AI customer service across WhatsApp, email and phone
A customer-communication software company
- Sector
- Software (SaaS)
- Service
- AI process automation
Problem
Support teams were answering customers in WhatsApp, email, social messengers and on the phone, each in a separate tool. Context was lost between channels, urgent or frustrated customers were spotted late, and every answer had to be written from scratch.
Approach
We built a single inbox for WhatsApp, email, Instagram, Messenger, Telegram, web chat and phone. On top of it sit AI features grounded in the conversation, the customer’s history and a knowledge base: reply drafts, thread summaries and triage for urgency, frustration and churn risk. An AI agent answers on its own inside chat flows, with clear rules for handing over to a person, and an in-app assistant can act through tools. An MCP server lets language models such as Claude read and answer the inbox. For voice, we built IVR call flows with text-to-speech, call recording, AI transcription and call summaries.
Outcome
Agents work from one place with the full history of each customer, start from a draft instead of a blank reply and see which conversations need attention first. Routine questions are answered by the AI agent, which hands over to a person by clear rules.
In numbers
- channels in one inbox, phone included
- 7
- integrations with CRMs, shops and work tools
- 25+
Stack
- NestJS
- PostgreSQL
- Elasticsearch
- BullMQ
- Socket.IO
- React
- Claude API
- SIP telephony
- Google Speech and TTS
- Stripe
Recognise a similar problem in your business?
Book a free call and tell us about it. We will explain how we would tackle it and how quickly a first version could be in production.