Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.trellistech.com/llms.txt

Use this file to discover all available pages before exploring further.

Stop doing the same things over and over

Every property manager has tasks they repeat dozens of times a week: sending check-in instructions, creating cleaning tasks after checkout, notifying the team about an issue. Automations handle these for you consistently once they are configured and tested.

What can you automate?

Guest messages

Send check-in instructions, checkout reminders, review requests, and follow-ups - automatically timed to each reservation.

Task creation

Create department-specific tasks when specific events happen - like a guest checkout or a new reservation.

Escalation alerts

Notify managers immediately when something needs human attention - a complaint, a refund request, or a conversation the AI cannot handle.

Operations digests

Send scheduled summaries of reservations, escalations, and tasks that need team attention.

Team notifications

Alert your team when tasks are assigned, conversations are escalated, or urgent issues come in.

Data updates

Automatically update records, change statuses, or tag conversations based on conditions you define.

How automations work

Every automation has three parts: a trigger (what starts it), optional conditions (when it should run), and actions (what it does).

Triggers - what starts the automation

The automation runs when something happens in Trellis:
  • A new reservation is created
  • A guest checks in or checks out
  • A conversation is escalated
  • A task is completed
  • A new message arrives
This is the most common type. For example: “When a guest checks out, create a cleaning task.”

Conditions - when it should run

Add conditions to make your automations smarter. For example:
  • Only create a cleaning task if the checkout is at a specific property
  • Only send a check-in message if the reservation is confirmed (not cancelled)
  • Only escalate if the guest has mentioned a specific keyword
Conditions let you build one automation that handles multiple scenarios intelligently.

Actions - what it does

Actions are the steps your automation takes. You can chain multiple actions together:
ActionWhat it does
Send a messageSend a templated message to a guest, owner, or vendor through supported connected channels
Create a taskCreate a department-specific task with all the details pre-filled
Update a recordChange the status, assignee, or any field on a property, reservation, task, or conversation
Notify your teamSend a notification to specific team members or roles
WaitPause the automation for a set time, until a specific date, or until something happens
Ask for approvalPause and ask a manager to review before continuing - perfect for high-stakes actions
Run the AI agentHave the AI agent analyze a situation, draft a response, or make a decision
Apply a ruleBranch the workflow based on a set of rules - useful when you have several conditions that should each take a different path
When you add an AI agent step with structured output, Trellis validates your configuration before saving. It checks that timeout values are long enough, that any referenced context is available, and that loop variables are only used inside loops. Fix any warnings to avoid unexpected failures at runtime.

Action access for AI agent steps

AI agent steps run with the agent you select for that workflow. Advanced workflow configuration can narrow what the agent can do in that step, which keeps it focused on the specific job the workflow needs. Trellis validates the configured action access before the workflow can run.
Keep AI agent steps focused. The more specific the task and available context are, the easier it is to test the workflow and understand what the agent will do.

AI effort level

For AI agent steps, choose an effort level that matches the job. Available levels depend on the step type:
LevelBest for
LiteThe simplest quick work, when available
FastShort, straightforward work where latency matters most
DefaultEveryday workflow reasoning and drafting
ReasoningMore complex analysis, branching decisions, and nuanced responses
PremiumHighest-quality drafting or analysis, when available
If you do not select an effort level, Trellis uses the default configured for the workflow or workspace.

Routing failed steps

Not every step has to take down the whole automation when something goes wrong. You can wire an on-error path from any step to another step on the canvas. If the step fails, the automation jumps to the step you connected - for example, sending a fallback message, notifying a teammate, or recording the failure to a task - and continues from there.
1

Open the step you want to protect

Click the step on the canvas to open its detail panel.
2

Connect an on-error path

Drag the on-error handle to the step that should run if this one fails. The connection is drawn as a separate path from the normal flow so you can see at a glance how errors are routed.
3

Save the workflow

Trellis validates the connections when you save. If a step you previously routed errors to is later removed, Trellis automatically clears the broken connection so your workflow stays valid.
On-error paths are perfect for graceful fallbacks - like sending a generic check-in message if a personalized one cannot be generated, or creating a follow-up task when a guest message fails to send. Steps without an on-error path mark the run as failed so your team can find it in the run history.

