osFoundry / Platform
Workflows
Visual builder for chained agents and tools — drag, drop, connect, ship.
What it is
Workflows is osFoundry's visual builder for chaining AI agents and tools. Instead of writing orchestration code, you lay out steps on a canvas, drag in the agents and tools each step needs, and connect them so the output of one feeds the input of the next. Drag, drop, connect, ship — the graph you draw is the pipeline that runs.
Because osFoundry is bring-your-own-key, each step in a workflow runs on the model and credentials you choose. One step can call a local model running on your machine, the next a cloud API, and another a self-hosted endpoint — all in the same chain, all billed to your own keys with no reseller markup on top. The desktop app is local-first, so a workflow's inputs and intermediate results can stay on your device rather than passing through a vendor's servers. Workflows live in the same workspace as your chat, agents, apps, code, knowledge, and data, so a chain can draw on what you already have instead of forcing you to move it somewhere else.
The point of a visual builder is that a multi-step AI process becomes something you can see, change, and hand off. You see where each step sends its output, swap a model on one node without touching the rest, and ship the finished flow as a repeatable thing rather than a one-off prompt. For anyone assembling multi-agent or tool-using pipelines, that turns orchestration from a coding task into a design task — while keeping the keys, the models, and the data under your own control.
Key capabilities
- Drag agents, tools, and data sources onto a canvas and connect them into a chain — no code required to define the flow
- Chain multiple AI models in one workflow: route a step to a local model, another to a cloud API, another to a self-hosted endpoint, each on your own key
- Every step runs on your BYOK credentials — no per-run reseller markup, so you pay providers directly at their rates
- Runs local-first on the desktop: inputs and intermediate results can stay on your device instead of a vendor's servers
- Ship a finished workflow into the same workspace as your chat, apps, code, knowledge, and data — no separate tool to wire up
Use cases
- Operations lead: chain a research agent, a summarizer, and a formatter into one repeatable pipeline, then ship it for the team to reuse
- Developer: route each step of an agent workflow to a different model — local for cheap drafts, a cloud API for the final pass — all on their own keys
- Privacy-conscious analyst: build a workflow that keeps document inputs and intermediate results on-device, sending only the steps that need it to a cloud model
Frequently asked questions
Do I need to write code to build a workflow?
No. You build the flow visually — drag agents and tools onto the canvas and connect them so each step passes its output to the next. The seed capability is drag, drop, connect, ship. The agents and tools themselves may do real work, but defining how they chain together is a visual, no-code step.
Can one workflow use more than one AI model?
Yes. osFoundry is bring-your-own-key across local, cloud, and self-hosted models, so different steps in the same chain can target different models. You might run one step on a local model and another on a cloud API, each on your own key, with no reseller markup added to either.
Where does a workflow run, and does my data leave my machine?
Workflows run in the osFoundry desktop app, which is local-first. Inputs and intermediate results can stay on your device. Any step that calls a cloud or self-hosted model reaches that provider directly on your own key — you control which steps go off-device and which stay local.