Kryoverse
Start a project

CRM software · AI development

A support chatbot that knows the customer's account.

A support chatbot built into a CRM product. It answers routine questions and passes the rest to people, chat attached.

A support bot that hands one case to a personAn inbox list where a bot answers the routine rows and routes one row to a person.InboxBotPerson

The problem

A CRM software company had a support team buried in repeat questions. Many had a known answer. They still waited in line behind the problems that needed a real person.

Customers waited longer than they should. Staff spent the day on easy questions and reached the hard ones late. The company wanted help inside the product itself. A separate chat tool would know nothing about the customer.

What we built

  • Chat inside the CRM

    Customers ask for help from inside the product they already use. There is no separate help site.

  • Account in view

    The chatbot sees who is asking and which account they are in. Answers fit that customer, not a generic FAQ.

  • Quick answers to routine questions

    Common questions get a clear reply right away, without waiting in the queue.

  • Handoff with the whole chat

    Anything the chatbot should not answer goes to the support team with the conversation attached. The customer never repeats themselves.

  • Built to fit their platform

    Written in Python, which suited both their CRM and the AI tools.

How AI helps

  • Understands everyday wording

    Customers type the way they talk. The AI works out what they mean, with no menus or keywords.

  • Remembers returning customers

    It uses the customer's profile, so someone coming back does not start from scratch.

  • Hands off instead of guessing

    When a question needs a person, it says so and passes it on. The support team checked its replies on real customer questions before launch.

How it works

  1. 01

    A customer opens the chat inside the CRM and asks for help.

  2. 02

    The chatbot reads the question and looks at the customer's account.

  3. 03

    Routine questions get an answer on the spot.

  4. 04

    Harder ones go to the support team, with the chat attached.

  5. 05

    Someone on the support team picks it up with the full story.

Built with

Backend
PythonFastAPI
AI
OpenAILangChain
Data
PostgreSQLRedis
Infrastructure
Docker

What changed

  • Routine questions get answered without anyone on the team stepping in.
  • The support team spends its day on the cases that need a person.
  • Customers get help inside the product, without leaving to find it.

Next case study

HR software · AI development

AI resume screening inside an HR platform

Ranks every candidate against the job and gives a reason for each rank. Recruiters make the final call.

Read the case study
See all work

Need something like this? Let's build yours.

Start a project