Artificial Intelligence (AI)

How NVIDIA Auto Ontology grounds text-to-SQL in an automatically generated semantic layer, what worked in our test and where it still falls short.

How we moved 310 Confluence pages into a Git-based company wiki that an LLM agent updates, cross-links and checks, with a human reviewing every change.

Copilot Studio loads SharePoint lists whole. Two agent flow tools fix that: one reads internal column names, one fetches only matching items via OData.

Run GLM-4.7-Flash, a 30B MoE coding model, on a 16GB GPU: build llama.cpp for NVIDIA or AMD, serve the Q4 GGUF and connect it to OpenCode.

Build graph-based Microsoft Agent Framework Workflows in Python: a four-step IT ticket pipeline, the DevUI, and conditional edges for urgent tickets.

Microsoft Agent Framework explained: tools, workflows, state and checkpointing, then a Python IT helpdesk agent with Dev UI, FastAPI and Docker.

Connect AI agents to BigQuery via MCP: Google's managed remote server vs. MCP Toolbox 1.16, with setup, guardrails, costs and a 7-question agent test.

RAGFlow 1.0 installed with Docker and tested on a 147-page PDF: setup, OpenAI models, DeepDoc vs. a vision LLM parser, RAM usage and answer accuracy.

How MCP sampling lets a FastMCP server hand its LLM calls to the client: ctx.sample on the server, a sampling handler on the client, with code.

What an MCP server is, how it differs from an API, and how to build one in Python with FastMCP: resources, templates, tools, progress reporting and a client.

Self-host Langfuse with Docker Compose and use it to trace OpenAI calls, group them into sessions, score outputs and version your prompts.

Connect Open WebUI to n8n with a webhook workflow and a Python pipe function, so chat messages trigger n8n workflows. Setup steps and full code.

Why multi-agent LLM systems fail, based on a study of 14 failure modes, and the design practices that make AI agent systems more reliable.

smolagents is Hugging Face's minimal agent library: code agents that act in Python, sandboxed execution, support for any LLM and tips for better agents.

Run DeepSeek-R1-Distill-Qwen-32B on two 24GB GPUs with SGLang: install, model download, Docker setup and OpenAI-compatible API calls in Python.

Self-host Opik, Comet's open-source LLM tracing tool, trace OpenAI, PydanticAI and LangChain calls, track costs and score traces with an LLM judge.

Build a self-hosted AI agent in n8n 2.46: order lookups in Postgres, chat memory, email drafts behind human approval. With screenshots, costs and pitfalls.

Add PostgreSQL-backed memory to a self-made Python AI agent: store every interaction, summarize long histories with GPT-4o and feed context into the prompt.

Extend a framework-free Python AI agent with an AgentState enum, a multi-step reasoning loop and clarifying questions to the user. Full code included.

Build an AI agent in plain Python with GPT-4o function calling: a PostgreSQL tool, a Wikipedia tool, a tool-call loop and PostgreSQL memory.

RAG on Azure, tested: index a 144-page PDF with Azure AI Search, run hybrid search with semantic ranker and answer with citations in Python. With costs.

Azure OpenAI content filters slow down requests and block legitimate business prompts. How they work, and how to set lower thresholds and asynchronous mode.

Build and run Crawl4AI in Docker, crawl pages via its REST API to get markdown, metadata and links, and extract data with an LLM through LiteLLM.

Late chunking embeds the whole document first, then pools token embeddings into chunks. Python code with jina-embeddings-v3 and a similarity comparison.

A hands-on DSPy tutorial: the 8-step workflow, metrics and optimizers, and a summarization example where optimization lifted the score from 42% to 68%.

LibreChat is an open-source, self-hosted ChatGPT alternative. We installed v0.8.8 with Docker and tested agents, MCP, file search, artifacts and web search.

langchain-autotools turns a Python SDK client into LangChain agent tools. Covers setup, CrudControls for picking functions and an Azure Blob Storage example.

Firecrawl turns websites into LLM-ready markdown. Python examples for scraping, crawling, Pydantic-based extraction, LangChain and self-hosting with Docker.

Check a new Azure OpenAI model against your current one before upgrading, using the Prompt Flow evaluation SDK, a GPT-4o judge and Azure AI Studio.

