Fleetworks saves 100+ engineering hours per month

Building AI Callers for Traditional Industries | Fleetworks x Vapi Case Study

đź’ˇ
Quick Stats
1. Powering 10,000+ AI-driven transportation conversations daily
2. Achieved realistic AI voice quality that satisfies AI-hesitant users
3. Saved 100s of engineering hours

Background

Fleetworks builds AI agents for the transportation industry, automating communications between transportation brokers and drivers through multiple channels including phone calls, emails, and WhatsApp. They focus on streamlining the day-to-day communication needs of transportation brokers.

The Challenge

When Fleetworks started their company in 2023, they found no suitable voice infrastructure solutions available in the market. This forced them to build everything from scratch. However, as the company grew, they realized they were spending excessive time on phone infrastructure plumbing rather than developing industry-specific features their clients needed.

Their particular challenge was compounded by their customer base's unique needs:

"A lot of these transportation brokers and a lot of people in the transportation industry are very old school. They are not used to technology. And so they expect a really realistic experience when it comes to phone calls," explains Quang, CTO and co-founder of Fleetworks.

Why Vapi?

After evaluating multiple offerings, Fleetworks chose Vapi for its:

  • Developer-friendly API-first approach 
  • Customizability for bespoke infrastructure
  • High-touch forward-deployed team

A key technical requirement was the ability to bring their own stack:

  • Custom transcription layer with pre-trained models
  • Ability to plug in custom LLM models
  • Existing infrastructure components

Solution Implementation

A critical challenge for Fleetworks was achieving natural interruption handling - the ability for AI agents to stop mid-sentence when interrupted and continue conversations naturally without losing context. This was especially important for their clientele of traditional transportation brokers who expect human-like phone interactions.

The Vapi team worked closely with Fleetworks to fine-tune this capability, iteratively improving the system's ability to detect and respond to interruptions. Through this collaborative effort, they achieved a level of conversation fluidity that matched their customers' expectations for natural dialogue.

"Early on when we were evaluating different providers, interruption handling was something missing," explains Quang. "This is something that Vapi has really improved over time, and now we see our communications not having many interruptions and flowing pretty smoothly."

The successful implementation enabled Fleetworks to:

  • Integrate their custom models with Vapi's platform
  • Achieve sophisticated interruption handling critical for their industry
  • Focus their development efforts on transportation-specific features

Results & Impact

  • Operational Scale. Successfully processing tens of thousands of automated calls daily through Vapi's infrastructure, enabling Fleetworks to operate at enterprise scale.
  • Development Efficiency. Eliminated all voice infrastructure work, allowing engineering teams to focus entirely on building vertical-specific features that drive client value.
  • Quality Improvements. Achieved such natural interruption handling that is critical for their traditional, technology-conservative clientele.
  • Business Impact. Voice AI through Vapi has become "one of the essential pieces of our offering to our clients," forming a core component of Fleetworks' value proposition in the transportation industry.

Looking Forward

Fleetworks continues to rely on Vapi as an essential piece of their offering to clients. The platform's flexibility and robust infrastructure support their ongoing growth and development of industry-specific features.

For other technical founders considering Vapi, Quang offers this advice: "If they are a tech company that has a lot of opinions on how to run their phone calls, then Vapi is a clear winner.”

Read more