Building AI Agents for Product Teams: A Practical Guide
How we built Milo, our AI Junior Product Manager, and what we learned about setting up agents for business workflows
Scaling Product Management Without Scaling Headcount
In high-growth, multi-product environments, product management can hit a wall. The frameworks and processes that worked at one product and one roadmap often struggle to scale to multiple platforms, varied customer segments, and competing priorities.
As Group Product Manager at TrustFlight, I oversee a portfolio of aviation safety and compliance products used by organizations with thousands of employees. My role is equal parts strategy, execution and systems design — ensuring that our teams operate from consistent frameworks while still moving quickly.
Recently, I’ve been exploring how AI could operate inside those frameworks — not as a novelty, but as a working team member that follows our framework, uses our tools, and executes our processes exactly as designed. That exploration led to Milo, our AI-powered junior product manager.
This post is both a how-to guide and a proof point that when you get the framework right, AI can slot in easily to handle high-volume, low-complexity tasks — freeing human PMs for higher impact work.
The Problem: Product Management Doesn’t Scale
Even the best PMs get bogged down in repetitive work — release notes, “what’s on the roadmap” questions, competitive intel logs, documentation tickets. At TrustFlight, running four platforms meant routine work was crowding out strategic time.
We didn’t just need more capacity — we needed a way to scale the operating model itself. Milo was the answer: an AI agent trained not just on generic PM tasks, but on our processes, our decision criteria, and our data.
The Solution: An AI Agent With Product Context
Most companies experiment with AI by giving their teams ChatGPT access or building generic chatbots. That’s a start — but it’s also where most stop. The real opportunity is when the AI operates inside a structured product management framework, with direct access to your systems, governance, and product data.
Milo isn’t just a chatbot. It can:
Defining Jira tickets
Updating Confluence pages
Generating release notes
Scoring feature requests with our prioritization framework
Maintaining competitive intelligence logs
This works because Milo was designed for specific workflows, not just open-ended tasks.
Step 1: Build a Knowledge-Base
The foundation of any AI agent is comprehensive, structured business, market, and product context. This is where most of the effort goes — and where the long-term value lies.
For Milo, that meant:
Clear prioritization systems (RICE, KANO, business value criteria)
Decision rights and escalation paths
Stakeholder communication patterns tailored for each audience
Consistent product reference pages with features, personas, competitive context, and workflows
A mapped tool ecosystem with authoritative sources and conflict resolution rules
Capturing this knowledge can prove as valuable as building the AI itself. In many organizations — large and small — key product knowledge lives only in people’s heads. Creating Milo forced us to surface, formalize, and structure it into a shared, reusable asset for the entire team.
Step 2: Define Targeted, Workflow-Driven Use Cases
Generic assistants are underwhelming (or fail) because they try to do everything. We designed Milo for specific, high-frequency workflows with clear triggers, steps, and outputs.
Core functions include:
Feature request processing
Competitive intelligence logging
Release notes creation
Monthly stakeholder updates
Stakeholder Q&A
Training agenda creation
Documentation gap detection
For each, we wrote step-by-step workflows with triggers, information gathering logic, quality checks, and output formats .
Step 3: Integrate Into the Toolchain
Milo uses Anthropic’s Model Context Protocol (MCP) to connect to Jira and Confluence — with both read and write capabilities, governed by our existing permission model.
Integration is where the real productivity gains happen — because the agent isn’t just answering questions, it’s maintaining your systems.
Step 4: Build Feedback Loops and Guardrails
We embedded automatic feedback detection, confidence-based escalation, and an improvement log to ensure Milo stays accurate and trustworthy.
Early Results (and Why This Post is Partly Milo’s Work)
We launched Milo last month. Early data suggests:
~75% reduction in workload for repetitive, individual tasks (e.g., defining a ticket, answering a question)
Even higher savings on batched work (e.g., writing a feature list from completed Jira tickets)
These aren’t just efficiency gains. Less time spent on operational overhead means:
Faster release cycles — features move from idea to delivery with fewer bottlenecks
More consistent customer communications — messaging and documentation stay accurate and aligned
Scalable capacity — we can grow product output without growing headcount
These are early numbers, but they confirm an important point: when AI is embedded into a solid framework, measurable gains appear quickly.
In fact — Milo co-authored sections of this post. That’s not a gimmick. It’s the point. Structured, repeatable work is exactly what AI should handle, so I can focus on the bigger picture.
Scaling Across the Organization
While Milo was designed for product management, the underlying model — codify workflows, embed governance, and integrate AI directly into systems — can apply across other functions. For instance:
Customer success could generate tailored onboarding guides.
Marketing could draft consistent, brand-aligned release messaging directly from product updates.
The pattern is the same: when AI works inside well-defined frameworks, you can scale operationallly without diluting quality.
Lessons Learned
Start with knowledge, not AI. Your agent is only as good as the information it has.
Workflows > intelligence. The value comes from codifying best practices into repeatable processes.
Integration drives productivity. If your AI can’t act in your systems, it’s just an expensive search tool.
Feedback loops matter. Build mechanisms for continuous improvement from day one.
Design for safety. Use guardrails so the AI can create freely but only modify with approval (and never delete).
Final Thoughts
AI agents won’t replace product managers — but they will reshape the PM role by absorbing the operational load that slows strategic work.
The real opportunity isn’t just automation. It’s scaling your product management operating model without scaling your headcount — something only possible when you know how to design the frameworks, governance, and tool integrations that let humans and AI work as one team.
And there are other benefits. Creating Milo forced us to surface, formalize, and structure product information into a shared asset for the entire team — a foundation that strengthens the product team long before the AI gets involved.
This isn’t just about AI. It’s about creating the conditions for scale — codifying knowledge, embedding workflows, and integrating technology so the business can grow faster without adding friction.
Aaron Stryd is a Group Product Manager who designs and implements product management frameworks. He leads product management for aviation compliance software at TrustFlight and writes about product management and AI at www.aaronstryd.com