RouteLLM sends each prompt to a cheap or a strong LLM. How it works, how to calibrate the router threshold, and how to run it as an OpenAI-compatible server.

Store embeddings in BigQuery, create them with ML.GENERATE_EMBEDDING, query them with VECTOR_SEARCH and run a full RAG pipeline with Gemini in one SQL query.

Parse PDF, Word and Excel files with GPT-4o vision: render each page as an image, extract the text in Python, for about 0.32 to 0.62 cents per page.

Python examples for sending images to GPT-4o via the OpenAI API to classify, describe and analyze them, plus detail modes and per-image costs.

Open WebUI is a free, self-hosted ChatGPT-style interface. We installed v0.13.0, tested RAG, MCP and web search, and explain the license and pricing.

What Vertex AI Model Garden is and how to use it: Claude 3.5 Sonnet and Llama 3.1 as pay-per-use API services, CLIP on a VM, and Hugging Face models.

Fine-tune GPT-4o mini in the OpenAI dashboard: the JSONL training format, how many examples you need, job settings, costs and calling the model from Python.

Set up the Neo4j LLM Knowledge Graph Builder on AuraDB or Docker, turn a web page into a knowledge graph and query it with GraphRAG in the built-in chat.

Build a text-to-SQL agent for BigQuery in Python: GPT-4o writes the SQL, errors are fed back for self-correction, and RAG picks relevant columns.

Finetune Llama 3 8B on a single GPU with QLoRA, bitsandbytes and TRL: dataset formats, LoRA and training parameters, full Python script and inference.

Extract text and tables from complex PDFs with an open-source vision LLM: render each page as an image and prompt MiniCPM-Llama3-V 2.5. Python code included.

Call OpenAI from PostgreSQL with the pgai extension: create embeddings, summaries and content moderation in plain SQL, plus triggers that embed new rows.

How Kolmogorov-Arnold Networks (KANs) replace MLP weights with learnable splines, what the paper shows on accuracy, and why MLPs aren't done yet.

Llama Guard 2 is Meta's 8B model for checking LLM prompts and responses. The main attack vectors, its 11 risk categories and how to run it.

Let GPT-4 write python-pptx code that builds a PowerPoint deck from your company documents via RAG, using your own corporate template.

What Vertex AI Agent Builder is and what Google calls it today, with a no-code example: index Slack messages and build a Gemini search app.

How Microsoft's BitNet trains LLMs with 1-bit and ternary weights, why it beats post-training quantization, and what it saves in memory and energy.

Build a Confluence chatbot in Python: load pages with PyAirbyte and LangChain's AirbyteLoader, store embeddings in PGVector, answer with GPT-4 Turbo.

HyDE has an LLM write a fake answer document and searches with its embedding instead of the query's. How it works, plus a LlamaIndex example.

GenAI's energy use is growing fast. The IEA and de Vries numbers, plus the techniques that cut it: TPUs, MoE, distillation, MoD, layer pruning and BitNet.

Eleven ways to cut LLM costs, from shorter prompts and semantic caching to quantization, distillation, RAG and chat summaries, plus which one to try first.

OpenAI's v3 embeddings can be shortened thanks to Matryoshka training. Adaptive retrieval uses that for faster vector search: 580 QPS at 99% accuracy.

How RAG works step by step: ingestion, chunking, embeddings, vector search and the LLM answer, and why it is easier to update than a fine-tuned model.

Add a ColBERT reranker to a LlamaIndex RAG pipeline: why semantic search ranks the wrong chunks first, and the Python code to rerank them before the LLM call.

Recursive retrieval in LlamaIndex links PDF tables to pandas query engines, so RAG can answer questions from table data. Step-by-step Python example.

Serve an open-source LLM with vLLM's OpenAI-compatible Docker image and put Caddy in front as a reverse proxy for automatic HTTPS with Let's Encrypt.

Evaluate a RAG pipeline with Ragas: generate a test set from your documents, then score faithfulness, answer relevancy, context recall and context precision.

Fine-tuning, training your own model or RAG: how each way of adding company knowledge to an LLM works, its pros and cons, and why RAG usually wins.