Workflow safety overview

Use the safety badge and risk popover on an automation to see, at a glance, whether the workflow is safe to run on autopilot. The safety overview summarizes:
  • Risky actions - steps that send messages, charge cards, modify reservations, or take other side effects that cannot be undone
  • Approval coverage - which risky steps require human approval before they run, and which do not
  • Suggested safeguards - gaps that Trellis recommends fixing before activating the workflow, like adding an approval step before a refund or narrowing what an AI agent step is allowed to do
Each step on the canvas also displays a safety badge so you can spot risky steps without opening the panel. Use the badges to quickly find places where an approval step might be the right safeguard.
Review the safety badge and risk details before activating a brand-new automation. Catching a missing approval at design time is much easier than catching it after a guest gets the wrong message.

Pinned test payloads

Most workflows have a handful of “what if” cases you’ll want to retest every time you make a change - a same-day check-in, a long stay, a cancelled reservation. Trellis lets you save these as pinned test payloads on the workflow so anyone on your team can reuse the same inputs during dry runs.
1

Open the test payloads panel

From the workflow detail view, open the test payloads area and choose Add test payload.
2

Save the payload

Add the JSON payload you want to reuse, name it clearly, and choose whether it should be the default dry-run payload.
3

Run a dry run

Select the saved payload when you dry-run the workflow. This makes it easy to retest the same reservation, sample message, or edge case after editing a step or adjusting a condition.
Test payloads are scoped to your workspace and respect the same permissions as the workflow itself - only members of the workspace can see, use, or remove them.

Human-in-the-loop approval

By default, automation steps - including ones that send messages or notify your team - run as soon as their conditions are met. Adding the workflow is itself the act of approval: you decided at design time what the automation is allowed to do. When a specific automation needs an extra checkpoint at run time, add an explicit approval step. The approval step pauses the workflow and asks a team member to review before continuing.
Messaging and team-notification steps are not gated automatically. If you want a manager to review a particular message before it goes out, drop an approval step in front of it. This keeps high-volume workflows (check-in reminders, departure messages, booking confirmations) running on autopilot while still letting you require review on sensitive flows like refunds or legal disclosures.
1

Automation reaches the approval step

The workflow pauses and creates an approval request visible in your inbox.
2

Reviewer is notified

The assigned reviewer gets a notification with all the context they need to make a decision.
3

Approve or reject

The reviewer approves the action (and the automation continues) or rejects it (and the automation stops or takes an alternative path). You can provide a reason when approving or rejecting - the reason is saved to the run history for your team’s records.
When to use approvals:
  • Before sending messages about sensitive topics (refunds, policy changes)
  • Before creating expensive maintenance tasks
  • Before sending messages that include sensitive financial data
  • Any time you want a human checkpoint in an otherwise automatic flow

Discuss approvals with the AI

Each pending approval card includes a Discuss button. Click it to open the AI chat sidebar with context about that specific approval already loaded. You can ask the AI to summarize the approval, explain what the workflow will do next, or help you decide whether to approve or reject.
Use Discuss when you want help understanding the workflow context before you make the approval decision.

Stale approval warnings

Long-running approval cards can display a warning indicator so you can spot overdue decisions at a glance. This helps prevent workflows from stalling when an approval is missed or overlooked.

Getting started with templates

You do not need to build automations from scratch. Trellis comes with pre-built templates for the most common property management workflows:
Trigger: 24 hours before check-in Action: Send the guest their check-in instructions, including property address, door code, Wi-Fi password, parking info, and house rules. Property details are pulled in automatically.
Trigger: Guest checks out Action: Create a cleaning task for the property, assign it to the default cleaning team member, attach the standard turnover checklist, and set the due date to the same day.
Trigger: Conversation is escalated (by AI or manually) Action: Send an immediate notification to the on-duty manager with the conversation summary, guest details, and a link to the conversation.
Trigger: Every morning Action: Send your team a summary of reservations, open escalations, and tasks that need attention that day.
Trigger: 1 day after checkout Action: Send the guest a friendly message thanking them for their stay and asking them to leave a review. Only sends if no issues were escalated during their stay.
Trigger: AI detects a maintenance issue in a guest message Action: Create a maintenance task, assign it to the on-call maintenance person, and notify the property manager.

Building your own automation

When you create a new automation, Trellis gives you two options: Build with AI or Start from scratch.

