// the find
openimsdk/wiseengage
Open-source Intelligent Customer Service System. An alternative to Intercom, Zendesk, Salesforce Service Cloud etc.
wiseengage is an early-stage Go backend for a self-hosted customer service system, pitched as an alternative to Intercom and Zendesk with LLM-backed answers. What is checked in right now is mostly the storage and protocol layer, so it suits teams who want to build on this base, not anyone expecting a working helpdesk to deploy this week.
The Redis layer is past the stub stage. It has a batch handler with tests, a Lua script test, a sharded Redis manager, and per-user and per-conversation sequence counters with tests. Sequence allocation per conversation is where homegrown chat backends usually break under concurrency, so starting there is the right call. The customerservice protobuf is checked in with a generation script, so the API shape is pinned before any handlers exist. The CI setup (golangci-lint, gosec, CodeQL, typecheck, coverage and e2e workflows) is heavier than the code it currently guards, so the scaffolding should not need retrofitting later.
The README is one line, with no setup steps, configuration reference, architecture note or API docs, so a newcomer cannot tell what runs. In this listing, which is marked truncated, the cmd, internal and config directories and the top-level agent and customer directories contain only placeholders, so nothing here starts a process yet. The topics promise chatbot support for OpenAI and DeepSeek, but no LLM provider client appears in the tree. The storage layer assumes Redis plus what looks like MongoDB (the mgo directory), which means two stateful dependencies before a first request can be served.