing right now: Working now, Waiting to start, Waiting for you, Stopped: needs you, Last task failed, Resting until its next duty, Ready for work, or Paused.
5. Spending warnings: Setu tells you once when company spending passes 50%, 80%, 90% and 100% of its limit, and again after you raise the limit. An employee with a monthly allowance gets the same warnings, and is paused automatically when its allowance is used up; raise the allowance, then press Start working.
6. A task that waits for another task starts on its own once you accept the task it was waiting for, if you had already tried to start it.
7. Reports → Download as CSV saves the audit trail (who did what, and when) to a file. Recorded events can never be edited.
8. Each employee knows who it reports to, who reports to it, how much freedom you gave it and which kinds of action need your approval, and uses this to plan and explain its work. It does not change what the employee may do: your Approvals settings still decide, and a changed setting reaches the employee on its next task.
9. Open a task and choose “Where this work comes from” to see the goal it serves, its project and milestone, the tasks it was handed down from (with who did each and who they report to) and the work handed on from it. With access to one department only, related work elsewhere is counted but not shown.
10. Monthly allowance: open an employee → Resources → “Monthly allowance” to see this month's allowance, what is spent, what is held for work in progress and what is left; the owner or an administrator can change it with a short reason. If it says monthly allowances are not switched on for this workspace yet, the amount is saved but not enforced.
11. Home starts with a short brief: “While you were away” after four hours or more, otherwise “This morning” or “Today so far”. It shows what your team finished, what waits for you, what stopped and why, what was spent, who is paused and what starts next; click a count to open the right page. It is counted from your records, uses no AI and costs nothing. Department members see their department only; a billing administrator sees the money lines only.
12. Settings → Approvals, an employee's lessons, the live status and the Reports download are also shown in Hindi when you choose हिन्दी as your interface language.
13. If a run fails for a temporary reason (the AI engine was busy, or its answer could not be read), Setu tries once more by itself, on your backup model when you have set one, and within the same spending limits. If the second try also fails, or the problem is not temporary, you get one message naming the employee, the task and the employee's manager, with what to do next. Setu never retries a run stopped by a spending limit, a permission or an approval.
14. Saved memory and lessons are explicit records you can inspect and edit. A correction or saved note is not proof of full or autonomous self-learning; check its wording and scope before relying on it. Passwords and codes should not be supplied as lessons.
15. Settings → Approvals: choose what your employees must ask before doing. For sending outside the company, publishing, deleting data and other changes inside an account, pick Always ask, Don't ask or Never allowed; sending also offers “Ask only for new recipients” (a recipient is known once you approved an action to that email address and it was confirmed as done). Paying always needs you in this version. Sending, paying, publishing and deleting start on Always ask, and a trusted tool does not skip that. “Different rules for a department or one employee” overrides the company setting; the employee's rule wins over its department's. Only the owner can change this page, and Setu checks it every time an employee prepares an action.
16. Your AI provider bills live model usage directly through your customer-funded OpenRouter account or your own provider API key. Your Setu subscription is separate. Practice fixtures do not call a live model; connected services and computer providers may charge separately.
17. More → Help & feedback opens this guide, feedback and WhatsApp options. Canvas search → Help opens the full Help page; public pages have their own Help button.
18. Use canvas search (Ctrl+K) for Employees, Work, Connections, Files, Reports, Spending and other pages. Modules open over the same company canvas. More holds account, appearance, AI engine and Help controls; the Company tree and workflow canvas remain interactive.
19. Find anything (Ctrl K, or the search box at the top): type a name, a task, a customer or any words from your knowledge. Results are grouped (people, tasks, drafts and results, knowledge, objectives, customer messages) and open where they live. You only find what your role and department let you open.

