Omnigent setup with Claude and GPT on Windows

This walkthrough introduces Omnigent as a harness for coordinating multiple AI agents and comparing their answers. The presenter covers installation for Windows users through a Linux environment, credential configuration, and launching the Debbie UI. The description calls this a WSL workflow; the transcript describes logging into a Linux box. Although the description says there is no macOS app, the speaker explicitly says one is available.

Credential options shown include subscriptions, API keys and a Databricks workspace. This is an orchestration setup using provider accounts, rather than a demonstration of how to run models locally. The presenter reports trouble with Codex and switches to an OpenAI API credential. GPT 5.5 returns a 404, and he says an earlier attempt with 5.4 also failed. A second Claude agent provides a fallback, so the session does not establish a reliable Claude versus GPT comparison.

The prompts ask whether AI will replace data engineers, request a script to ingest five days of NOAA weather data for San Francisco, and explore moving from software engineering into data engineering. The presenter reviews differing career forecasts and adds his own views on SQL depth, data modeling and analytics optimization. These are agent responses and the speaker's judgments, rather than verified employment predictions. The weather prompt demonstrates task dispatch, but the transcript does not establish that the generated script ran successfully.