v0.3.0
Fuzzbucket v0.3.0: Social Publishing, Workspace Polish, and Steadier AI Runs
Fuzzbucket now carries more of the marketing workflow in one place: generated assets can become scheduled Meta posts, teams can inspect post performance, and the workspace gives clearer feedback while AI runs are in progress.
Fuzzbucket

New
- Agent chat now supports mid-run steering, so you can redirect the agent while it works with your “btw...” or “NO NO STOP” messages.
- You can now select specific models in the workspace for image generation, video generation, and the agent you talk to. We’ll add more models very soon as well.
- Added social publishing workflows for connected Facebook, Instagram, and Threads targets, including connection management, publishing targets, scheduling, and disconnect actions.
- Added a Distribution workspace for managing social posts by platform, with queued work, published posts, lifecycle actions, and post-level performance details.
- Added artifact-to-social scheduling so generated content can move from the workspace into a planned social post with its caption and media context intact.
Improvements
- Refined the workspace canvas, chat panel, zoom controls, artifact controls, and editing overlays so creative work feels smoother and easier to steer.
- Improved image and video generation/editing flows so selected models, source assets, and direct edit instructions are handled more consistently.
- Images now load faster. Fr, that’s it.
Reliability
- Made long-running AI runs more durable by preserving progress, task snapshots, resume/cancel state, and stream events through more edge cases.
- Improved social publishing reliability, especially around scheduled jobs, analytics sync, generated media publishing, and Facebook image uploads.
- Reduced chat startup overhead and tightened usage accounting recovery so sessions are less likely to stall or retry unnecessarily.