We’ve recently launched AI Agents into Beta, alongside our existing open-source app-building and automation tools.
We built Budibase Agents for teams that want to leverage AI within real-world workflows, using their own LLMs, data, and APIs. Because of this, our Agents can be powered by any LLM with an OpenAI-compatible API, including open-source and locally-hosted models. This means you can build Agents that connect to your existing toolstack, within your own environment.
Some more details:
- Agents' behavior is configured using natural language instructions, within existing Budibase Workspaces.
- You explicitly control which data sources, APIs, and automations an agent can access.
- End-users can interact with your agents via Budibase Chat, or using existing chat tools like Slack and Discord.
- Agents can be called from Automations, and vice versa, enabling complex workflows, including interacting with end-user apps for manual human approvals.
The Budibase Agents Beta is available now for all self-hosted and cloud users.
As this is a Beta release, we’re actively looking for input and feedback on how we can improve Agents, in terms of experience and functionality for building using AI within real-life workflows.
Feel free to let us know via our GitHub Discussions: https://github.com/budibase/budibase/discussions
sneg55•1h ago
Curious how you handle tool permissions in practice. Is it just a whitelist, or can you scope down to specific rows/fields?
mjashanks•57m ago
The way to scope down access, would be to use an Automation (our RPA, zapier-like section). You can give an automation parameters and control exactly what it does, then use it as an agent tool.
In future, we'll be looking into user-delegate permissions. I.e. you give your users a role (RBAC) in Budibase, which gives them access to certain resources. Any agent calls will be done under the user's permissions. This is not always desirable - sometimes you want your agent to have more permissions that the user - so this will need to be configurable at some level.