A good AI-assisted run starts with a well-organized workspace. This guide covers what to do before FilepadAI or an external agent gets to work.
Upload your source material
Bring in the documents the work should be grounded in:
- Upload the documents you need, in whatever folder structure makes sense for the work
- Use descriptive filenames (date + topic is a reliable default)
- Remove obvious duplicates
- Prefer final versions over drafts or long email threads
You can add more sources between runs — FilepadAI and connected agents always see the current file tree.
Organize with folders, not conventions
There's no required folder scaffold. Structure the tree the way you'd structure a shared drive: by client, by project, by document type — whatever keeps things findable for the people (and agents) working in it.
If you're sharing the workspace, remember contributors draft into their own space and submit into the shared tree for review (see Reviewing diffs and approvals).
Write down what an external agent should do
If you plan to connect an external agent, write a short note somewhere in the workspace describing:
- What the agent is: name, model, purpose
- What it should do and how it should behave
- What it is explicitly not supposed to do
- The Agent Access scopes it's been granted
Be specific. Vague instructions produce vague behavior.
Curate workspace memory
Workspace memory is for durable context that shouldn't need restating in every conversation:
- Project-specific terminology or naming conventions
- Decisions that have been made and shouldn't be revisited
- Standing preferences (output format, tone, language)
- Facts FilepadAI or connected agents should treat as ground truth
Keep memory small and explicit. A few precise facts beat a long, vague memory dump.
Next: Connecting an agent.