2. Start in five minutes

1. Voice: press the microphone beside “Tell your company what you need” and speak in English or Hindi; your words are added to the box for you to check and send. Open any draft and press “Read this draft aloud” to hear it in the most natural voice your device has. Voice works in Chrome and Edge; nothing you say is sent to Setu until you press Send.
2. From an empty company canvas, choose Make your first hire for a guided conversation about one person. Answer five core questions, review the summary, and choose when to save an inactive draft; it starts without a budget or access grant. The saved hiring history stays visible only to you as its author. You decide whether and when to activate that person.
3. Just looking? “Explore the demo company” on the landing page gives you your own copy of a running company: edit anything, it is saved. A Setu plan provides platform access; live AI work also needs your funded model connection.
4. For a whole team, choose Describe my team. One prompt starts a bounded proposal: one independent role, several independent roles reporting to you, or a team with reporting lines. Edit and re-check the proposal, then explicitly create drafts; no employee drafts are created before confirmation, and new drafts have no tools or spending. Review each role, access and limit before activating anyone. This is separate from Make your first hire, which guides you through one person.
5. Create a company in three short steps: company details, first outcome, then review and create. You land directly on the company canvas; Make your first hire, Describe my team and Start guided setup are optional next steps. Your typed setup saves to your account as you type; check its status to confirm each save. If a save cannot be confirmed, your text stays in the form but may not be on the server, so copy any changes you need before replacing it with the saved version.
6. Settings → AI engine: connect your customer-funded OpenRouter account or your own supported provider API key using the options shown. For an OpenAI-compatible API, enter the provider's HTTPS chat completions URL and supply its model name, price and key. Practice mode, when available, produces structured fixture briefs without a live model call. Saved keys are encrypted and never shown again. The Getting started card on Home opens this tab for you.
7. Spending → set a company spending limit (the “Authorized limit”). Live AI work stops at that ceiling; practice mode needs no limit.
8. Settings → General → Interface language switches Setu's own menus and headings to Hindi on this device; “Language your employees write in” is a separate, company-wide choice for drafts and replies.
9. Home → type what you need in the box (“write a WhatsApp reply for a delivery question”). Setu suggests an employee, or you pick one; the result appears under Work and, when ready, under Approvals.
10. Example: type “Draft a 3-line WhatsApp reply to a customer asking where their order is.” Expected output: one task assigned to your support employee, and a few minutes later a draft reply waiting under Approvals for you to accept or send back.
11. Each chat request has a task spending limit within the employee and company limits; it is a ceiling, not provider credit. If it stops with “Needs a higher limit”, open the task and press “Raise this task's limit from … to … and run” (you confirm the new amount), or raise the company limit under Spending. Limits are never raised on their own.
12. If nothing happens after a request: check that at least one employee is active (Employees), that a key is saved (Settings → AI engine) and that the company is not paused (Settings → General → Operating boundaries).

3. Give your company its knowledge

1. Project memory: when you add knowledge or a document, choose “One project (objective)” and pick the project. Only tasks under that objective use it, so one client's brief never reaches another client's work. Everyone who can see company knowledge sees it, labelled with the project. Facts an employee learns from it stay with that project and are not added to the employee's memory for other work.
2. Knowledge → “Write a procedure, a handbook, or custom knowledge”. A handbook is company-wide standing knowledge: vision, mission, policies, brand voice, how you work. Write it first.
3. A procedure is a step-by-step way of doing a job, for the whole company, one department, or one employee. Employees carry handbooks and procedures on every task and follow them.
4. Custom knowledge (price lists, product details, client facts) is used when it is relevant to a task. Tick “always include” if it must be there every time.
5. Knowledge document upload is controlled by a feature flag and accepts original files up to 2 MB where enabled. This is not a general form preview for PDF or Word files; verify supported formats and extracted content in the feature itself. Access follows company permissions.
6. Example: add a price list as custom knowledge, then ask “Quote 50 units of product A with delivery to Pune.” Expected output: a draft quote using your exact prices, with anything it had to assume marked for you.
7. Writing knowledge does not start a model run. Live AI use of knowledge is billed by your connected provider; connected services may charge separately. Without knowledge, employees draft with assumptions and mark them for you.
8. Facts and saved notes wait here and on each profile for review where the feature is available; check their recorded source, scope and approval state before relying on them. When two sources or facts disagree, “Information that needs your decision” opens on this page and both records are kept out of work until you decide; it stays folded while nothing is open.

