What Is Flowith?

Instead of a single chat thread, Flowith lets you place prompts and outputs as nodes on a canvas and grow many branches in parallel. An orchestrator agent can plan multi-step work, and reusable ‘seeds’ store prompts, files, and context you can drop into future sessions.

How to Use Flowith: Step by Step

1. Create a workspace

Sign in and start a new canvas for your project. Name it after the goal so you can find it later among your other gardens.

2. Understand the node model

Every prompt you send becomes a node. From any node you can spawn child branches, compare two models side by side, or merge results back together.

3. Set up seeds for reuse

Add your brand voice, reference documents, or standing instructions as seeds. Attaching a seed to a node injects that context automatically, saving repetitive setup.

4. Use the orchestrator mode

Switch to orchestrator mode and describe a larger objective. Flowith plans sub-tasks, executes them across branches, and assembles a final answer you can review node by node.

5. Compare models

Branch a prompt across different models to see how each responds, then keep the strongest output as your working branch.

6. Organize the canvas

Group related nodes, add labels, and collapse finished branches so the canvas stays readable as the project grows.

7. Share and export

Publish a read-only view for collaborators or export the branch you need into a document. Comments let teammates discuss specific nodes without derailing the thread.

Tips for Getting the Most Out of Flowith

• Start broad with an orchestrator run, then drill into the branches that matter.
• Store your best prompts as seeds so good results become repeatable.
• Keep one canvas per project to avoid context bleed between unrelated work.
• Use comparison branches before committing to an expensive long-form generation.

Final Thoughts

Flowith removes the friction between an idea and a working result, whether you are automating research, writing code, or orchestrating multi-step AI workflows. Start with a small, concrete task, review every step the agent takes, and expand from there. With the workflow above you can go from installation to a finished, verified result in a single session.