Build with AI

The fastest way to create an automation is to describe it in plain language and let the AI build it for you. Type what you want the workflow to do and the AI chat sidebar handles the rest - asking any clarifying questions it needs and assembling the complete automation.
1

Go to Automations

Navigate to Automations in the left sidebar and click New Automation.
2

Describe your workflow

Type what you want the automation to do in the “Describe what this workflow should do…” input and click Build.
3

Answer any follow-up questions

The AI chat sidebar opens seeded with your description. If it needs more detail - like which properties apply or when the workflow should trigger - it will ask clarifying questions. Answer in plain language or pick from the choices it presents.
4

Review and activate

Once the AI has enough information, it builds the complete automation and opens it in the visual builder. Review the workflow, make any adjustments, and activate it when you are ready.
Build with AI is the recommended way to create automations. It handles trigger configuration, condition setup, and action wiring automatically - so you get a working automation in seconds instead of configuring each step manually.

Start from scratch

If you prefer full manual control, choose Start from scratch to open the visual automation builder directly. The builder uses a canvas where you can see your entire workflow at a glance. A collapsible sidebar on the left lets you toggle between your workflow list and the canvas, giving you more room to work when you need it.
1

Choose a trigger

Select what starts your automation - an event (like a checkout), a schedule (like every Monday), or manual activation. Click the trigger node on the canvas to configure it in the detail panel.
2

Add conditions (optional)

Narrow down when the automation should run. For example, only for specific properties, reservation sources, or guest types.
3

Add actions

Define what the automation should do. Add as many steps as you need - send a message, create a task, wait, notify, and more. Click any step on the canvas to open its detail panel and configure it. Steps run in order.
4

Run now or activate

When you are ready, click Run now to execute the automation immediately, or toggle it on for automatic execution. Use the Builder, Runs, and Versions tabs at the top to switch between designing, monitoring, and reviewing past versions of your automation.
Trellis validates your workflow when you save it. If a condition or transform step references an input that hasn’t been declared, or if a scheduled automation uses date and time values without a timezone configured, you’ll see an error immediately - so you can fix the issue before the workflow ever runs.
Always test a new automation with a single property or reservation before rolling it out to your entire portfolio. This helps you catch any issues before they affect guests.

Monitoring automation runs

After an automation runs, you can review exactly what happened from the Runs tab. Each run shows its status, the steps it completed, and any errors that occurred.

List and calendar views

The runs tab offers two ways to view your automation history:
  • List view - the default table showing each run as a row with status, steps progress, trigger type, and timestamps. The list defaults to showing runs from the last 7 days so you see recent activity first. Adjust the date range at any time using the inline calendar. Best for scanning recent activity and filtering by status.
  • Calendar view - a month-based calendar that plots runs by date. Switch to this view using the toggle at the top of the runs list. Use the left and right arrows to navigate between months, and click any day to see the runs that executed on that date.
The calendar view is especially useful for scheduled automations. At a glance you can confirm that your daily or weekly workflows are running on the expected days - and spot any gaps where a run was missed.

Filtering runs

Use the filter options at the top of the runs list to find specific runs quickly. Filters apply to both the list and calendar views:
  • Status - filter by queued, running, completed, failed, or cancelled runs using the status tabs
  • Date range - pick a start and end date using the inline calendar. By default, runs from the last 7 days are shown.
  • Trigger type - show only runs started by a specific trigger (event, schedule, or manual)
  • Dry run - show or hide test runs (marked with a flask icon) so you can focus on real executions or review test results separately

Run details

Click any run to see a step-by-step breakdown on the canvas. Each step shows whether it succeeded, failed, or was skipped, along with its output. A Steps progress column in the run list gives you a quick visual summary of how far each run got before completing or encountering an error. A detail sidebar opens alongside the canvas so you can inspect each step without losing sight of the full workflow.

Step summaries

Click a step on the canvas to open its summary in the sidebar. Each step type shows information tailored to what it does:
  • Trigger steps - display the trigger type (manual, event-based, AI-started, or scheduled) and the event that started the run
  • Condition steps - show whether the condition evaluated to true or false with a clear badge
  • Action steps - list the entities the step affected, such as reservations, tasks, properties, or contacts
  • Loop steps - show the entities being iterated over and how many items were processed
When a step involves reservations, tasks, properties, or contacts, the sidebar shows clickable entity rows with key details at a glance - for example, a reservation row displays the guest name and check-in/check-out dates. Click any entity row to jump directly to that record in Trellis.

