Inferable

Inferable is an open-source platform for creating production-ready AI agents and workflows. It enables developers to build, orchestrate, and manage AI-driven automations using existing codebases, APIs, and data, without needing to learn new frameworks.

Inferable acts as an orchestration layer between your internal systems (APIs, databases, functions) and human intent. You can set up conversational AI agents that interact with your internal tools via chat (Slack, email), APIs, or programmatically.

Yes, Inferable is completely open-source and can be self-hosted on your own infrastructure. This gives you full control over your data, compute, and security.

Inferable's key features include Workflow Versioning to evolve workflows with backward compatibility, Managed State eliminating the need to provision or manage databases, and Observability with end-to-end monitoring and integration with existing observability stacks. It also supports On-premise Execution, allowing workflows to run on your infrastructure without deployment, and offers Security with outbound-only connections requiring no inbound ports. Human-in-the-loop functionality enables pausing workflows for manual approval, and it provides Integration with native support for Slack, email, Zapier, and HTTP APIs.

Inferable ensures data security and privacy by processing all data through AI models with zero data retention, not allowing user-generated content on the platform, and by not selling or sharing data with third parties. Additionally, on-premise execution and outbound-only connections further enhance security.

Yes, Inferable integrates seamlessly with your existing codebases, APIs, and databases. You can use your current programming primitives and control flow.

You can return an 'Interrupt' from your function code, which pauses the workflow until manual approval or intervention is provided. The control plane handles this, not the LLM.

Inferable supports integrations with Slack, Email, Zapier, HTTP APIs, and also offers custom authentication and authorization.

Yes, Inferable offers a free tier to get started. Paid, consumption-based services are also available for larger-scale or production use.

You can get support or technical guidance through Inferable's Documentation, the GitHub repository, or via their Community and support page.

Common use cases for Inferable include conversational interaction with internal APIs and databases, automated processing of alerts such as from Datadog with enriched context, enriching support tickets from systems like Zendesk by pulling context from multiple sources, and replacing custom scripts or internal tools with conversational agents.

Inferable handles workflow state and versioning by ensuring workflow executions maintain version affinity, so updates to a workflow version do not disrupt ongoing executions. Furthermore, all state required for durable workflows is managed by Inferable, is API-driven, and requires no database management.