🧱 No-Code Rovo Agents + Forge Actions = Real-World AI Workflows

[#1]

One of the most powerful things about Rovo is how quickly you can go from idea to working solution — even if you’re not a developer.

In our team, we explored building no-code agents to support workflows in HR and project time tracking. The goal: create a conversational interface for users to retrieve and analyze structured data from tools like Personio and Moco — all through natural language prompts in Rovo.

⚙️ What We Built

We created a no-code agent in Rovo Studio that:

• Answers org-related questions (e.g. team roles and supervisors)

• Pulls in absence data from Personio

• Connects to Moco to analyze logged hours

• Generates visual summaries — such as charts showing workplace distribution

To do this, we paired the no-code agent with custom Forge actions that acted as connectors to the external systems. These actions fetched the data and formatted it so Rovo could interpret and respond meaningfully.

🧩 Why No-Code + Forge Just Works

Rovo Studio made it easy for non-technical users to:

• Write prompt instructions tailored to specific tasks

• Attach Confluence spaces for context

• Choose actions — including the custom Forge actions built by our devs

And since custom actions can be shared across agents, we only had to build once — and enable many.

On the dev side, Forge gave us the flexibility to:

• Securely connect to external APIs

• Preprocess and filter the data

• Return it in a structure Rovo could use for summaries, decisions, or even charts

This created a strong hand-off between the developer-defined logic and the user-facing no-code agent.

✅ Result: Rovo Agent in Action

Here's a snapshot of the user experience inside Rovo, showing how the agent responds to organizational queries.

🗨️ Sample prompts:

• "Who is the supervisor of Barbara?"

• "What is Irmgard Heinrich's role?"

• "Create chart of people’s working places"

🤖 Responses:

• The agent identifies Barbara Vogt’s supervisor as Irmgard Heinrich.

• It explains Irmgard’s role: Teamlead Sales.

• When asked for a chart, the agent uses a Forge-powered action to retrieve data and generate a bar chart showing the distribution of employees by workplace — complete with a link to the underlying report.

This combines:

• Natural language processing

• External system integration

• AI-powered visualization
  All in a single conversational flow.

💬 Have you tried combining no-code agents with Forge actions?
Which systems would you love to connect?

Let’s share ideas — and maybe some code!

[Talent flame]

What about developers , this looks good for prototyping, what would you do for full fledge application development, what would you suggest for developers like to work on ? 

[Paul Pasler _Seibert_]

As the custom actions have to be provided via a Forge app, there is still work for developers :)

I suggest you begin by learning how develop with Forge and concentrate on the Rovo modules (in case you want to something with AI).

You might start with this video from @Caterina Curti: https://www.youtube.com/watch?v=WwhEq8MFFeE

[Robert Santogrossi]

I am trying to do just this.  Add a Forge hosted function as an action that a Studio build Rovo Agent can use.  Since your post, things have changed.  Now Agents use Skills and despite my efforts to follow the guide here: Action using the read:chat:rovo scope I am unable to get a Studio built agent to use a Forge hosted function.  Any tips from the most current Atlassian Rovo Studio interface?

[Paul Pasler _Seibert_]

Hi @Robert Santogrossi ,

just checked it and I was able to add my custom Rovo action as skill in the Rovo Studio.

Feel free to share your manifest or DM me.

Cheers,
paul

[Robert Santogrossi]

Paul,

Thanks for the reply.  I deployed to production, installed, and even enabled sharing within my org but can't see it show up for the Agent skills.  My org has Rovo enabled and Beta features enabled.  I must have done something wrong.  Any help would be appreciated as this is my preferred workflow rather than duplicating code within multiple Forge built agents.  Here is my manifest:

modules:

  action:

    - key: find-critical-path-action

      name: "Find Critical Path"

      description: "Gets Critical Path and Other Project Info"

      function: find-critical-path

      actionType: GET

      inputs:

        projectKey:

          title: Project ID

          description: "The uniqie ID of the project"

          type: string

          required: true

      outputs:

        success:

          description: Indicates if the analysis was successful

          type: boolean

          nullable: false

        projectKey:

          description: The Jira project key that was analyzed

          type: string

          nullable: false

        totalAnticipatedFunds:

          description: Sum of all anticipated funds across all issues

          type: number

          nullable: false

        totalFundsSpent:

          description: Sum of all funds spent across all issues

          type: number

          nullable: false

        milestones:

          description: Array of milestone durations with completion timelines (JSON string)

          type: string

          nullable: false

        criticalPath:

          description: The critical path with longest duration from root to end (JSON string)

          type: string

          nullable: false

        alternativePaths:

          description: Top 5 alternative paths for comparison (JSON string)

          type: string

          nullable: false

        allIssues:

          description: Complete list of all project issues with dependencies and metadata (JSON string)

          type: string

          nullable: false

        projectTotals:

          description: Aggregated project metrics (time, budget, status breakdown) (JSON string)

          type: string

          nullable: false

        summary:

          description: Analysis summary with counts and statistics (JSON string)

          type: string

          nullable: false

  function:

    - key: find-critical-path

      handler: index.handler

  automation:actionProvider:

    - key: find-critical-path-action-provider

      actions:

        - find-critical-path-action

app:

  runtime:

    name: nodejs20.x

  id: ari:cloud:ecosystem::app/384fd278-e10b-4570-b456-93d995e39293

permissions:

  scopes:

    - read:jira-work

    - read:chat:rovo

[Robert Santogrossi]

@Paul Pasler _Seibert_  Looking into this more, the following questions come to mind:

1. Am I allowed to have outputs?  The Rovo->Action  Action page does not list it as a property but the Automation->Action Action page does.

2. Do I need the automation:actionProvider: property?  It is not mentioned in the Rovo action page but is mentioned in the Automation Action page.

3. Although it doesn't seem likely, do I need to request and enable this Rovo Dev Automation beta feature: