*Alex Perevosnikov/Product developer
← Selected work
02 / OMNICHANNEL CUSTOMER SUPPORT

Support that learns.

A realtime support aggregator that brings every channel together, automates replies, and learns from the operators who use it.

MY ROLESolo product developerProduct definition → UX/UI → full-stack development → messaging integrations → AI feedback loop
Realtime messagingWhatsAppTelegramWebsite widgetsAI & knowledge base
THE CHALLENGE

Support arrives through different channels, but the answers should stay consistent. I built a shared workspace where AI handles routine questions and operators can intervene, correct it, and improve the guidance it uses next time.

Product & experience

Designed the operator workflow across a unified inbox, conversation board, customer context, analytics, and AI settings.

Realtime engineering

Connected messaging channels and website widgets into a shared conversation system with live updates and operator handoff.

AI & continuous feedback

Built instruction- and knowledge-based replies, plus an in-chat teaching workflow that turns operator feedback into reusable guidance.

Explore views Click to switch

WhatsApp, Telegram, and website conversations in one live operator workspace.

01 / 06
THE CORE AI FEATURE

Every correction
becomes useful guidance.

Operators teach the AI in the context of a real conversation. The improvement lives in the instructions and learned knowledge used for future replies.

  1. 01

    Answer

    The AI responds using instructions, conversation context, and the knowledge base.

  2. 02

    Correct

    An operator selects a reply and explains what is wrong or missing.

  3. 03

    Learn

    The feedback becomes guidance that can be checked and applied.

  4. 04

    Reuse

    Future answers draw on that guidance, keeping human expertise in the loop.

What I built

MY ROLE IN THE PRODUCT
01 / FULL-STACK & INTEGRATIONS

One conversation workspace

Connected WhatsApp, Telegram, and website widgets to a realtime inbox. Operators can review context, send replies, and take over from the AI without changing tools.

02 / AI WORKFLOW

Answers grounded in the business

Built automatic replies around support instructions and a knowledge base. Teams control the guidance and response mode, with human handoff when a conversation needs an operator.

03 / FEEDBACK & KNOWLEDGE

Operators teach the assistant

Made it possible to select a reply, explain what should change, and turn the correction into learned guidance. Future replies use that guidance alongside the knowledge base.

04 / PRODUCT & OPERATIONS

Visibility beyond the inbox

Added conversation analytics, response-time views, and operator workload reporting. A Kanban view gives the team another way to organize active support work.

LET’S WORK TOGETHER

Have a project in mind?

info@alexperev.com