4. Shape each employee

1. Start from a ready role in the library, or write your own. Review and edit the title, instructions and skills before saving a draft (chief of staff, customer support, social media, SEO, content, performance marketing, email marketing, sales development, bookkeeper, HR recruiter, operations, research, data, product, software, catalogue, contract reviewer, executive assistant). Every one drafts for your review; change anything before saving.
2. On Company view, select an employee card to open its inspector. Open profile and Connect tools (or View tools) are available there; Edit role & access is for owners and administrators. Use the reporting and department controls on Company view or Employees to change placement. Review and confirm a proposed change before it is applied; the change is recorded in Reports. The Company graph supports keyboard navigation as well as pointer controls.
3. Employees → click an employee → “Edit role & access”. Instructions say what the job is; Skills say how to do parts of it (one per paragraph, “Name: how to do it”). The operating limit is the most one run may spend.
4. When editing a paused or draft employee, availability starts unchecked. Save the revised role first, or explicitly select Make available for work now when you want that employee to become eligible for work.
5. Skill import supports Markdown and text files or a supported GitHub source, subject to separate size limits (up to 48 KB for a file and 12 KB for imported text). Review the preview and choose what to add; imported skills become editable instructions.
6. Memory and lessons: “What it has learned from its work” shows explicit records from corrections, failures, owner instructions and work accepted first time. Owners and administrators can correct, archive or block a lesson; only the owner can use “Share with everyone” on an active employee lesson to make its scope company-wide. The same correction or failure is flagged as a repeated mistake at three occurrences. These records can guide later tasks but do not prove autonomous self-learning or prevent every future error. Passwords and codes should not be supplied as lessons.
7. Storage: notes the employee keeps for later tasks, plus its computer files if it has a computer. You can edit, remove or add text files there and download other files. Give employees a computer only when a computer service is separately configured and access is granted to that employee. The “desktop” template adds a screen; confirm its actual capabilities, credentials and charges under Connections. A computer with the desktop template also keeps the last three screenshots the employee took under Computer files, so you can see what it saw.
8. Access & tools: tick only what the role needs. Anything not in the list is not connected yet; connect it under Connections first.
9. Never do: Edit an employee → Responsibilities → Never do instructions for written limits (up to 20, 300 characters each). Written instructions guide the AI; they are not an automatic action block. Under Tools & accounts → Blocked tools, choose restrictions that override assigned access, including company defaults. A named app-tool block wins over access to its whole account, and blocking the whole account prevents every tool there. Blocking does not grant any new access. Save, then read the restrictions in Access & tools and the saved role revisions in Activity. Earlier revisions may say Not recorded. Clear a block explicitly to remove it; pause the employee to stop all work. A read-only check of an earlier uncertain external action can still establish what already happened, without sending it again.
10. Outside agent (owner-only, where enabled): connect an agent you already operate through the supported endpoint and protocol. For an OpenAI-compatible chat completions address, enter the provider's HTTPS URL, then provide its model details, price and key. It stays off until configured; this is not a one-click marketplace install. Its own provider charges and connection upkeep are separate. Setu’s access and review controls still apply.
11. Standing duties (employee profile → Duties): give an employee a routine, such as a morning inbox check every day at 08:30, every Monday at 09:00 in your time zone, or every 4 hours. Each duty has its own spending limit per run, and a task is created and started on schedule; the result comes to you for review. While the last result is still open, that run is skipped; a paused company or employee also skips it.
12. Only on working days: tick it on a standing duty and it skips days off, company holidays and times outside working hours set under Settings → General → Work calendar (working days, holidays such as 2026-10-02, and optional hours, in the company time zone). The duty shows “Day off: …” and runs at its next time; duties without the tick run every day.
13. Example duty: “Every day at 08:30: read new customer messages, draft replies, list anything the owner must decide.” Expected output: one task each morning under Work marked as a standing duty, with a draft under Approvals. If a duty cannot start, you are notified (Settings → Notifications) and the reason shows on the duty.
14. Performance (employee profile → Performance): accepted first time vs after changes, acceptance rate, cost per accepted result, typical time to your acceptance, sends that left the company, customer messages handled, quality-review criteria met, and accepted results per week. Choose the period at the top.
15. Saved role revisions: Employees → employee → Activity → Load saved revisions. Open a revision to read the recorded name, role, instructions, permissions and per-run limit; Load older revisions continues the history. This is read-only and does not restore old settings or restart work. Department-scoped members see only employees currently in their department and revisions recorded there; missing earlier numbers do not mean records were deleted. Changes are compared with the earlier visible revision named on screen. If loading fails, retry; access refusal clears the loaded history. To change the employee, use Edit role & access and review the new settings.
16. Customer messages reach your employees: on the profile → Access & tools → “Connect incoming requests”, choose WhatsApp (Meta Cloud API) or Email (Brevo inbound parsing) and follow the steps shown. Each message becomes a task with the sender's address and earlier messages, with its own spending limit per message; the reply is drafted for your review and goes out only through an approved send. Work shows an Inbox of these messages.
17. If an employee's task stops with “Needs access”, open Employees → the employee → Edit role & access and tick the tool, or grant the employee on the connection under Connections; then press “Grant access, then run” on the task.

