Who can use this feature?
👤 Organization Admins
🔑 Contact your Customer Success Manager if you're unsure whether this feature is available on your plan.
In this article:
What are AI Workflows?
AI Workflows let your legal operations team design and automate multi-step AI-powered processes — without writing code. You define a sequence of steps (nodes) that the AI follows whenever the workflow is triggered. Each step can generate text, make decisions, use LawVu data, or wait for user to approve next step.
Common use cases include:
Use case | Details |
Matter triage auto-assignment | Use AI to review incoming matter details, determine the nature, urgency, and legal area of the request, then automatically route the matter to the appropriate team, practice group, or legal professional. |
Contract triage auto-assignment | Use AI to analyse contract requests, identify contract type, region, complexity, and business context, then assign the contract to the most suitable lawyer or contracts specialist. |
Auto generate tasks and assign | Use AI to identify key actions, deadlines, approvals, and obligations from matter or contract information, automatically creating and assigning tasks to the relevant stakeholders. |
Contract risk analysis on contract creation | Use AI to evaluate new contracts for potential legal, commercial, compliance, or operational risks, highlighting high-risk clauses and recommending areas for review. |
Matter risk analysis on matter creation | Use AI to assess new matters for legal, regulatory, financial, or reputational risk, helping teams prioritize resources and escalate high-risk matters early. |
General matter and contract housekeeping | Use AI to monitor and maintain data quality by identifying missing information, outdated records, duplicate entries, incomplete tasks, and opportunities for process improvement. |
Standardize matter and contract naming conventions | Use AI to automatically generate consistent naming formats for matters and contracts based on defined business rules, improving searchability, reporting, and governance. |
Auto assign new matter type | Use AI to analyse new matter details, classify the request, and assign the correct matter type based on its content, context, legal area, and business purpose. |
Creating a workflow
Go to Settings in the left navigation.
Select AI Management from the settings menu.
Click the Create a new workflow. New workflow can be created from the top “Create +“ button too.
Enter a name for your workflow (up to 200 characters) and click Create.
You will be taken directly into the workflow editor where you can begin building your workflow.
Tip: Give your workflow a descriptive name that reflects what it does — for example, "New Contract Summary on Upload" or "Matter Triage – Employment". This makes it easier for your team to identify workflows at a glance
What workflows can I create?
If you're unsure what Workflow Builder can do, click the "What can I do?" prompt within the AI Planner. This option displays the latest available capabilities and supported actions to help you build your workflow more effectively.
Understanding the workflow editor tabs
The workflow editor has three main tabs at the top of the page. Each gives you a different view into your workflow.
Workflow tab
This is where you design your workflow. The canvas shows a visual graph of all the steps (nodes) and the connections between them. You can:
Use the AI Planner panel to generate a workflow from a conversation
Add, move, and delete steps using the AI Planner
Click any step to open its configuration panel on the right
Use the Versions panel to view, manage, and revert to previous versions
Use test mode to emulate your workflow without impacting production data
Use the Publish button in the toolbar to make the current draft version live. A yellow "Unpublished changes" indicator appears whenever you have edits that have not yet been published.
Executions tab
The Executions tab shows a history of every time the workflow has run. For each execution you can see:
When it ran and how long it took
Whether it completed, failed, or is still in progress
A step-by-step breakdown showing how long each step took and its outcome
Use this tab to troubleshoot unexpected behavior — if a run failed, clicking into it shows you exactly which step went wrong and why.
Observability tab
The Observability tab provides aggregate performance data across all workflow executions. It highlights:
Overall execution outcome mix (completed vs. failed)
Median and slowest step durations
Trends over time to help you identify performance regressions
This is useful for monitoring workflows running in production and identifying where to optimize prompts or node configuration.
Building your workflow steps
Every workflow is made up of steps connected in sequence. These nodes represent individual steps the AI will take. The following step types are available:
Trigger step
The Trigger step is the entry point of every workflow. It defines when the workflow is triggered. You can configure one of three trigger types:
Trigger type | When does the workflow run? |
Run Now (Manual) | Triggered manually by a user on demand |
Scheduled (coming soon) | Runs automatically on a recurring schedule you define |
Event-based | Runs automatically when something happens in LawVu (e.g. a matter or contract is created) |
Event-based triggers currently support:
Matter created — fires whenever a new matter is created in your workspace
Contract created — fires whenever a new contract is created in your workspace
Note: A workflow can only have one Trigger node, and it must always be the first step in the graph.
