Hello, I'm Rijul. I'm building git-lrc, a micro AI code reviewer that runs on every commit. It's free and source-available on GitHub. Star git-lrc to help more developers discover the project. Do give it a try and share your feedback
Back in the day, when someone mentioned the word framework, it usually referred to developer-oriented frameworks like React, Next.js, Angular, or Django.
Today, that term has expanded.
When people talk about frameworks in AI, they are often referring to agentic frameworks as well.
If you're not familiar with agentic frameworks, let's first understand what they are.
What Is an Agentic Framework?
Just like React is a framework for building web applications, agentic frameworks are software toolkits and libraries that help developers build autonomous AI agents.
These agents can pursue goals, reason, plan, and perform tasks with minimal human intervention.
You can think of an agentic framework as the middle layer between an LLM and a production-ready AI system.
The LLM generates responses.
The agentic framework turns that model into something that can repeatedly reason, take actions, observe results, and continue until a goal is achieved.
Traditional generative AI is reactive.
You give it a prompt, it generates a response, and the interaction ends.
Agentic AI is goal-driven.
An agent can:
- Receive a high-level goal (for example, "Research this topic and write a report")
- Break it into smaller tasks
- Decide which tools or APIs to call
- Observe the results
- Adjust its plan if necessary
- Continue until the goal is completed
Common Capabilities of Agentic Frameworks
Most agentic frameworks provide these capabilities out of the box.
| Capability | Why it matters |
|---|---|
| Orchestration | Controls the flow of reasoning and actions, turning one-shot prompts into multi-step workflows. |
| Tool Integration | Lets agents call APIs, databases, browsers, code interpreters, and more. |
| Memory / State | Maintains short-term context and long-term memory across tasks. |
| Planning & Reflection | Helps agents break down tasks, recover from errors, and improve reliability. |
| Multi-Agent Support | Allows multiple specialized agents to collaborate on complex tasks. |
| Human-in-the-Loop | Lets humans review or approve actions before they are executed. |
| Observability | Provides logging, tracing, evaluation, and debugging for production systems. |
Now that you have an idea of what agentic frameworks are, let's look at some of the major players.
Major Agentic Frameworks
Some of the most popular agentic frameworks today are:
- LangChain + LangGraph
- OpenAI Agents SDK
- Google ADK
LangChain + LangGraph
LangChain is a broad framework for building LLM-powered applications.
LangGraph is the modern layer that extends LangChain into a production-grade agentic framework.
Core Model
You model your agent as a graph.
Nodes
Nodes represent units of work, such as:
- LLM calls
- Tool execution
- Custom functions
Edges
Edges define the control flow between nodes.
State
A shared state object flows through the graph, allowing different nodes to exchange information.
OpenAI Agents SDK
The OpenAI Agents SDK is a lightweight, production-oriented SDK from OpenAI, available in both Python and TypeScript.
It is the successor to OpenAI's earlier experimental Swarm project.
Core Model
- An Agent is an LLM configured with instructions, tools, and optional guardrails.
- Handoffs allow one agent to transfer control to another specialized agent.
- A built-in agent loop continues until the task is completed.
- Support for sandboxed agents.
Google ADK (Agent Development Kit)
Google ADK is Google's open-source, code-first framework for building production AI agents.
It is model-agnostic and deployment-agnostic, but is optimized for Gemini models and the Google Cloud ecosystem.
Core Model
- LLM agents for reasoning and dynamic decision-making.
- Workflow agents for deterministic execution.
Workflow agents include:
- SequentialAgent
- ParallelAgent
- LoopAgent
Wrapping Up
These are some of the major agentic frameworks available today.
As a simple guideline:
- Need maximum control, long-running agents, or complex graphs? → LangGraph
- Want a clean developer experience, especially with OpenAI models? → OpenAI Agents SDK
- Building on Google Cloud or working with Gemini models? → Google ADK
Each framework has its own strengths, so the right choice depends on the kind of AI agent you're building.
That's it for this article, see you on the next one.
AI agents write code fast. They also silently remove logic, change behavior, and introduce bugs -- without telling you. You often find out in production.
git-lrc fixes this. It hooks into git commit and reviews every diff before it lands. 60-second setup. Completely free.
Any feedback or contributors are welcome! It's online, source-available, and ready for anyone to use.
Give it a ⭐ star on Github
Top comments (0)