Ensuring Information Consistency Between Website and Phone Channels

A customer calls your business because they could not find a specific detail on your pricing page. If your automated phone system tells them to visit the website for more information, you have created a loop that ends in a lost lead. This occurs when a user falls into the specific question trap because the information available to your support channels is disconnected from the documentation your company produces.

The Problem of Knowledge Drift

Maintaining a consistent set of answers across your website, WhatsApp, phone lines, and physical office branches is a coordination problem. Most businesses try to solve this by training humans on documents or writing static scripts for phone menus. Both methods fail when a product update or a policy change occurs. The website gets updated, but the phone script remains the same for weeks. This phenomenon, known as knowledge drift, creates friction for the customer and extra work for the support team.

Duvi addresses this by using your website as the primary source of truth. When you point the platform at a URL, it crawls the pages and parses uploaded files into a vector store. This creates a single knowledge base. When a customer asks a question on your website chat or calls your phone number, the AI agent pulls from the same verified content. This architecture reduces the contact us friction by ensuring your documentation is accessible on every channel simultaneously.

Grounded Responses and Real Time Actions

This approach prevents the agent from guessing. If a customer asks about a specific warranty clause that does not exist in your uploaded documents, the agent will state it does not know. This is a deliberate design choice: in a sales or support context, a wrong answer is more expensive than an admission of a knowledge gap.

The utility of an agent extends beyond reciting text. Moving beyond static answers requires the agent to do more than point to a file. If a customer on WhatsApp asks for a quote, the agent needs to calculate a value based on user input and record that lead. Duvi agents use tools to bridge this gap. They can call your own HTTP APIs, connect to MCP servers, update Salesforce records, or interact with other external databases.

Channel Persistence and Human Handover

A typical workflow might begin with a phone call. A customer asks if a specific item is in stock. The agent uses an API tool to check your inventory database in real time. If the item is available, the agent asks for the customer's name and contact number to create a lead in Salesforce. If the customer prefers to receive the quote in writing, the agent can continue the conversation via WhatsApp.

Duvi routes replies back to the channel the customer used. If they reached out via the web script, the reply stays there. If they called, the agent speaks the answer. This maintains the context of the interaction without forcing the customer to switch devices or repeat their problem to a new person. For conversations that require a human touch, the system allows for a handover. A live chat can be passed to a team member when the logic determines a person is better suited for the complexity of the request.

Launching an agent does not require an engineering team. You connect your own phone number or WhatsApp business account through providers like Twilio or Meta. Usage is managed through credits. This setup ensures that the same logic and knowledge are applied whether the customer is typing a message at midnight or calling during a lunch break.