Independent software. Connected by design.Discover the ecosystem

LOCAL AI & AUTOMATION

Local AI and automation on your terms

Make your model a working part of a process, not a separate chat window.

A model connection is the beginning, not the workflow

A useful automation often needs to retrieve data, transform it, call a model, inspect the result and update another system. OAIY brings those operations into a visual workflow while keeping the model or service choice explicit. The inference endpoint can be local or remote; the surrounding application should know which path it is using.

FormLogic can supply the business records and permissions around that workflow. This is useful when the outcome belongs in a team workspace rather than only in a console on one machine.

An illustrative review-and-draft process

A new record enters a FormLogic queue. A configured flow prepares only the fields needed for the task, sends them to a selected model through the available OAIY path and produces a draft for review. An operator decides whether the result should be saved or acted on.

Keeping an approval step is a design choice, not a limitation of the model connection. It provides a place to handle uncertain output, check context and avoid treating generated text as an authoritative business decision.

  • Choose and test the actual inference provider.
  • Limit workflow input to the data the operation needs.
  • Record failures and results in a place the operator can inspect.
  • Use human approval for actions that should not be delegated silently.

Browser, desktop and headless are different operating modes

The browser editor is useful for constructing a flow. Desktop manages supported local processes, models and plugins. CLI and headless paths allow work to run without the editor. Pick the mode that matches the availability the process needs, rather than assuming closing a browser leaves every flow running.

For use from another device, the linked FormLogic account path lets OAIY connect outward over HTTPS. The local computer must remain online. Sleep, restarts and model startup time belong in the operational plan.

Local-first is a configuration, not a universal privacy claim

Local inference can keep model processing on the selected computer. A remote endpoint receives the input needed to perform its task. FormLogic can separately hold records on its host. Read the integration’s encryption boundaries instead of labelling all traffic end-to-end encrypted.

Credentials, allowed service operations and app pairings should be reviewed deliberately. OAIY is a local capability layer, so access to it deserves the same attention as access to the services it controls.

Prove a small workflow end to end

Start with one provider and an output that is easy to check. Measure whether the model responds, how long the complete workflow takes and what happens when the service is unavailable. Only then add records, remote access or device actions.

Compare local and remote options using your actual input size, quality needs and hardware. A configured GPU service may accelerate supported work, but a local endpoint or a model download alone does not establish GPU execution.

Project references: OAIY · FormLogic · Aokie. Their project pages link to the underlying implementation documentation.

BUILD SOMETHING THAT FITS

Find your starting point.

Explore the platform, the local capabilities or the engine. You do not need the entire stack to do something useful.