Record at a glance
- Record type
- Portfolio record: scope and delivery, no measured results
- Client
- Business, name withheld
- Region and sector
- Not published; filed under professional services for navigation
- Channel
- WhatsApp Business API
- AI layer
- OpenAI language models (GPT-4 API) with Dialogflow or Amazon Lex for intents and conversation flows
- Business systems
- The client's CRM: lead capture, customer data and workflow automation, synchronised both ways
- Services
- Agentic AI and automation and systems and API integration
- Solutions
- Customer service AI and CRM and workflow automation
- Duration
- Six milestones over roughly four to eight weeks, then 30 days of support
- Published metrics
- None
The problem
Customers expect to message a business on WhatsApp and get a useful answer straight away. Answering every message by hand does not scale, and a simple keyword bot frustrates people as soon as they phrase a question differently. Just as important, conversations that stay inside a chat app never reach the sales pipeline: leads are lost and customer details have to be typed into the CRM again.
What Netbase built
-
Conversational AI
A chatbot that uses OpenAI language models to understand customer questions written in everyday language and reply naturally.
-
Intent and flow management
Dialogflow or Amazon Lex recognise what the customer wants and steer the conversation through defined flows, so the bot stays within the business's use cases.
-
WhatsApp Business API
Real-time messaging through the official API rather than a personal phone.
-
CRM integration
New contacts and leads are captured in the CRM, customer data is kept in sync in both directions, and CRM workflows are triggered from conversations.
-
Tuning
Conversation logs were analysed to refine intents and flows during optimisation.
-
Documentation and knowledge transfer
The client's team received technical documentation and a handover session to run and extend the chatbot.
How the work was delivered
Delivery followed six milestones, each with its own deliverable: design of the conversation objectives, flows and a prototype; the core AI chatbot on OpenAI; the WhatsApp Business API connection; CRM integration and workflow automation with real-time data sync; natural-language tuning, testing and optimisation; then documentation and knowledge transfer. Thirty days of post-launch support followed.
Where people stay in control
This is a description of good practice for the category, not a claim about the client's configuration: an AI chatbot should answer only within approved topics, leave a clear route to a human for anything it cannot resolve, and keep customer data governed by the CRM's own access rules. The documents for this project do not describe a hand-over to staff, so this record does not claim one. Our guide on AI agents versus workflow automation explains where each approach fits.
What the record does not disclose
- The client's name, sector, country or CRM product.
- Conversation volumes, response times, conversion or any other result.
- Prompts, conversation content, customer data or screenshots.
Netbase works with models from several AI vendors and chooses them per project; naming OpenAI here describes this project's stack only.
Planning a similar assistant
The customer service AI solution covers answering, routing and escalation, and CRM and workflow automation covers what happens after the conversation. The models and tools behind them are on our data and AI stack page. See more Netbase work or contact Netbase to review a similar project.
Discuss a project
Netbase JSC helps organizations design, build, modernize, and operate digital products and AI-enabled business systems.+84 937 869 689
91 Nguyen Chi Thanh, Dong Da, Hanoi, Vietnam
Get in touch
Tell us what you want to build, modernize, or operate.