5. Connect the outside world

1. Connections → pick a card → Connect. Every card walks you through the steps: the account, the key with a “where to find it” guide, an automatic check, and who may use it. Keys are encrypted; check the connected provider's charges and account requirements.
2. The Connections page shows the services and tools available in your deployment. A catalogue listing does not mean an app is installed, configured or ready to use. SETU plans do not include runtime hardware; outside runtimes, messaging and connected providers may require separate accounts, setup and payment.
3. The app directory at the top of Connections lists catalog options. Check each card for availability and setup; a listing is not proof that it is connected. Apps that support sign-in use your account; apps that use a key ask for it during setup. You choose employee access; supported changes need separately reviewed setup, approval and result verification.
4. Search the catalogue for Google Calendar, Gmail, Google Sheets or Google Ads. Check support to see advertised tools and whether the current account has a reviewed route for them. A listing or connected account does not establish that a particular report or change works.
5. A SETU plan does not include a computer or code sandbox. If a separately configured runtime is available in your deployment, review its own storage, access, network and billing terms before assigning work; its service credential and usage are separate from your model key.
6. If a separately configured computer service is available, its “base” or “desktop” templates and exact capabilities are shown in the connection. They may depend on a service credential and can incur separate charges. Do not assume screenshot, browser, filesystem or internet access without checking that runtime’s actual configuration.
7. Before assigning work to any runtime, confirm its actual access and available files; do not infer that a web address or computer action is available from a product listing.
8. Shopify, HubSpot, GitHub, Brevo email and WhatsApp use a key from that service; the guide tells you exactly where to copy it. A read-only database connection runs one query at a time with a ten-second limit and at most 200 rows; nothing can be changed through it.
9. Software teams on GitHub can connect GitHub from the app directory before configuring ready actions: “Create a work branch”, “Save a file to a work branch” and “Open a pull request”. These reviewed routes require exact-request approval, branch checks and result verification; other advertised GitHub tools need their own support check.
10. “Everyone, including employees added later” gives the connection to the whole team and to future hires. Each granted employee receives a skill note explaining the tools.
11. Trusted tools: on a connected tool press “Trust: run without asking” and employees can use it without waiting for your approval. Leave it off for anything that sends, posts or spends.
12. Pause one connection: Connections → Pause next to an account stops every read and action through it at once, including approved actions not yet sent, without removing it. Resume puts it back exactly as it was. Revoke removes it for good.
13. Webhooks (Connections page, bottom): a website form, Zapier, your shop or any script can hand a task to a chosen employee under a chosen objective. The token is shown once; copy it then. Each delivery becomes a task reviewed like any other work.
14. If a connection stops working (“Needs access” on a task): open the connection under Connections, run its check again or paste a fresh key, then press “Grant access, then run” on the task. You can revoke any connection there at any time.

