Playbooks in Brazil | Conversational, MCP-Enabled, and AI-Powered

[kristymerriam]

Playbooks have always been a powerhouse for workflow orchestration on the ServiceNow platform, but they've often stayed behind the scenes. In this Developer Passport Brazil episode, Sai Venkatesh Manadava, PM for Playbooks, showed us three major capabilities coming in Brazil that bring playbooks front and center, making them conversational, interoperable with external AI tools, and intelligent enough to handle complex processes on their own.

Quick Navigation

• Conversational Playbooks
• Playbooks as an MCP Tool
• Hybrid Agentic Activities
• What's Next for Playbooks

Conversational Playbooks

Chat has become a critical interface for runtime users in modern service operations. Until now, playbooks have been available in workspace, portals, and mobile, but not in chat. That changes with conversational playbooks.

This new capability exposes playbooks directly in chat, allowing both fulfiller and requester perspectives to access and complete playbook activities through conversation. But it goes beyond just moving playbooks to chat. With conversational playbooks, you can also expose playbook actions as client tools so that users can take action through chat while leveraging the intelligence of ServiceNow's AI Agent Orchestrator and Otto.

How It Works

Creating a conversational playbook is simple. If you already have a playbook that exists, you create a skill for that playbook. For each skill, you provide three things:

• A label (the skill name)
• A description (what the skill does)
• The target assistant (which assistant should expose this playbook)

The label and description should closely match what users would type when expressing their intent in the chat. For example: "Start an incident root cause analysis and remediation by incident number." This makes the playbook discoverable and intuitive in a conversational context.

↑ Return to Table of Contents

Playbooks as an MCP Tool

One of the most exciting capabilities unveiled in this session is playbooks as an MCP (Model Context Protocol) tool. This means you can take any playbook you've built in ServiceNow and expose it to external AI tools like Claude, ChatGPT, or other LLM-powered applications.

Why does this matter? It breaks down silos between your ServiceNow platform and the broader AI ecosystem. Your playbooks become tools that external AI agents can invoke, turning your workflow orchestration into callable capabilities for any AI-powered application.

Current Scope and Future Expansion

In the Brazil release, playbooks as MCP have some restrictions on which activities and fields can be used. However, Sai indicated that the team is actively working on removing these constraints. The future vision is that any playbook you create, whether built with custom activities, out-of-the-box activities, or any automations, will be supported as an MCP tool.

↑ Return to Table of Contents

Hybrid Agentic Activities

The third major feature brings AI agents directly into your playbooks. In any interactive activity within a playbook, you can now configure AI agents to intelligently handle steps in your workflow. This is what Sai calls a hybrid agentic activity, where AI and humans work together.

Two Operational Modes

Collaborative Mode is the starting point if you're new to this feature. In collaborative mode, AI fetches data and shows it to you in the form, but asks for your permission before saving data or taking any subsequent actions like marking an activity complete.

Autonomous Mode lets AI agents work more independently. If you're comfortable with AI taking more initiative, autonomous mode allows agents to fetch data, process it, fill forms, and even complete activities automatically, making your processes significantly more efficient.

Custom AI Agents and Natural Language Instructions

What makes this feature flexible is the ability to select your own custom AI agents. These work alongside the two out-of-the-box agents that ServiceNow provides, giving you control over which agents handle which activities.

Configuration is straightforward. You provide natural language AI agent instructions that tell the system what to do. For example, you might say: "Look through all available data in the parent record and find a relevant knowledge article. Extract the root cause and solution, and fill that data back into this form." The AI agent does the rest.

Real-Time Progress Visibility

When an AI agent is working on an activity, users see a banner indicating that AI is active. A "View Progress" button opens the Otto panel, showing all the tools the AI is executing and a summary of what it's doing. The form fields are blocked during execution to prevent conflicting changes. This transparency builds trust in the automated process.

↑ Return to Table of Contents

What's Next for Playbooks

Sai shared some exciting forward-looking capabilities (subject to change, of course) that the Playbooks team is investing in for future releases:

Conversational Playbooks

The team wants to automatically populate form data based on existing context when users interact with playbooks in chat. Imagine starting a playbook conversation and having key fields pre-filled based on what the system already knows about the situation.

Playbooks as MCP

The goal is to remove all current restrictions. Custom activities, out-of-the-box activities, and any automations you use should all be supported as MCP tools. No limitations.

Hybrid Agentic Activities

Currently, agentic activities work only with form-based activities. The future roadmap extends this to all interactive activities. Want to generate a list intelligently using AI agents? Summarize a knowledge article and show it to the user? That's all coming. The vision is to add intelligence to every type of activity in your playbooks.

↑ Return to Table of Contents

Get Involved

Have questions about conversational playbooks, playbooks as MCP, or agentic activities? Want to share feedback with the Playbooks team? Leave a comment on this post. You can also reach out on LinkedIn or email; the Playbooks team is actively listening and iterating based on your real-world usage.

Thanks to Sai Venkatesh Manadava for joining us on Developer Passport Brazil to dive deep into the future of Playbooks. We can't wait to see what you build!