Details tab

Switch to the Details tab in the sidebar to see deeper execution information for any step. For steps that use the AI agent, this tab includes timing details and a plain-language timeline of what the agent did. A run-level execution log at the bottom shows every step in chronological order, making it easy to review the full sequence of events.

Run detail sidebar

When you view a run, a detail sidebar opens alongside the canvas. It gives you a complete picture of what happened without leaving the page. Summary tab The Summary tab shows at a glance what the automation touched:
  • Entity rows - reservations, tasks, properties, and contacts involved in the run are listed as clickable rows. Each row shows key details (like guest name, check-in and check-out dates for a reservation, or task title and status) so you can quickly identify the affected records and click through to see more.
  • Trigger summary - see what started the run: a manual trigger, a scheduled run, an event-based trigger, or an AI-started action.
  • Step-by-step results - each step displays its type badge, name, duration, and outcome. Condition steps show whether they evaluated to true or false. Loop steps list the entities they iterated over.
Details tab The Details tab provides deeper review information for each step:
  • Execution log - a chronological list of every step in the run, with timestamps and durations, so you can see the exact order of execution.
  • Step timing - for AI-powered steps, the sidebar shows response times and status details. This helps you understand where time is being spent and whether a step needs adjustment.
Step details load on demand - they are only fetched when you expand a step in the Details tab, keeping the sidebar fast even for runs with many steps.

Troubleshooting step errors

When a step fails, Trellis shows the error details directly on the canvas. Two actions are available on every failed step to help you resolve the issue quickly:
  • Ask AI to fix - Trellis analyzes the error, explains what went wrong in plain language, and suggests a fix. In many cases it can apply the correction for you automatically, so you can re-run the automation without digging into the details yourself.
  • Report issue - If the error looks like a bug or something you cannot resolve on your own, click to open a feedback dialog that is pre-filled as a bug report with the error details already attached. Submit it directly to the Trellis support team - no need to copy error messages or describe what went wrong manually.
Try Ask AI to fix first - it resolves most configuration issues instantly, like a missing field reference or an incorrect condition. Use Report issue when the error is outside your control or keeps recurring after a fix.

Run failure notifications

When a workflow has chronic failures, Trellis can notify the workspace so your team can respond before the automation keeps failing silently. Failure notifications appear in the notification bell in the top navigation bar. Click a failure notification to jump into the workflow runs view, where you can inspect the failed runs and use the troubleshooting tools described above.
Chronic failure alerts are deduplicated so a noisy workflow does not spam the team with a notification for every failed run.

Who can access automations

Automations are a manager-level feature. Only workspace admins and managers see the Automations entry in the navigation and can create, edit, or run workflows. Members and viewers do not see the Automations section - they only see the outcomes of automations (a task created, a message sent, a notification received) in the parts of Trellis they already use.
If a teammate cannot find Automations in the web sidebar or the web More menu on small screens, check their workspace role. Promote them to manager or admin if they need to build or manage workflows.

Cancelling a running automation

If an automation is currently running and you need to stop it - for example, if it was triggered by mistake or conditions have changed - you can cancel it directly from the run detail view. Click the Cancel button in the run header. A confirmation dialog appears to make sure you intended to stop the run. Once cancelled, the run is marked with a cancelled status and no further steps are executed.

Managing automations in bulk

Select multiple automations from the list to enable, disable, or delete them in one step. This is useful when you need to pause all automations for a season, clean up old workflows, or quickly activate a set of automations for a new property group.

Tips for great automations

Start with the templates

Templates are battle-tested and cover the most common workflows. Customize them to match your operation rather than building from scratch.

Use approval steps for high-stakes actions

Messages about refunds, policy changes, or financial data should always have a human review step.

Layer conditions for precision

The more specific your conditions, the smarter your automations become. A checkout cleaning automation that only triggers for confirmed stays avoids unnecessary tasks for cancellations.

Review automation logs regularly

Check the run history to make sure your automations are working as expected. Look for skipped runs or errors that need attention.

Use dry runs to test safely

Run your automation in test mode before activating it. Dry runs are clearly marked with a flask icon so you can distinguish them from real executions.

Filter runs to find issues fast

Use the status, date, and trigger type filters on the Runs tab to quickly narrow down to failed or unexpected runs.