Introducing CraftCX

Support conversations tell you what customers need. They show where instructions cause confusion, which product tasks take too much effort, and what people keep asking for.

When your team handles those conversations directly, your team learns from your customers throughout the day. As AI takes on more of that work, fewer requests pass through a teammate's hands. You still need a way to understand what customers experience, and what they need.

We built CraftCX to help Support, Product and Engineering teams learn from the conversations their AI agents handle.

Everyone needs accountability

When I introduced an AI support agent to my team in a Support Ops role, I found myself spending hours each day reading its responses and checking its performance. Reviewing the agent was taking time away from the rest of my job. I needed to know customers were getting good support, but I couldn't keep checking every conversation myself.

We already make time to review the work of human support agents. We use QA, customer feedback, peer reviews, and performance reviews to recognise good work and help people improve. AI agents need that same ongoing attention. The team running them still owns the quality of support customers receive.

That experience helped shape CraftCX. I wanted a way to build trust in an AI agent through regular review, without spending hours reading threads every day. CraftCX helps teams see where their agent performs well, find problems that need attention, and check whether a change improves the customer's experience. That gives teams evidence and clear reporting to base their confidence on as they give AI more responsibility.

Start with the question your team needs to answer

For a support leader, that question might be whether customers receive accurate answers and useful handoffs. For a product team, it might be why customers keep asking for help with a task that should be straightforward.

Both need evidence they can inspect.

CraftCX connects quality findings and customer themes to their source conversations. Your team can read the request, assess the finding, and decide what to change.

Review the quality of AI support

An AI agent can answer a question correctly while missing another part of the request. It can give clear instructions that the customer has already tried. A transfer can reach the right team but leave the customer to explain everything again.

CraftCX's Support Quality evaluates resolution accuracy, customer effort, and handoff quality. Findings include the conversation evidence and evaluation reasoning so your team can check the assessment.

Policy Monitoring adds review against the rules your support team relies on. When a finding needs attention, the next step is to inspect the evidence and identify who can correct the problem.

Learn what customers need from the product

Support requests contain product feedback even when customers never submit a feature request.

Consider a hypothetical example. Customers keep asking how to download invoices for several workspaces. The agent gives accurate instructions for downloading them one workspace at a time. Support has answered the question, but the repeated request may point to a product improvement.

Support Intelligence brings recurring customer needs and problems together and provides a unified intelligence signal. A product manager can inspect those examples before deciding whether the pattern deserves further research.

Work with the support tools you already use

CraftCX connects with Intercom, Front, HubSpot, and Gorgias. Your team continues to handle support in its existing helpdesk and uses CraftCX to review quality and investigate customer patterns.

The work can start with one topic. Choose an area your team wants to understand, inspect the conversations behind a finding, and assign a concrete follow-up. After the change, check new conversations to see whether the same problem remains.

That is what we want CraftCX to make easier. Find the issue, understand it well enough to act, and return to the evidence when you assess the result.

Explore CraftCX's integrations to find your support helpdesk and see how to connect it.