6. Work, review, approve

1. Verification evidence: open a draft or an approval, then expand “Verification evidence”. Text and format, sources and domain rules, external actions, quality review and acceptance have separate results. A passed check covers only what its explanation says. Practice output stays labeled; missing checks say so. Source changes, revised criteria and newer drafts can make earlier evidence historical. Accepting a draft never means that it was sent or that every quality check passed.
2. Home shows what needs you first: “Needs your attention” groups stopped tasks by cause with the one action that fixes it. “Needs a higher limit → Raise this task's limit (confirm the new amount), then run”, “Needs access → Grant access, then run”, “Failed → Open the task and retry”, “Needs changes → Run again with your saved feedback”.
3. More → Company progress shows how many pieces of work are finished, waiting for you or stuck. Its percentage uses real deliverables, excluding reviews, planning steps and cancelled tasks. Open it to inspect the work on Work; a configured employee is not proof that a task has finished.
4. Work lists every task with its state and the reason if it stopped. “Run with connected AI” starts it (costs AI usage); “Run local fixture” makes a practice brief without a live model call. A task's own limit is a ceiling the product never raises by itself: when the estimate is above it the task stops with “Needs a higher limit” and the dialog offers “Raise this task's limit from … to … and run”, which you confirm. A task assigned without a limit takes the employee's per-run limit at its first paid run (practice runs leave it unchanged) (recorded in Reports as a System change). A run above the employee's per-run limit stops until you raise that limit under Employees → employee → Edit role & access. If the AI engine fails or does not answer, the task shows “Outcome needs checking”: the amount reserved for that attempt is held, and running it again counts the full reserved amount as estimated spending because the provider's final usage is unknown. Cancel task stops later model calls. Any model call or knowledge lookup already sent can still incur a cost; known usage is counted as measured, and an unresolved lookup keeps its estimate. An answer that arrives after cancellation is withheld.
5. Choosing the AI model: Settings → AI engine sets the company's model. When assigning work you can pick another approved model of the same provider for that task (for example a faster, cheaper one for simple work); its own approved price applies. A backup model (Settings → AI engine) takes over new runs automatically while the main model's last three attempts have failed, and the run history says so.
6. How the work was done: open any task and expand “How the work was done” to see each run (practice or connected AI, when, its state and measured cost). “View steps” shows what the employee looked up, what came back and what each step cost. People placed in a department see their department's runs.
7. If a run fails: open the task, read the reason, and press “Run with connected AI” again to retry. If it fails again, check Settings → AI engine (key and model) and the employee's access. After three attempts in a row end without an answer from the AI engine, new paid runs pause for ten minutes (the message says until when) so no more of your limit is held against an engine that is down; practice runs and scheduled duties carry on, and duties simply try again later.
8. Approvals shows each draft exactly as it will be accepted. Accept it, or Request changes with notes; the employee revises with your notes next time. “Changes to connected accounts” below lists sends and account changes waiting for your approval, one by one.
9. Department sign-off (Settings → General → “Department heads sign off first”): a person you placed in a department as manager or reviewer signs off that department's drafts before anyone accepts them, and is emailed about them. The approval card shows “Waiting for sign-off from …” and then “Signed off by …”. Heads can sign off or request changes but not accept. An owner or administrator can still accept without the sign-off by writing why; the reason is kept and shown in Explain this decision.
10. Managers can review their team's drafts first (Settings → General; one to three rounds). The manager's verdict appears on your approval card. Managers cannot accept drafts or approve sends for you.
11. Explain a decision: open any accepted or sent-back draft (task → Output history) and press “Explain this decision”. It shows only what was recorded then: who decided and when, the reason (optional when accepting: write it in the notes box), the manager's note, the other versions and why they were sent back, the brief, limit and measured cost, and the sources and remembered facts exactly as they were used. Anything not captured says “Not recorded”. It lists what has changed since (a source revised or removed, the employee paused, the limit changed) with dates. “Reconsider this decision” starts a new task for the same employee with that record and your note; the earlier decision stays as it was and nothing is sent without review. For example, after sending a supplier proposal back and changing its brief or buyer, the earlier decision still shows the original brief, employee and limit. A draft created later is not listed as an option considered earlier. If the connection fails, keep the same note and retry in the same browser tab: the pending request returns its existing task even if the original information changed. Reloading that tab keeps the request reference when browser storage is available; your note text is not stored there, so enter the same note again. A changed note or a new confirmed request starts separate work. The complete original brief stays editable, while your note, the recorded reason and changed assumptions travel as read-only decision context to the employee and manager. Reconsidering again keeps that earlier context too. If the complete history exceeds the supported limit, no new task is created: create a focused assignment with the instructions you want reviewed. Earlier tasks already shortened by an older version are not reconstructed. The normal model and spending limits still apply. Very old requests whose original input cannot be matched ask you to review the existing task rather than creating a duplicate. Older decisions made before context capture say “Not recorded” for missing details; create a new task with your chosen instructions instead of copying today's task into their history.
