Reducing wasted field service visits by automating pre site technical checklists
Reduce wasted truck rolls by using AI agents to conduct technical pre site checks grounded in your existing documentation and manuals.
A technician pulls into a driveway and walks to the back door. They realize the installation cannot proceed because the homeowner lacks a specific 240V outlet. The company pays for the truck roll while the technician loses an hour. This sequence repeats across service businesses because human dispatchers lack the time to walk every customer through a technical checklist before the van leaves the warehouse.
Duvi handles this by anchoring a conversation in the actual installation manuals and site requirements documentation. When a customer schedules a visit, the business can trigger an automated reach out via WhatsApp or a phone call. The agent asks specific questions derived from the business's uploaded files rather than just confirming the appointment time. This process is essential for reducing human triage time by enforcing data collection in automated conversations.
If a customer says their basement floor is unfinished, the agent checks the vector store to see if that affects the mounting hardware required. If the documentation specifies a concrete requirement, the agent can use an HTTP API to tag the service order in Salesforce. This alerts the human dispatcher to pack different anchors before the technician departs. Enabling the agent to modify records prevents the read only tax where agents become bottlenecks by failing to update systems.
Voice calls provide a different layer of verification for these scenarios. A customer might struggle to describe a part over text, but a phone call allows the agent to listen for descriptions or guide the customer through a physical check in real time. Duvi uses the same knowledge base for both voice and WhatsApp, so the technical constraints remain identical across both channels. This precision is necessary because RAG is insufficient for transactional customer support that requires specific logical outcomes.
The implementation requires no engineering work. The business crawls its site and uploads PDF manuals into Duvi Studio. The builder writes the prompt based on these sources. When a customer interacts with the agent, the responses are grounded in these specific facts. If the conversation hits a point where the agent cannot find a specific answer in the documentation, it can hand the conversation over to a human manager. This structured approach helps in solving technical compatibility friction before the technician even starts the engine.
Completed conversations are analyzed to show exactly where customers get stuck in the pre qualification process. If multiple people fail to understand the requirement for a specific circuit breaker, the business can update the documentation on its website. Duvi will crawl the updated page and use the new information in its next conversation. Evolving the agent logic this way is a core part of closing sales in the chat by moving beyond the FAQ.
This automation reduces the number of dry runs technicians make. It ensures that when a person arrives on site, the environment is ready for the work to begin. Usage is tracked via credits, which allows the business to scale the volume of pre site checks during peak seasons without hiring temporary staff. Connecting a Twilio or Meta account ensures the communication comes from the business's own recognized phone number to maintain trust with the homeowner.