Agently is an open-source Python framework for building generative AI applications and agent backends that hold up in production. Maintained by AgentEra on GitHub, where it has earned more than 1,600 stars, it is actively developed, with the latest code pushed in October 2026, and is released under the Apache 2.0 license.
Agently positions itself as an AI application runtime framework: instead of gluing prompts together, developers define runs with stable contracts. An AgentExecution owns the prompt, strategy, actions, skill bindings, process stream and result views for a single run, while the framework normalizes provider setup, response parsing and action execution behind one request contract. That design lets teams switch between model providers such as GPT, Claude, Gemini, DeepSeek and others without rewriting application code. For longer processes, the event-driven TriggerFlow manages complex working logic with signals, branching and recovery, and model-generated plans can be turned into validated task graphs. The framework installs from PyPI in seconds and targets Python 3.10 and above.
Agently is free and open source under the Apache 2.0 license. You can install it from PyPI with pip and use it in commercial projects under the license terms; the project notes that trademark and official certification rights are not granted by the open-source license and can be arranged separately. There is no hosted tier required to use the framework: you bring your own model API keys and infrastructure.
Agently suits product engineers building assistants, internal copilots, knowledge tools and AI-backed APIs, platform teams that need clear extension points for model providers, tools, MCP servers and observability, and technical leads who want structured contracts and restart-safe workflows instead of fragile prompt glue.