Skip to content

Workflows · Agentic use cases

Agents do the legwork. Your team makes the call.

Fourteen workflow patterns for your business, across API integrations, AI agents, automation and custom apps. Each shows what starts the work, what the agent does and what your team gets back.

Map your workflow Browse the patterns
  1. Request
    “Plan next week's website changes and flag anything blocked.”Asked in team chat
  2. Agent work
    Linear tasks created with owners, priorities and due datesChecks the current Linear plan, GitHub and recent notes
  3. Approval
    Team lead confirms priorities before anything movesSensitive steps wait for a person
  4. Output
    Plan shared, blockers raised, status kept current in LinearReply posted back to the same chat
Sample request: team chat becomes a planned workflow, with priorities and delivery tracked in Linear.

See a connected journey from brief to billing: watch the Linear agentic invoicing demo →

All workflow patterns by business function

Customer follow-up

Enquiries, calls and campaigns that reach the right person while the interest is still warm.

Often involves Website forms, CRM, phone, Google Ads, and whatever else your workflow needs.

Also in this group: Turn website enquiries into sales follow-up

  1. Capture phone enquiries before they vanish

    Inbound calls can become visible follow-up work, with call alerts, durable logs, and a clear path into CRM follow-up.

    Steps & tools
    1. Starts with · 01A prospect calls your business number
    2. Agent work · 02The call event is captured securely
    3. Agent work · 03The team gets an immediate alert and durable log
    4. Business result · 04The follow-up record is queued for CRM update

    Tools in this patternTwilioTeam alertsCRMOpenClaw

  2. Build Google Ads campaigns from a plain-English brief

    Describe the objective, offer, audience, budget, and landing page in natural language. The agent turns it into a full campaign structure ready for review.

    Steps & tools
    1. Starts with · 01You describe the campaign objective, offer, budget, and landing page
    2. Agent work · 02The agent creates the campaign, ad groups, ads, keywords, negatives, and tracking
    3. Agent work · 03A human reviews the full draft, evidence, and spend controls
    4. Business result · 04The campaign is published to Google Ads paused and ready for activation

    Tools in this patternGoogle AdsReview dashboardLanding pagesOpenClaw

    Built for teams who want campaign creation automated while approval, publishing, and activation stay under human control.

  3. Generate campaign drafts for review

    Your team can turn a campaign brief into a structured draft pack with copy, keywords, tracking notes, and a clear approval path.

    Steps & tools
    1. Starts with · 01Your campaign brief is given in chat
    2. Agent work · 02The agent creates draft assets and rationale
    3. Agent work · 03The draft is placed in your review workspace
    4. Business result · 04A reviewer validates the campaign before publishing

    Tools in this patternCampaign workspaceGitHub/PRGoogle AdsOpenClaw

    Useful when marketing teams want repeatable campaign creation without giving an agent permission to spend money.

Team, inbox & admin

Everyday requests, email and documentation handled from one governed place, without the admin drag.

Often involves Team chat, Microsoft 365, Outlook, Jira, Confluence, and whatever else your workflow needs.

  1. Ask for help from chat

    Send a plain-English request and get a useful answer, action, or status update without opening five different systems.

    Steps & tools
    1. Starts with · 01Your team sends a request in the approved chat channel
    2. Agent work · 02The agent works out which system is needed
    3. Agent work · 03It checks the right tools and records what it found
    4. Business result · 04Your team gets a clear reply with the action taken or the next decision needed

    Tools in this patternChatOpenClawAI agentsBusiness tools

  2. Daily email briefing and review queue

    Important emails, calendar context, receipts, and cleanup candidates are surfaced without handing the agent unlimited inbox control.

    Steps & tools
    1. Starts with · 01A scheduled check runs
    2. Agent work · 02The agent reads only the approved mailbox and calendar scope
    3. Agent work · 03It groups what needs attention or review
    4. Business result · 04The owner receives a briefing and approves any cleanup actions

    Tools in this patternMicrosoft 365OutlookTeam alertsPrivate dashboard

  3. Update Jira and Confluence without the admin drag

    Backlog updates, project notes, documentation changes, and status comments can be handled from one governed agent surface.

    Steps & tools
    1. Starts with · 01Your team asks for a Jira or Confluence update
    2. Agent work · 02The agent checks the right workspace, page, or issue
    3. Agent work · 03It drafts or applies the approved change
    4. Business result · 04The record is updated with a traceable explanation

    Tools in this patternJiraConfluenceOpenClawApproval rules

Linear planning & delivery

Linear tasks that move from request to tested, reviewed work, with blockers raised before they stall.

Often involves Linear, GitHub, browser QA, scheduled checks, and whatever else your workflow needs.