Action step
The Action step gives the AI access to a specific LawVu capability or data source. An Action step actively fetches or writes data from LawVu on the AI's behalf. You can pass additional prompt and instructions to the tool to make it behave in a certain way.
Examples include retrieving matter details, searching contracts, or looking up contact information to enrich the workflow's context.
AI Task step
The AI Task step uses AI to analyze and process information collected from earlier workflow steps. It helps transform raw inputs into structured, useful data that can be used by later steps in the workflow. It can:
Review data
Extract information
Summarize content
Classify or categorize information
Identifies important details, trends, or intent.
Formats information for use in conditions, actions, or outputs.
Skills step
A reusable business capability that delivers a specific business outcome. For example, a Contract Summary skill can analyze a contract and generate a summary tailored to the organization's requirements, without workflow designers needing to recreate the underlying logic, prompts, or actions each time.
Skills are managed independently of workflows and can be reused across multiple workflows. Learn more about custom skills
Condition step
The Condition step branches the workflow based on a condition. The AI evaluates the condition and routes execution down one of two or more paths.
Use Condition nodes to build conditional logic — for example:
If the contract value exceeds a threshold → escalate for legal review
If the matter type is "Employment" → assign to the Employment team
If key dates are missing → create task for the matter manager
Output step
The Output step is the final step in a workflow branch. It defines what the AI produces at the end of that path. There are two output types:
Output type | What it produces |
Text | A text response — for example, a summary, a list of recommendations, or a notification message |
Chart | A visual data chart — useful for scheduled reporting workflows |
Every branch in your workflow must end with an Output node. The workflow editor will flag branches that are missing an Output node when you validate before publishing.
Configuring a step: Config and Advanced tabs
When you click on any step in the canvas, a configuration panel slides open on the right side of the screen. This panel has two tabs:
Config tab
The Config tab contains the core settings for that steptype:
Prompt: Defines the task the AI should perform in this step (the what).
Example: “Create a follow-up task for this matter.”Instructions: Define the rules or constraints for how the task should be carried out (the how).
Example: “Set priority to High. If the task relates to legal work, set the due date to next Friday; otherwise, set it to Monday.”Tool / Trigger settings — depending on the step type, you may see dropdowns to select the event type, tool, or output format.
Using the AI Planner to build a workflow
The AI Planner is a built-in chat assistant that generates a complete workflow graph from a conversation. This is the fastest way to get started if you're unsure how to structure your workflow.
How to use the AI Planner
Open your workflow in the editor (on the Workflow tab).
Click AI Planner in the right-side panel toolbar (the ✨ wand icon).
The AI Planner panel opens with a greeting: "Hi! I'm your workflow assistant. Tell me what you would like to automate."
Type a description of what you want the workflow to do and press Enter (or click the send button).
The AI Planner builds a suggested workflow graph on the canvas and replies with a summary of what it generated.
Review the generated nodes and connections, then ask follow-up questions to refine the result — for example, "Add an approval step before the output" or "Split the summary and the extraction into two separate steps."
Conversation history is preserved. When you reopen a previously planned workflow, the AI Planner loads your previous messages so you can continue refining from where you left off.
Tip: You don't have to accept the AI Planner's output exactly as generated. Treat it as a starting point. If you want to make changes, ask the planner to modify it.
Test mode in Workflow Builder
Test mode lets you safely try out a workflow before turning it on for everyone. It helps you confirm the workflow is working as expected without affecting your live production process.
What Test mode does
When a workflow is in Test mode:
you can validate that the workflow triggers correctly
you can review the steps and outcomes in a controlled way
your wider team or live process is not impacted in the same way as an active workflow
Why use Test mode
Use Test mode when you want to:
check that the workflow logic behaves as expected
confirm conditions, actions, and notifications are configured correctly
troubleshoot issues before enabling the workflow
Suggested process
Build or update your workflow.
Turn on Test mode.
Run through a sample scenario.
Review the results.
Make any needed changes.
When you’re satisfied, switch the workflow to active.
Best practice
We recommend using Test mode whenever you create a new workflow or make significant changes to an existing one.
Dashboard widgets
Two dashboard widgets let you monitor and interact with AI Workflows directly from a matter or contract dashboard, without opening the workflow editor.