12. Files keeps accepted outputs with their version. Where export is supported, formats include DOCX, XLSX, CSV, JSON and patch; each file is limited to 4 MiB. Open a file, download it, or save it as knowledge. These outputs do not establish arbitrary video editing or delivery capability.
13. Reports is the company audit trail: every action, approval, tool call and spend, newest first, with details on each; “Load older events” at the bottom shows earlier ones.
14. Run several companies (agencies, groups): All companies in canvas search shows each company's drafts, sends and stopped runs, work in progress, accepted results, employees, customer messages, spend and plan, sorted by what needs you. Open a company from there to act.
15. Templates: browse and search 18 company packs, even when your team already exists. Preview the roles, objectives and first tasks before applying a pack to a new, empty company. Choices include digital agency, D2C brand, clinic, real estate, coaching, software, recruitment, hospitality, restaurant, nonprofit, events, logistics, manufacturing, consulting, publishing, travel, construction and agriculture. Installation creates draft employees and supporting records, with scheduled duties disabled. Review the handbook and budgets before activating employees, and enable each duty separately when ready. If an install was interrupted, use its recovery action to confirm the same request. Select starter tasks and connect workflow dependencies separately.
16. Plans and billing (Spending): Starter is ₹1,999 for 30 days with 300 runs; Team is ₹4,999 for 30 days with 1,500 runs. Both have unlimited role profiles. These are platform fees; your AI provider, messaging, runtime and connected tools may charge separately. Cashfree purchases cover 30 days without automatic renewal; if a Stripe subscription is offered, check its recurring terms before checkout. Add-ons add runs to the current period. Refund eligibility is shown with the invoice: requests are available within 7 days when fewer than 10 runs were used.
17. Notifications (Settings → Notifications): SETU can notify you through configured email or WhatsApp connections when a draft waits for review, a send waits for approval, a run stops, a limit is reached, a customer writes in or a duty cannot start. Several things at once arrive as one message. The email carries a one-use link: open it on your phone, read the draft, and accept it or send it back with notes without signing in. If your company has not connected its own WhatsApp Business account, WhatsApp alerts come as a short message from Setu's own number (when Setu has one set up).
18. People (Settings → People): invite members with a one-use invitation link and see who belongs to the company. The owner can hand the company to another member with “Make owner”; it asks for a fresh sign-in, the previous owner stays as an admin, and both people are told.
19. Open Company chat to choose Company for shared discussion or Private chat and one employee from your current authorized roster. Conversation visibility follows your role and employee access; private chat is not owner-only. An unsent message or in-progress request locks switching. Clear an unsent message to switch, or retry an uncertain request in the same conversation before moving on.
20. Employees hand work to each other: a draft can include work for a colleague. Once you accept the draft, a direct report's task runs on its own; a peer's task waits for you to start it. Each shows “Handed over by”.
21. Anything that changes data outside Setu (send an email, create a product, post a message) is a proposal you approve one by one, unless you trust that tool.
22. Risk labels: every change an employee proposes shows Low risk (only looks at information), Medium risk (changes something inside your account, like a draft or a record) or High risk (sends, publishes, pays, spends or deletes; hard to take back). Under Connections you can change a tool's label, and the owner can switch on “High-risk actions (send, publish, pay, delete) always wait for a person”, so even a trusted tool or a manager cannot approve those.
23. Approve several at once: tick the boxes on waiting drafts or account changes and press “Accept selected” / “Approve selected” (or “Send back selected” / “Reject selected” with a note). Each one is still checked and recorded on its own; high-risk changes are always decided one at a time.
24. Autonomy (Employees → employee → Settings): Watch only (reads and reports back), Prepare only (drafts, but cannot send, publish, pay or delete), Ask every time (a person approves each change; trusted tools do not skip this) or Works on its own (today's behaviour: trusted tools go ahead). For a manager or CEO employee, the owner can also choose which risk levels it may approve for its own team (“Save approval powers”); you still see who approved what.
25. Slack: under Settings → Notifications, paste a Slack incoming-webhook address (it starts with https://hooks.slack.com/services/) to get the same alerts in a Slack channel. Slack posts never include the one-use decision links, because a channel is shared.