  1. Keep delivery work moving in Linear

    Linear tasks and projects become the operating view for progress, blockers, approvals, quality checks, and what is ready to ship.

    Steps & tools
    1. Starts with · 01A Linear task, project, or comment changes
    2. Agent work · 02The agent checks the work state and next dependency
    3. Agent work · 03It updates the delivery record and flags blockers
    4. Business result · 04Stakeholders see what is ready, blocked, or waiting for approval

    Tools in this patternLinearOpenClawTaskFlowGitHub

  2. Turn a delivery request into tested work

    A feature request can become a branch, tested change, review note, and status update without someone manually coordinating every step.

    Steps & tools
    1. Starts with · 01A task is raised or assigned
    2. Agent work · 02A specialist agent works in the correct codebase
    3. Agent work · 03It builds, tests, and captures evidence
    4. Business result · 04The team gets a pull request or status update ready for review

    Tools in this patternGitHubOpenClaw agentsBrowser QALinear

  3. Check a website in a real browser

    The agent can load pages, capture screenshots, check responsive states, and report what actually rendered instead of trusting static tests alone.

    Steps & tools
    1. Starts with · 01A page or feature needs review
    2. Agent work · 02The agent opens it in a browser
    3. Agent work · 03It captures screenshots, layout checks, and interaction evidence
    4. Business result · 04The team gets a concrete QA finding or approval note

    Tools in this patternBrowser automationWebsitesLinearGitHub

  4. Keep projects warm between conversations

    Scheduled checks keep an eye on active work, stale blockers, review gates, and tasks that are ready for a decision.

    Steps & tools
    1. Starts with · 01A scheduled heartbeat runs
    2. Agent work · 02The agent checks the current project state
    3. Agent work · 03It identifies blockers, dirty work, and approval-ready items
    4. Business result · 04The right records are updated or a decision is requested

    Tools in this patternScheduled checksLinearTaskFlowMemory/docs

Finance & governance

Receipts, separated workspaces and outside events handled inside clear access boundaries.

Often involves Microsoft 365, SharePoint, webhooks, review queues, and whatever else your workflow needs.

  1. Send receipts to the right expense record

    Receipt capture becomes a structured back-office workflow with the original attachment retained and exceptions routed for review.

    Steps & tools
    1. Starts with · 01A receipt arrives by chat or email
    2. Agent work · 02The agent extracts supplier, date, GST, and line-item detail
    3. Agent work · 03It checks for missing data or duplicates
    4. Business result · 04A finance record is created with the receipt attached

    Tools in this patternChat/email intakeAI document readingMicrosoft 365SharePoint

  2. Run separate brands or tenants safely

    Different brands, departments, or client environments can use the same operating model without mixing data, tools, memory, or credentials.

    Steps & tools
    1. Starts with · 01A request arrives for a specific brand, tenant, or client
    2. Agent work · 02The agent uses only that workspace and credential set
    3. Agent work · 03It updates the right records for that environment
    4. Business result · 04Work stays separated from every other operating lane

    Tools in this patternTenant workspacesJira/ConfluenceCRMIsolated OpenClaw gateway

  3. Connect external events without opening the whole system

    External services can trigger narrow, signed workflows while private dashboards stay protected behind the right access boundary.

    Steps & tools
    1. Starts with · 01A trusted external event or review need appears
    2. Agent work · 02The event enters through a narrow secure route
    3. Agent work · 03The agent records, reviews, or acts on the event
    4. Business result · 04The result is stored, shown in a private UI, or sent for approval

    Tools in this patternWebhooksWorkflow intakeReview queuesOpenClaw

Questions

Before you map your own

Every pattern here is scoped again around your tools, permissions and approval rules.

What are these patterns?

Representative workflows we design and build for businesses: customer follow-up, team operations, delivery and finance. Each one is scoped around your tools, permissions and approval rules, then built, tested and handed over.

Can these workflows be adapted to our tools?

Yes. The examples describe the operating pattern. The actual implementation is scoped around your tools, permissions, approval rules, and data boundaries.

Why keep approval gates in the examples?

Because these workflows touch customers, money, records, and reputation. The agent prepares or performs bounded work; sensitive actions stay reviewable.

Private intake

What would you like to connect or improve?

A few sentences are enough to start. Request a workflow review and we will reply by email with a practical next step. No booking is made by submitting this form.

What happens next:
  1. Nebulas reviews your context.
  2. We check fit and urgency.
  3. We reply by email with the most practical next step.

We use your details to review and reply to this enquiry, and keep a record of it. Privacy details

Call us directlyCall the team

Name, email, a short description and consent are all we need. Please do not include passwords or sensitive records.

Add details (optional)