Workflows widget
The Workflows widget shows a live list of all AI Workflows configured for your organisation. Use it to check the status of recent runs and trigger manual workflows on demand.
What the widget displays
Each row in the widget shows:
Workflow name — the display name set in the workflow editor.
Trigger icon — indicates how the workflow starts:
Clock icon — scheduled trigger (coming soon)
Bolt icon — event-based trigger (e.g. fires when a contract is created).
Hand icon — manual trigger (must be started by a user).
Trigger detail — a short description of the trigger schedule or event (shown for scheduled and manual triggers; hover the trigger icon for the full tooltip).
Last run status — a coloured dot and status label (e.g. Succeeded, Failed, Running) with the date and time of the last execution. If the workflow has never run, the row shows Never run.
Current step — when a workflow is actively running, the name of the step currently being executed is shown below the last-run line.
Pending approval indicator — a people icon appears when the latest execution is paused waiting for a human approval decision.
Running a workflow manually
Click Run on any row to trigger an immediate execution of that workflow. The button is only active when the workflow has a published version. While the run is in progress, the button shows a spinner and is disabled to prevent duplicate runs.
Note: Only published workflows can be run. If the Run button appears greyed out, check that the workflow has been published in the editor.
Opening the workflow editor
Click the external-link icon (↗) next to any workflow name to open that workflow directly in the editor in a new browser tab.
Workflow Output widget
The Workflow Output widget displays the latest output from a chosen AI Workflow execution, including text summaries, charts, and structured data produced by the workflow's Output nodes.
You can add multiple instances of this widget to a single dashboard — each instance tracks a different workflow.
Selecting a workflow
When you first add the widget (or if no workflow has been selected yet), it shows a list of available workflows. Click a workflow name to view its latest output. The selected workflow is remembered — the widget reopens on the same workflow next time.
To switch to a different workflow, click the back arrow (‹) in the widget header.
Output content types
The widget renders whatever the workflow's Output nodes produce:
Text / Markdown — plain text or formatted markdown (headings, lists, bold, etc.).
Chart — a bar or line chart rendered from structured data returned by the workflow.
JSON — raw structured data displayed as formatted JSON.
Viewing execution history
Click the Executions tab (or the history icon in the widget header) to see a list of all past runs for the selected workflow. Each entry shows the run's status, start time, and duration. Click any entry to view the output from that specific run.
Click the back arrow (‹) to return to the latest output view.
Execution statuses
Status | Meaning |
Succeeded | The workflow completed successfully and output was produced. |
Failed | The workflow encountered an error. See the Executions tab for details. |
Running | The workflow is currently executing. The widget auto-refreshes. |
Cancelled | The execution was cancelled before it completed. |
Cancelling a running execution
If a workflow execution is in progress and can be cancelled, a Cancel button appears in the widget header. Click it to stop the current run.
Auto-refresh
While an execution is in progress, the widget polls for updates automatically. A thin loading bar appears at the top of the widget during a refresh to indicate background activity.
FAQ
Who can create and edit AI Workflows?
AI Workflows can be created and edited by Organisation Admins only. Other user roles can trigger manual workflows if that option is made available in the product but cannot create or modify workflow definitions.
Can I have more than one workflow active at a time?
Yes. You can create as many workflows as you need. Each workflow runs independently.
What happens if a workflow run fails?
Failed runs appear in the Executions tab with a Failed status badge. Click into the run to see which step failed and why. Common causes include misconfigured prompts, missing required data, or a tool being unavailable. Fix the issue, publish the updated workflow, and future runs will use the corrected version.
Is my data used to train AI models?
No. LawVu uses Microsoft Azure OpenAI Service. Your data is not used to train any public models. All processing is transient — no data is stored or retained by the AI service beyond the duration of the request. For full details, see LawVu's Trust Centre.
Can I test a workflow before publishing it?
Use the Run Now trigger type (available on the Trigger step) to trigger a manual execution of the workflow on demand. This lets you test the workflow's behaviour using real data before enabling event-based or scheduled triggers.
How do I know when my workflow ran?
Every execution is recorded in the Executions tab. You can see the timestamp, duration, and outcome of each run. For event-based workflows, each new matter or contract creation that matches the trigger will generate a new execution entry.
Where can I get more help?
Contact LawVu Support via the chat icon at the bottom right of any page in the app.