Introducing Lyra Agent: The AI Agent for Customer Feedback Analysis

Today we're announcing Lyra Agent, the AI agent that works across Chattermill to run automated analysis on your customer feedback.

Lyra Agent is the CX-trained agent that thinks and acts like a seasoned analyst. It spots problems, identifies opportunities, and proactively reports new insights — continuously monitoring signals across every touchpoint. Query it, analyze with it, delegate to it — the same way you would work with your best analyst.

‍

What led us here

Over the last couple of years, we've seen technological advancements at a scale never seen before. Agentic AI has moved from experimentation to execution, drastically improving productivity and the speed at which work gets done — from automating everyday tasks to building software itself. AI agents are now core to how every team operates, reshaping the pace at which ideas turn into action.

But at Chattermill, we believe doing more, faster, doesn't create a meaningful impact on its own — if anything, it introduces new risks.

When it's easier than ever to act on any idea, jumping into every opportunity makes it harder for Product, CX, and Revenue teams to step back and ask whether this is the most important thing to work on. Speed without direction just gets you to the wrong place sooner. 

Agentic AI might be making execution easier — but knowing what to build and where to invest is, if anything, getting harder.

That decision ultimately comes down to one thing: understanding your customers, identifying blind spots, and prioritizing your next steps based on facts — the type of research every team values most, and almost no one has time to do well.

This is because customer research is fundamentally hard — the feedback data is messy, unstructured, subjective, and full of nuance, buried across millions of conversations, reviews, and surveys.

Agents don't magically solve this. They work well with structured data, but their ability to process unstructured feedback reliably starts to break down at around ten thousand data points — and the average enterprise generates far more than that in feedback every month.

So even the smartest agents, and the humans directing them, need help turning that volume of feedback into something reliable to act on. The good news is that the same technology can give agents the right tools, guardrails, and infrastructure to do exactly that, at scale.

That's exactly what we've spent years building at Chattermill.

What Lyra Agent is 

If agents can help teams execute faster, they should also help them decide better. Lyra Agent was built on that principle. 

Lyra is a CX-trained AI agent for every team whose decisions depend on understanding customers, from Customer Experience and Product to Marketing and Operations.

It continuously monitors your customer feedback, spotting problems, identifying opportunities, and reporting what's new before you go looking for it — monitoring signals across every touchpoint.

Ask it a question, analyze your data with it, or delegate a task to it — the same way you'd work with a seasoned analyst. 

Here are a few examples of what that means in practice.

Ask any question, get analyst-quality answers

Most teams don’t have the bandwidth to keep up with the volume of feedback their customers generate. Lyra Agent is the front door to that intelligence.

Ask it anything about your customer experience — why NPS or CSAT shifted, what's driving support ticket volume this week, which complaints deserve attention first — and it returns structured answers, complete with charts, quantified evidence, supporting customer quotes, and the ability to dig deeper.

Product teams can find out what customers think of a new release in seconds. CX teams can discover what's driving a spike in complaints, or which issue is generating the most tickets. And revenue teams can use it to understand how the last campaign landed with customers, or what competitors' customers are saying this month. Everyone across the business has access to an agent designed specifically for customer experience intelligence.

Unlike general-purpose agents, Lyra Agent doesn't analyze feedback from scratch. Every signal has already been classified, scored, and validated by Chattermill's proprietary AI engine — giving Lyra structured, consistent, and instantly verifiable intelligence to work from. That's why when two people across the business ask the same question, they get the same answer, every time.

Discover what you didn’t know 

One of Lyra Agent's superpowers is telling you something important before you ever think to ask. That's because it was built to be proactive by design.

It works like an AI assistant monitoring your data around the clock. It identifies patterns and trends, notices important events and volume changes, and delivers insights without waiting for a prompt — using context and past patterns to know what actually matters to you and your team.

Each team gets a different set of insights, delivered proactively and curated for its specific product, project, or focus area.

Imagine Product teams automatically flagged on issues and bugs worth escalating, or feature requests gaining traction. CX teams automatically flagged on issues gaining momentum. Marketing automatically flagged on what customers love, and the initiatives worth doubling down on. Lyra Agent monitors context to surface the next most important thing that needs your team's attention.

Access customer intelligence wherever you work 

At Chattermill, we believe Lyra Agent will be the best place to understand your customers. But we also know teams will want that same intelligence inside the agents they're already building into their own workflows.

That's already possible today, through our MCP. Whether you're using Claude, Claude Code, ChatGPT, Notion AI, or Glean (see the full list here), you can connect Chattermill to your agent of choice. 

Chattermill feeds it with quantified customer feedback, so every answer comes grounded in fact — supported by charts, breakdowns, a list of prioritized issues, and direct customer quotes. It's the same verified, structured intelligence behind Lyra Agent, now available in the tools your teams already use. 

See how to connect Chattermill to your AI agent here.

How Lyra differs to other AI agents

There are many general-purpose AI agents out there, most of them generalists, not specialists. Lyra is different — custom-trained to excel at one thing: customer experience intelligence.

The insights engine behind Lyra Agent runs on a proprietary AI model, blending multiple technologies to deliver the highest levels of precision: frontier LLMs for open-ended reasoning, fine-tuned specialist models for precision tasks, and Aspect-Based Sentiment Analysis built in by design. 

That engine classifies, scores, and validates every piece of feedback before an agent ever queries it — so Lyra Agent isn't inferring from raw text, it's retrieving structured, pre-verified intelligence.

The result is deterministic. Query it twice, get the same answer twice — consistent, traceable, and grounded in evidence, which is what makes it reliable and trustworthy to base business decisions on.

Models of this caliber take years to build, and a world-class team to build them. Backed by years of experience working with the world's most forward-looking companies, Chattermill built an agent global enterprises can depend on.

Learn more about why world-leading teams choose Lyra Agent.

Coming next

Our team has spent over a decade advancing the science of natural language processing. That same expertise is now driving what comes next for agentic AI. Early adopters are already uncovering new applications for Lyra Agent, and we believe it will reshape how businesses approach customer experience. 

Future updates will teach Lyra to learn your business context over time, maintain a complete audit trail for compliance and governance, and build a connected memory of every insight your organization has ever produced. The more you use it, the smarter it gets. 

Early access to Lyra Agent opens today. Sign up below to join the waitlist and we'll get you set up to run it against your own customer feedback data.

Please note: we recommend your account admin manage early access requests on behalf of your whole organization.