As we bid farewell to another groundbreaking year in AI development, LangChain presents its highly anticipated State of AI 2024 Report. With the privilege of observing nearly 30,000 new users signing up for LangSmith every month, LangChain offers unparalleled insights into the evolving landscape of AI technology and development practices. This report by Anablock delves into the significant shifts, emerging trends, and changing preferences that are shaping the future of AI applications.
The Changing Face of LLM Providers
OpenAI's Continued Dominance
OpenAI maintains its position as the titan of LLM providers, outpacing its closest competitor by a factor of six in terms of usage among LangSmith users. This dominance underscores the continued trust and reliance on OpenAI's models for a wide range of AI applications.
The Rise of Flexible Deployment Options
A notable trend in 2024 is the emergence of providers offering more flexible deployment options. Ollama and Groq have made significant strides, breaking into the top 5 LLM providers. This shift indicates a growing demand for customizable AI infrastructure and the ability to run models in various environments, from local setups to cloud deployments.
Vector Stores and Retrieval Systems
Consistency in Leadership
In the realm of vector stores and retrieval systems, Chroma and FAISS continue to lead the pack, maintaining their positions as the most popular choices from the previous year. This consistency suggests that these tools have established themselves as reliable and efficient solutions for vector storage and retrieval tasks.
New Entrants in the Top 10
The landscape is becoming more diverse, with Milvus, MongoDB, and Elastic's vector databases entering the top 10 this year. This diversification indicates that developers are exploring a wider range of options to meet their specific needs in vector storage and retrieval, potentially driven by factors such as scalability, performance, or integration capabilities with existing systems.
Development Trends and Language Preferences
Python's Continued Dominance
Python remains the undisputed king of AI development languages, accounting for an impressive 84.7% of LangSmith usage. This dominance reflects Python's robust ecosystem of AI and machine learning libraries, as well as its ease of use for data manipulation and model development.
JavaScript's Growing Influence
While Python leads the pack, JavaScript is making significant inroads in the AI development space. The report notes a threefold increase in JavaScript usage compared to the previous year, now representing 15.3% of LangSmith usage. This growth suggests an increasing interest in AI development for web and Node.js environments, potentially driven by the desire to create more interactive and dynamic AI-powered web applications.
Cross-Platform Observability
An interesting finding is that 15.7% of LangSmith traces come from non-LangChain frameworks. This statistic highlights a growing need for observability tools that can work across different development platforms and frameworks. As the AI development ecosystem becomes more diverse, the ability to monitor and analyze AI applications regardless of the underlying framework becomes increasingly important.
The Evolution of AI Applications
Shift Towards Complex, Agent-Based Systems
One of the most significant trends observed in 2024 is the shift from predominantly retrieval-based workflows to more complex, agent-based AI applications. The report indicates that 43% of LangSmith organizations are now sending LangGraph traces, signifying a substantial adoption of controllable agent frameworks.
Explosion in Tool Calls
The average percentage of traces involving tool calls has seen a dramatic increase, jumping from 0.5% in 2023 to 21.9% in 2024. This staggering growth reflects a move towards more sophisticated, multi-step AI workflows and agentic behavior in AI applications. It suggests that developers are creating AI systems capable of performing more complex tasks, interacting with various tools and APIs, and making autonomous decisions.
The Open-Source Revolution
Democratization of AI
The rise of open-source models is a key highlight of the 2024 report. Providers like Ollama, Mistral, and Hugging Face are at the forefront of this trend, making it easier for developers to run powerful open-source models on their platforms. This shift is democratizing access to advanced AI capabilities, allowing a broader range of developers and organizations to leverage state-of-the-art AI technologies.
Conclusion
This report paints a picture of an AI landscape in rapid evolution. From the continued dominance of established players to the rise of new, flexible solutions, the industry is becoming more diverse and sophisticated. The shift towards complex, agent-based applications and the growing adoption of open-source models are particularly noteworthy trends that are likely to shape the future of AI development.
As we look towards the future, it's clear that the AI industry is moving beyond simple language model interactions to more intricate, multi-step workflows capable of handling complex tasks. The democratization of AI through open-source models and the increasing need for cross-platform observability tools suggest a future where AI development becomes more accessible, customizable, and integrated across various platforms and languages.