7. Objectives, workflows and schedules

1. Handoff instructions: while assigning work, editing an unstarted task or building a later workflow step, record constraints, open questions and a target date. They travel with the task and its AI request; they do not grant access or automatically cancel work at the deadline. Open a task → Load handoff records to see each run's exact accepted input version, original brief, criteria, employee revision and reserved budget. A prepared record is not proof of success. Older runs may have no complete record; nothing is invented from today's settings.
2. Goals & projects → “Plan an objective”: describe the outcome and what a good result looks like; a manager proposes the steps, who does what, and how much each may spend. Planning with a connected AI model uses your key and limits; the practice template makes no live model call.
3. Example: objective “Launch the Diwali offer”. Expected output: a plan with steps such as write the offer, design the post, draft the customer email, each with an employee and a limit, for you to review and save as a workflow.
4. Workflows: publish a workflow before running it (editing creates a new version). Steps run in order and in parallel where possible; each step still comes to you for review. Dependent steps wait for accepted inputs; a decision step runs only the route you accept.
5. You can save a workflow plan with draft or paused employees. Saving the plan does not activate them or start work. Before publishing, every step owner must be active and have permission to draft. Running also checks the employee versions pinned when you published.
6. Schedules: put a workflow on a repeat, every day at a local time or at a regular interval, in your company's time zone. A previous run must finish before another starts; choose what happens to runs missed while the service was offline.
7. If a workflow step stops for a low limit, open its task and press “Raise this task's limit from … to … and run” (you confirm the amount); the rest of the workflow continues from there.

