Function-Calling
PulseAugur coverage of Function-Calling — every cluster mentioning Function-Calling across labs, papers, and developer communities, ranked by signal.
6 day(s) with sentiment data
-
AI Agents: Distinguishing Skills, MCP, Function Calling, and A2A
The article clarifies the distinctions between Agent Skills, Model Context Protocol (MCP), Function Calling, and Agent-to-Agent (A2A) communication in AI development. Agent Skills are described as self-contained folders…
-
LLM Agents Enhanced with Tool Use and Function Calling
Large Language Models (LLMs) can be enhanced with tool use and function calling to overcome their static knowledge limitations and lack of real-time data access. This capability allows LLMs to interact with external API…
-
New Function Hijacking Attacks Threaten Agentic AI Models
Researchers have identified a new security threat called Function Hijacking Attacks (FHA) that targets agentic AI models utilizing function calling capabilities. These attacks manipulate the model's tool selection proce…
-
Prompt Engineering Outpaces Fine-Tuning in Cost-Effectiveness for LLMs
In 2026, prompt engineering is generally more cost-effective and easier to iterate on than fine-tuning for most LLM applications. Advances in cheaper frontier models like DeepSeek V4 Flash, larger context windows, and r…
-
AI Agents vs. Function Calling: Execution Control is Key
This article discusses the distinction between AI agents and function calling, highlighting that while both utilize the same Large Language Models (LLMs) and tools, the key difference lies in the execution control. The …
-
LLM API Integration: Beyond Demos to Robust Production Systems
This article discusses the challenges and best practices for integrating Large Language Models (LLMs) with real-world APIs, moving beyond basic demonstrations to robust production systems. It emphasizes treating LLM API…
-
LLM function-calling tests should focus on schema compliance, not model behavior
This article discusses the importance of contract testing for function-calling schema compliance in LLMs. It differentiates between testing model behavior (e.g., whether it calls a specific tool) and testing contract co…
-
AI agents pose security risks when given tool access; mitigation strategies outlined
Giving AI agents access to tools like function calling, code interpreters, or APIs introduces significant security risks, as models cannot reliably distinguish between instructions and data. This means a prompt injectio…
-
New method predicts AI task difficulty without simulations
Researchers have developed a method to predict the difficulty of AI tasks without needing to run simulations, which can be computationally expensive. This approach analyzes task descriptions to forecast success likeliho…
-
OpenAI chatbot evolution: From text to tools and verifiable actions
This article explores the distinction between a simple chatbot and an instrumental assistant, focusing on the threshold where a language model requires tools for external actions. It argues that tools are only justified…
-
Model Context Protocol (MCP) shifts AI agent interaction from APIs to self-describing tools
The Model Context Protocol (MCP) is a new approach that changes how AI models interact with APIs. Unlike traditional APIs, which require developers to explicitly instruct models on how to use each tool, MCP allows tools…
-
LLM structured output guarantees syntax, not semantics, leading to 'well-formed lies'
Structured output for LLMs, such as JSON schema enforcement, guarantees that the generated text is syntactically correct but does not ensure semantic accuracy. While these tools eliminate parsing errors, they can lead t…
-
Developers prioritize LLM API patterns over prompt engineering for reliability
Developers are shifting focus from prompt engineering to more robust API integration patterns for large language models (LLMs). Key strategies include using structured output like JSON via function calling or schema val…
-
MCP vs Function Calling: Latency and code size benchmarks favor MCP for scaled tool reuse
A data-driven comparison between MCP and Function Calling for tool reuse in AI agents reveals that while MCP incurs a 2ms communication overhead per call, this is negligible for typical agent tasks where LLM inference d…
-
MCP protocol centralizes AI agent tool integration, reducing code duplication
The article introduces MCP (Model Context Protocol), a new system designed to streamline the integration of AI agents with external tools like Jira. Unlike traditional function calling, where tool definitions and execut…
-
Google's A2A protocol finds niche in agent-to-agent communication
Google's Agent2Agent (A2A) protocol, introduced in April 2025, aims to standardize communication between independent AI agents from various vendors and frameworks. Initially met with skepticism due to market saturation …
-
AI Function Calling Tutorial: DeepSeek V4-Pro Integrates with APIs
This tutorial demonstrates how to implement function calling in AI models, enabling them to interact with external APIs. It provides a practical example using the DeepSeek V4-Pro model to call a weather API based on use…
-
Developer builds agentic RAG system from scratch using Python and minsearch
A developer detailed their experience building an agentic RAG system from scratch as part of the LLM Zoomcamp 2026. The process involved creating a retrieval-augmented generation pipeline using Python and a lightweight…
-
LLM function calling explained: How models use tools and avoid errors
This article explains function calling, a key capability for LLMs to interact with external tools and data. It details how models decide which tool to use and with what arguments, moving beyond simple text prediction to…
-
Developer clarifies RAG, Function Calling, MCP, and Semantic Kernel
This article details a developer's journey in understanding and applying Retrieval-Augmented Generation (RAG), Function Calling, MCP, and Semantic Kernel within .NET AI applications. It aims to clarify when to utilize e…