Bentowise Weekly: Rethinking How AI Fits Into the Workspace
AI integration has been on the Bentowise roadmap for a while. One question has become increasingly important:
Where should the AI actually live?
Instead of adding another built-in chat panel, we are currently exploring a different approach:
Bentowise provides the workspace. Your AI client does the AI work.
Many users already have an AI client they trust, such as Codex, Claude Code, OpenCode, or another tool capable of working with local files and Skills.
Bentowise Boards are already stored locally inside a Vault, so an external AI client can potentially work with them directly. The missing layer is a Bentowise Skill that teaches the AI how to understand and safely update a Board.
The workflow we are designing is intentionally simple:
Start AI Session → Go to AI Client → Use Bentowise Skill → Check Results
When an AI Session starts, Bentowise will create a restore point and temporarily lock the current Board.
You then switch to your AI client, open the Vault as its work folder, and ask it to use the Bentowise Skill.
For example, the AI could:
- reorganize research into a clearer structure
- summarize notes into new blocks
- build a comparison section
- identify unanswered questions
- add risks or counterarguments
When the task is complete, return to Bentowise and choose Check Results.
Why Lock the Board?
The Board will remain visible, but temporarily read-only while the AI is working.
This avoids Bentowise and the external AI client editing the same file at the same time.
The interface will likely show a simple overlay:
AI Session Active
- Go to your AI client
- Use the Bentowise Skill
- Return when it finishes
Check Results
The goal is to keep the process obvious without exposing unnecessary technical details.
Restore Points Become More Important
Restore Points were introduced in Bentowise V2 as a useful feature on their own, but they also provide an important safety layer for AI editing.
Before an AI Session starts, Bentowise can automatically save the current state.
AI changes can then be reviewed, adjusted, or rolled back instead of being treated as permanent.
Use the AI You Already Trust
We also want to avoid requiring another AI subscription inside Bentowise.
If you prefer Codex, use Codex. If you prefer Claude Code or OpenCode, those should be valid choices too.
The goal is to make Bentowise a workspace that different AI tools can safely work with.
There are still details to solve around Skill installation, validation, recovery, and setup.
But the core interaction is becoming much clearer:
Start the session. Use your AI. Come back. Review the result.
Everything underneath—Skills, work folders, Board locking, and restore points—should support those four steps rather than get in the way.
The first preview of this AI workflow is planned for the next Bentowise release, currently expected in the V2.2.x series.
More details soon.
