Moltbot Slack setup and PraisonAI with Ollama

This tutorial covers Moltbot setup through Slack, followed by a PraisonAI alternative that uses Ollama to run models locally. The speaker explains how messages reach a bot on the user's computer, pass to an agent and return to the messaging app. Agents can also use tools such as a browser or command line.

The Moltbot walkthrough installs a gateway service and selects an OpenAI model during quick start. Slack configuration takes much of the guide: creating an app, setting OAuth permissions, enabling Socket Mode and subscribing to bot events. The setup uses a bot token and an app-level token. An optional App Home setting enables direct messages. The speaker tests the web interface, invites the bot into a channel and notes that channel messages require an @mention.

The second setup installs PraisonAI's bot package, exports the Slack tokens and connects to Ollama for local LLM inference. The speaker recommends choosing a model to suit the computer and suggests an 8B model for a typical machine, but gives no precise hardware requirements. The demonstration includes chat replies and a folder listing requested through Slack.

A later example defines a searcher and summarizer in agents.yaml and installs ddgs for web search. The guide also covers MCP tools and Python configuration. The speaker enables command execution and cautions that users should understand the access they grant. Local inference does not make the Slack or web-search workflow offline.