Skip to main content
An initiative is what happens when you commit to acting on an automation opportunity. Fluency bundles everything the agent needs — the workflow map, step-level instructions, required systems, and projected outcomes — and deploys it as an autonomous AI worker inside your existing tools. No new infrastructure. No custom integrations. The agent works in the environment your team already uses.

​
Creating an initiative

1

Open Opportunities and find the process you want to automate

Navigate to the Opportunities module from the left sidebar. Use filters or the search bar to locate the workflow you’ve decided to act on. Click the row to open the full opportunity detail.
2

Review the full workflow map and projected outcome at each step

Study the proposed automated path Fluency has generated. Verify that the agent steps match your understanding of the process and that the projected outcomes are realistic against your operational baseline.
3

Click Commit to convert the opportunity to an initiative

Click the Commit button in the top-right corner of the opportunity detail view. Fluency immediately creates a new initiative and opens it in the Initiatives module.
4

Fluency generates the agent configuration automatically based on the workflow data

The platform constructs the agent’s full configuration from the execution data it has already collected — step-level instructions, decision logic, required system access, and expected outputs are all pre-populated.
5

Review the agent's plan

Examine each restructured workflow step in the initiative detail view. For every step you can see the agent’s intended action, its expected output, and which system or team it interacts with. Make any adjustments before proceeding.
6

Click Deploy to activate the agent

When you’re satisfied with the agent’s plan, click Deploy. The agent begins processing new workflow runs immediately. You can monitor its activity in real time from the Initiatives dashboard.

​
What’s in an initiative

Every initiative contains a complete package of operational and business context:
  • Full workflow map — the original discovered process alongside the restructured automated version
  • Step-by-step agent actions — the specific action the agent takes at each node, written in plain language
  • Required inputs and outputs — what the agent needs to begin each step and what it produces when the step completes
  • Connected systems — every tool the agent accesses, drawn from observed execution data
  • Projected hours saved per month — calculated from run frequency and the time eliminated at each automated step
  • Cycle-level impact summary — the expected reduction in end-to-end cycle time across all runs
  • Business case narrative — a plain-language summary of the initiative’s value, ready to share with stakeholders

​
How deployed agents work

Once an initiative is live, the agent operates autonomously inside your existing tools. Most executions complete end-to-end without any human involvement — the agent reads context, takes action, and advances the workflow just as a human worker would. When the agent encounters a situation outside its configured parameters — an unexpected input, a missing data field, or an ambiguous decision point — it surfaces that specific run as a discrete human task in the Initiatives dashboard rather than guessing or stalling. Your team reviews only the exceptions, not every execution. Agents also monitor their own outputs continuously. If an agent detects that its results are diverging from expected patterns, it flags the anomaly automatically so you can investigate before the issue compounds.

​
Monitoring an initiative

The Initiatives dashboard gives you a real-time view of every active and completed initiative across your organization. From the dashboard you can:
  • See the total number of runs completed, in progress, and flagged for review
  • Drill into any individual run to see the outcome of each step, the time taken, and any notes the agent recorded
  • Review exception tasks in the dedicated Human Tasks queue, where flagged runs are held until a team member resolves them
  • Track performance metrics — cycle time, exception rate, hours saved — as they update with each new run
Metrics accumulate automatically. The longer an initiative runs, the more precise its impact data becomes.

​
Updating or pausing an initiative

You can modify a live initiative at any time without losing historical run data. To update the agent’s configuration, open the initiative detail view, make your changes, and click Save. The updated configuration applies to all runs that begin after you save. To pause an initiative, click Pause from the Initiatives dashboard. The agent stops processing new runs immediately, and all run history is preserved. When you’re ready to resume, click Re-deploy — no reconfiguration required unless you’ve made changes.
Before deploying an initiative to production, use the Initiatives preview mode to review the agent’s proposed actions against a sample of real workflow runs. Preview mode shows you exactly what the agent would do at each step without executing any actions. This ensures the agent’s behavior matches your expectations before it operates autonomously.

​
Next steps

Assistant

Query initiative results, ROI metrics, and workflow performance in plain language — and export structured summaries for ops reviews or board presentations.

Deploying your first automation

Follow the end-to-end guide to committing an opportunity, configuring the agent, running a preview, and going live with confidence.