Reviewers see LangChain as a powerful but sometimes heavy framework for building LLM apps. The biggest strengths are broad integrations, easy model switching, solid retriever and tool abstractions, and a strong community that helps teams move fast without wiring everything from scratch. The main complaints are complexity: deep abstractions, a steep learning curve, uneven docs, versioning friction, and harder debugging when too much is hidden. Founders, including the makers of Databox, especially praise LangGraph for stateful, multi-step agent control and reliability.

+112

Summarized with AI

Pros

agentic workflow support (20)

LangGraph framework (11)

model integration (11)

LangSmith observability (7)

RAG workflows (6)

rapid prototyping (6)

scalable AI development (6)

flexible framework (5)

complex data source integration (4)

community support (3)

Pros

Cons

Reviews

All Reviews

Most Informative

Founder Reviews (108)

Other Reviews (7)

•3 reviews

What's great

flexible framework (5)rapid prototyping (6)agentic workflow support (20)model integration (11)complex data source integration (4)RAG workflows (6)

Model integration is really the killer feature for us swapping between providers (we've gone between Anthropic and OpenAI more than once depending on the task) without rewriting the whole pipeline saved real time. RAG workflows and the retriever abstractions are solid enough that we didn't have to hand-roll our own chunking and embedding logic in the early months, which mattered when we were still validating the product idea instead of infra. Community support is underrated too when something breaks, there's almost always a GitHub issue or forum thread that's already been through it.

What needs improvement

The abstraction layers can get in the way once you actually need to know what's happening under the hood debugging a chain five layers deep is harder than debugging a raw API call, by design. And LangGraph Platform, the hosted piece, has gotten more gated than the open framework itself, so it's not always clear where the free tier ends and where you're expected to pay.

vs Alternatives

We looked hard at LlamaIndex first, since it's more purpose-built for retrieval, but LangChain's broader agent and tool-calling surface won out once we realized we needed more than a RAG pipeline we needed the model taking actions, not just answering from documents. We also seriously considered skipping a framework entirely and building directly against the Anthropic and OpenAI SDKs, which is tempting early on because it feels like less overhead, but that broke down the moment we needed to swap models mid-project or add a second agent step. LangChain's abstractions cost you some directness, but they bought us the ability to change our minds about the architecture without a rewrite, which mattered more at our stage than raw performance would have.

Ratings

Ease of use

Reliability

Value for money

Customization

Report

79 views3mo ago

•1 review

What's great

LangChain is fantastic in the sense that it has a huge ecosystem, many integrations, and it helped move the whole LLM app space forward. I respect that part.

What needs improvement

For me, the biggest issue is developer experience.

LangChain often feels too deep, too wrapped, and over-designed. Simple things can require too much code and too much framework-specific knowledge.

I would love to see a much simpler default path, clearer docs, less boilerplate, and more transparency around the actual agent loop.

vs Alternatives

I also considered using simpler agent SDKs, like openai-agents-sdk\claude agent sdk.... and direct model APIs. For many common agent use cases, those feel much lighter and easier to control.

Ratings

Ease of use

Reliability

Value for money

Customization

Report

95 views4mo ago

•33 reviews

What's great

LangChain was basically the default starting point for anyone building LLM apps back in 2022-2023, and tbh it still has the widest integrations of anything out there. Connecting LLMs to vector stores, APIs, memory modules, custom tools, it handles all of that without you having to stitch libraries together yourself.

The honest pain point is the learning curve once you go beyond hello-world. The docs have improved a lot but they can still lag behind new releases, and if you pin versions poorly, upgrades can break things in weird ways. For teams who know what they're doing it saves real dev time. For newcomers, the abstraction can feel like too much magic happening under the hood.

Also worth flagging: parts of LangChain are being sunset in favor of LangGraph, especially for agent use cases. Not a dealbreaker, but something to be aware of before you build something production-critical on the older chains.

Ratings

Ease of use

Reliability

Value for money

Customization

Report

37 views4mo ago

•2 reviews

LangChain is an open-source framework designed to simplify the creation of applications powered by large language models (LLMs). My thoughts on it are that it serves as a powerful but sometimes complex tool that has significantly influenced the development of LLM-powered applications. It excels at connecting LLMs to external data sources and enabling them to interact with their environment.

What's great

complex data source integration (4)

Report

30 views1yr ago

•4 reviews

Langchain is one of those interesting frameworks that really help you build out an agentic flow though at the last minute all sorts of things are gated behind Langraph Platform. For instance to use the Langraph React hook you will have to reimplement their API.

What's great

agentic workflow support (20)

What needs improvement

Langraph platform gating (1)

Report

32 views1yr ago

•4 reviews

LangChain makes it easy to build complex LLM applications with simple, modular components. Its plug-and-play design supports agents, tools, and memory out of the box—ideal for rapid, production-ready AI workflows.

What's great

rapid prototyping (6)modular toolset (3)agent creation (2)

Report

23 views1yr ago

•8 reviews

Documentation can be overwhelming at first but once you get the hang of it, productivity goes way up. The community ecosystem around it is pretty solid too

What's great

community support (3)rapid prototyping (6)

Report

18 views1yr ago

Exstats

•7 reviews

What's great

LangSmith observability (7)LangGraph framework (11)agentic workflow support (20)

I use LangChain and their implementation of DeepAgents in my products, and I can honestly say this is a huge contribution to the AI community!

It makes building capable agents much more practical. Memory, skills, reasoning, and todos well designed and usable in actual products. I know it was inspired by Claude, but the implementation is excellent and under open-source. Well done.

Debugging with LangSmith is a special kind of love haha. Love it! Highly recommend it.

What needs improvement

Do not automatically include all LANGSMITH_* and LANGCHAIN_* env variables in LangSmith traces when tracking is enabled.

Ratings

Ease of use

Reliability

Value for money

Customization

Report

156 views6mo ago

used Langchain to build Super

•8 reviews

Langgraph helped build Super by providing a standardized framework that made the complex LLM/source interactions more manageable through a declarative graph structure. It enabled better parallelization which significantly improved response speed, while also providing full graph state/sequence debugging capabilities that made troubleshooting easier. The migration to Langgraph unlocked new capabilities like human-in-the-loop pauses for clarification and better fallback mechanisms when models fail. Additionally, Langgraph's functional API allowed developers to write more natural, sequential code for complex subgraphs while still maintaining the benefits of the graph framework.

What's great

flexible framework (5)LangGraph framework (11)scalable AI development (6)complex data source integration (4)

Report

177 views1yr ago

used Langchain to build Warestack

•15 reviews

LangGraph's graph-based architecture with nodes and edges was the key on how we coordinate complex AI workflows. We're orchestrating multi-agent conversations where each node handles specific tasks (analysis, decision-making, action execution) and edges manage the flow between them. Memory lets us build operations that maintain context across conversations, while parallel execution nodes handle multiple AI operations simultaneously. The Pydantic formatted outputs ensure type-safe data flow between nodes, and the checkpoint system with PostgreSQL persistence lets us maintain state across these graph executions, so conversations can resume seamlessly even after server restarts.

Report

136 views1yr ago