8. Money and safety

1. Every AI request reserves an estimated cost before it runs and settles the measured cost after. Setu checks its estimate against company, employee (per run), department and task limits. These Setu limits are estimate-based admission ceilings, not prepaid money or a provider-side spending cap. Review your provider's own spending controls and set a separate provider spending limit where it offers one.
2. Spending shows the authorized limit, what was used, what is reserved and what is available, plus every usage record by employee and task. Reports shows every action, approval and tool call.
3. When the company limit is reached: raise it under Spending → “Adjust limit”, then re-run the stopped task. When only one task is short, use its “Raise this task's limit from … to … and run” button instead.
4. Open Spending for the plans and eligibility available to your company. A SETU subscription or eligible platform trial does not supply AI provider credit. Live AI work needs your own funded provider connection and the provider bills it separately. When offered to an eligible verified Google account, the platform trial is one 3-day period, needs no mobile number or payment method, and does not restart for another company. It is not open unless the account’s Spending page offers it.
5. Saved provider keys use AES-256-GCM encryption with tamper checking, bound to the company and purpose. Settings show connection status instead of returning the saved key. SETU decrypts the key on its server to make an authorized request to your provider; the provider receives that request under its own terms.
6. Use the dedicated connection fields for keys. Keep passwords and keys out of chat, instructions, knowledge and uploaded documents, because task content may be sent to the AI provider. Connecting a model does not grant access to your email, files or other accounts; each tool needs its own connection and permissions.
7. The owner can replace or remove a saved company model key in AI engine settings. Removing SETU's copy does not revoke the key at the provider: use the provider's account settings to revoke it or set a provider spending limit. Disconnect tools in Connections and pause the company under Settings → General → Operating boundaries when needed.
8. Settings → Advanced: export your organization as a portable file (private instructions, sources and keys are left out; workflows travel as structure, without step briefs or connections), import one, see the “Behind the scenes” health panel, and archive or delete the company.
9. Need help from Setu? In “Behind the scenes”, press “Create support link”: a read-only link that works for 24 hours and shows run states and stuck work only (no content, prompts or passwords). “Withdraw link” stops it at once.
10. Customer-funded AI work still follows Setu's configured company, employee and task limits. These are spending controls, not provider credit. Uploaded files are checked first: PDFs with scripts or embedded files and Word files with macros are refused (save a plain copy, for example print to PDF).
11. System health (Settings → Advanced → “Behind the scenes”): how much work is waiting and for how long, what is running against your company's limit of parallel runs, successes and failures in the last day, typical run time, and today's spend compared with a usual day. If one company is very busy, other companies' work still starts in turn.

9. Agency walkthrough: from brief to delivered work

1. Start with one client and a clear deliverable, such as a campaign analysis and three social post drafts. Create the company, then preview the digital agency template. Review each role, its manager, access and budget before activation; keep scheduled duties off until the first workflow works as intended.
2. Add the client brief in Knowledge: audience, brand voice, approved claims, channels, source data and what the team must not invent. Give the manager, analyst, content specialist and reviewer access to the sources they need.
3. Connect your funded AI provider in Settings → AI engine and review its model and pricing. Set a company ceiling in Spending, then smaller employee and task limits. An AI key does not connect a social account or give an employee computer access; configure those capabilities separately when needed.
4. Build a workflow with an analyst, content specialist and separate reviewer. Define the output and acceptance criteria for every step. Link accepted analysis as an input to the writing step, then link the writing to review. Save the plan, check the owners and limits, and publish it when the team is ready.
5. Run one workflow and inspect the actual outputs. Check calculations and source claims, request one correction, and accept only the version you intend to use. If work stops, open the task's reason, resolve the missing access or limit, and resume through the offered action.
6. Open the accepted result in Files and download it. If you also need publishing, first configure a supported tool for the intended account and verify its access. Review the proposed destination and exact content, then check the recorded outcome and the destination itself. A prepared draft or queued request is not a published post.
7. Review the completed tasks, handoffs and measured usage in Reports and Spending. Save approved reusable knowledge, then enable a repeat schedule or standing duty only after this complete cycle succeeds. Add more clients and employees in measured steps so quality, permissions and spending remain clear.

What each page does

Try it

Explore the demo company View plans