Introducing AI Agents: A More Intuitive Way To Manage AI Answers

AI Answers is Help Scout’s AI-powered tool that uses your support content and help docs to answer common customer questions, right when they’re asked. Customers get answers immediately, while your team focuses on more complex conversations.

As teams started relying on it for more of their support volume, they needed more visibility and control. They wanted to understand what shaped each reply, make updates with confidence, and keep things consistent as AI expanded across their workflow. We knew that need would only grow, so our team built a solution: AI Agents.

One system behind every AI response

AI Agents are a new way to set up and control how AI Answers responds to customers. Every response is shaped by three things: what the AI knows, how it responds, and where it’s used.

• Knowledge - What the AI knows. Your docs, website content, and Improvements. If a question isn’t fully covered, AI flags the gap and suggests an Improvement, so future answers are more complete.

• Behavior - How the AI responds. The tone and guidelines you set. You can keep replies short and direct, more detailed, or aligned with your brand voice.

• Connections - Where the AI is used. The same agent can power multiple Beacons, so responses stay consistent wherever customers reach out. As AI expands to new channels, that setup carries over.

AI Agents bring these together, so you can define them once and reuse them across your support experience.

Understand every answer and make them better

When AI suggests or sends a reply, you'll be able to clearly see what went into it, including the sources used and how they shaped the response. For example, if an answer references outdated pricing, we’ll surface which source provided the information, you can update the content once or add an improvement, and every future response will reflect that change.

Test responses before they go live

Seeing how a change to a knowledge base article or a new Improvement affects an AI Answers response used to mean publishing it first. With AI Agents, you can test changes before committing to them.

Say you decide to update your billing documentation. You can create the change, simulate questions like “How do I change my plan?”, and see how your agent responds. Not feeling 100% confident in the answer? Make adjustments, test again, and only roll changes out to customers once you’re confident in the results.

Use one agent across your operations, with flexibility where you need it

A single agent can power AI Answers across multiple Beacons, keeping behavior consistent across your support experience.

You can also create different agents for different parts of your business — like a support agent to manage customer requests, a sales agent to speak to prospective customers, or a billing agent to answer questions about pricing. Once configured, each agent can be reused across multiple Beacons, and over time, across additional channels like email.

At the same time, each Beacon can control how knowledge is used. Your sales site might exclude internal documentation, while your logged-in product Beacon includes it. This gives you shared behavior where it matters, with the flexibility to adapt to different products, brands, or use cases.

Your existing setup has been moved into AI Agents automatically, so everything continues to work the same way. Each Beacon now runs AI Answers using its own agent, with the same knowledge and behavior preserved. Over time, you can reuse or consolidate agents to apply the same behavior across multiple Beacons where it makes sense.

Not using AI in Help Scout yet? Visit our AI product page to learn more.