Case Study · AI Sales Agent
Inside 13 Months Of AI Sales
Operations At K1 Speed Canada.
262,690 customer and AI messages. 44,181 sessions. How the world's largest karting brand runs an always-on sales layer built by DSE Group: guests answered, events booked, and every conversation turned into operational intelligence.
K1 Speed.
K1 Speed is the world's largest indoor go-kart racing operator, with all-electric karts and entertainment centers across four continents. Its Canadian locations serve a constant stream of racers, families, corporate groups, and party bookings. For many of them, the buying journey starts with a question: hours, pricing, group sizes, age requirements, availability.
Entertainment and hospitality: indoor electric kart racing venues.
AI sales and service agent, guided booking agent, and a connected analytics dashboard, built and operated by DSE Group.
Thirteen months of production history, July 2025 through August 2026, reported from aggregated system data.
An entertainment venue's demand does not keep office hours. Racers, families, and event planners decide in the evening and on weekends, and every decision starts as a question: pricing, racing rules, age and height requirements, availability, group packages. Answered by staff alone, those questions compete with the guests standing at the counter, and free-form inquiries arrive without the structure a booking team can act on. The production data confirmed the pattern: 65% of all agent sessions began outside weekday 9 to 5, and one in three started on a weekend.
DSE Group built K1 Speed Canada a connected, three-part system. A General Agent, engineered in the K1 Speed brand voice and loaded with venue-specific knowledge, answers guest questions in natural conversation at any hour. A guided Booking Agent turns event interest into structured, decision-ready records: location, event type, date, guest count, package, and contact details, written into the CRM. And a shared analytics dashboard normalizes every conversation into operational intelligence: 14 tracked topics, 29 keyword signals, and over 9,000 classified intent signals that show operators what customers want, when they need help, and how booking demand develops.
65% of all agent sessions began outside weekday 9 to 5, and one in three started on a weekend. In practical terms: nearly two thirds of AI-assisted customer demand arrived during the periods when a conventional office-hours service model is least equipped to respond immediately. These were real conversations, not deflections; 95% of General Agent sessions contained both customer and AI messages, averaging nearly six exchanges each.
86.5% of Booking Agent sessions entered a structured booking state, and 77% generated a CRM deal record, capturing location, event type, date, guest count, and package in a median of 6.3 minutes. Half of all recorded confirmations occurred outside weekday business hours. The operational meaning: 44,181 customer interactions did not have to begin with an employee answering the same recurring questions, and event interest arrived at the sales team as decision-ready records instead of free-form inquiries.
Nearly 9,000 sessions were classified by topic, with near-complete coverage since June 2026. Four interests, reservations, racing requirements, pricing, and locations, generated roughly two thirds of all classified demand. K1 operators now use that map for website prioritization, location-level staffing, and promotion planning: intelligence that previously evaporated with every closed chat window.
Among sessions entering the defined booking workflow, the recorded in-flow confirmation rate increased from 18.4% in May 2026 to 68.7% in August 2026 as the flow was tuned in production. We report this as an observed operational trend, not a causal product claim; the denominator is monthly Booking Agent sessions, not website visitors.
What these numbers are, and are not.
Every figure above comes from aggregated, read-only analysis of the production databases, with no personal message content involved. Sessions measure conversations, not unique customers. Booking confirmations are recorded in-flow states, reconciled to the workflow rather than to attended events. And the confirmation-rate climb is reported as an observed trend, not a causal claim. We publish what the data supports; when it does not support a claim, we say so. That standard applies to every case study we release.
One deployment inside a measured portfolio.
Across DSE Group's current deployments, AI sales agents resolve 94% of customer conversations entirely on their own, with the remainder handed to human teams with full context attached. Every deployment is tracked, and every number we publish is measured, not promised. Explore what an AI sales agent would look like on your traffic.
What Would Yours Resolve?
Tell us about your business and the questions your team answers all day. We will show you what an AI sales agent changes.
Start The Conversation