Introduction
A ticket gets marked resolved, but the work isn't actually finished. A refund still needs processing. An account record still needs updating, usually in more than one place: the billing tool, the CRM, and a couple of other systems that all need to agree on what happened. Someone still has to do that follow-through work, and right now it's almost always a person doing it by hand, ticket after ticket.
Ballet launched publicly today to close that gap. It's built by Daniel Kimber, who is also CEO of Brainfish, and Ajain Vivek. We think it's directly relevant to how support and CX teams work, so we're introducing it here.
At Brainfish, we spend most of our time on one layer of the CX stack: knowledge, meaning what a team and its AI agents actually know. Ballet operates on the other half of the problem, doing something with that knowledge once a ticket comes in.
TL;DR
Ballet is an agentic orchestration platform built for ops teams, not engineering teams. Describe a workflow in plain English, and Ballet builds and runs it end to end, the same way every time.
Founder Daniel Kimber built it because teams doing this work today are stuck between legacy automation tools that break the moment a step needs judgment, AI chat tools that can't be trusted to run a live system write, and "build it yourself" efforts that turn into a graveyard of half-maintained internal agents.
For support specifically, the clearest proof point so far: a mid-market B2B SaaS company is auto-resolving 40% of support volume through a Ballet workflow, not deflecting it, resolving it, with a human in the loop where it counts, for roughly $120K in annual savings and 2 to 3 FTEs of freed-up capacity.
What Ballet Is
Ballet lets operations teams automate their hardest business problems in minutes. Describe a workflow in plain English, and Ballet builds and runs it end to end. Its category is agentic orchestration, built for ops teams, not engineering teams, and that distinction matters. Most automation tooling assumes someone with engineering skills is setting it up and maintaining it. Ballet is built to be described, adjusted, and owned by the person actually doing the operational work, including the people running a support queue.
Here's founder Daniel Kimber on why it exists:
"Every ops team I talked to over the past year had the same problem. The market handed them three bad options. Legacy automation like Workato or n8n breaks the moment a step needs judgment. AI chat tools like Claude and Cursor can't be trusted to run a live Salesforce write. And 'we'll just build it ourselves' turns into a graveyard of half-maintained internal agents. Ballet is the fourth option: orchestration built specifically for the ops folks who need this to actually work in production."
Laid out side by side, the three options teams already had:
Ballet is built specifically for ops teams who need a workflow to run reliably in production, without hitting any of the three breaking points above.
Daniel's quote is grounded in twelve months of research and hundreds of customer conversations. Three patterns came up consistently:
- Ops professionals describing themselves as human middleware. The connective tissue manually moving information between systems that don't talk to each other.
- Every team building its own siloed AI agents. None of them coordinated, none of them able to hand work off to each other.
- A universal demand for graduated control. Teams wanted to earn autonomy gradually, letting a workflow prove itself before trusting it with more, rather than an all-or-nothing switch.
Support teams know this trifecta firsthand: ticket and account data locked in systems that don't talk to each other, a bot that can't see what the CRM already knows, and a queue nobody's ready to hand over completely until it's earned that trust. That's exactly where Ballet's clearest results are starting to show up.
Why This Matters Now
Coordinating AI agents matters more now than it did even a year ago. Most companies are moving fast to deploy agents across the business, and very few have built anything to manage how those agents actually work together.
Gartner projects the average Fortune 500 company will run more than 150,000 AI agents by 2028, up from fewer than 15 in 2025. That's not more employees using AI tools. It's thousands of distinct, purpose-built agents, most of them stood up independently by different teams solving different problems, with nothing coordinating how they interact.
At the same time, research from MIT's NANDA initiative found that 95% of enterprise GenAI pilots deliver zero measurable P&L impact: no real, trackable effect on revenue or cost. Most pilots never get past the demo stage into something that changes a real number.
Together, these numbers point to the same problem. Companies are accumulating agents faster than they can prove any of them are working. Support teams feel this first, because support is already the department playing the missing coordination role by hand, one ticket at a time.
The CX Use Case
A support agent solves the customer's problem, marks the ticket resolved, and moves to the next one. That's the moment most teams stop measuring, but it's not the moment the work stops. A refund still has to get processed. An account record still has to get updated, and not just in one system: in the billing tool, the CRM, and often two or three other places that all need to agree on what happened. Someone still has to do that follow-through, and today it's almost always a person doing it by hand, ticket after ticket.
That's the exact slice of work our early support customer handed off. A mid-market B2B SaaS company is now auto-resolving 40% of its support volume through a Ballet workflow. The distinction matters: resolved, not deflected. The customer's actual problem gets handled, not routed around, with a human still in the loop wherever it counts.
What that frees up isn't just time, it's a different kind of job. As routine execution gets automated, the work left for a support team shifts from doing the resolution to designing how resolution happens. Kimber is speaking on this directly in a talk called "The Rise of the Support Engineer" at Support Driven Expo in Chicago, August 24 to 26.
How Ballet Fits Next to Brainfish
Brainfish and Ballet solve two different halves of the same problem, which makes sense once you know they share a founder in Daniel Kimber. The CX stack splits cleanly into two layers.
The knowledge layer is what a team and its AI agents know: the help center, the SOPs, the product docs, kept current. That's Brainfish.
The execution layer is what happens once a ticket, question, or trigger event actually comes in, and everything that has to happen after it's marked resolved. That's Ballet.
Great execution on top of a stale or fragmented knowledge base still gives a customer the wrong answer. A perfect knowledge base with no execution layer still leaves a human doing the after-ticket work by hand. Built together, the two close both gaps at once.
The Result
A mid-market B2B SaaS company is auto-resolving 40% of its support volume through a Ballet workflow, saving roughly $120K a year and freeing up the equivalent of 2 to 3 full-time roles.
Where This Goes Next
Execution and knowledge are two halves of the same problem for CX teams. An agent, human or AI, is only as good as what it knows, and only as fast as what it can actually do without a person running five systems by hand after the ticket's already "resolved." Ballet is a new answer to the execution half. If you're already working on the knowledge half with Brainfish, it's worth understanding how the two connect.
Get a Personalized Workflow Review
Curious what this looks like for your own support queue? Ballet is offering workflow reviews right now: a working session to map one of your team's processes and pinpoint where an agentic workflow could take over the parts that eat up the most time after a ticket closes.
Daniel is a product and customer experience leader with over a decade of experience solving user experience challenges at scale. As CEO of Brainfish, he is redefining how users interact with technology - championing a new era of proactive, AI-driven support that anticipates user needs before they arise