AI Intelligence Archive

We study how far a simple statistical pipeline can go on univariate time series anomaly detection under a strict selection protocol. The method extracts a small pool of statistics over sliding windows, scores each window with a transductiv…

Defenses against jailbreak attacks on Large Language Models (LLMs) operate at different pipeline stages, such as input modification or output guard, but it remains unclear which defenses to deploy at each stage and how to combine them. Pri…

As deep learning models continue to advance, knowledge tracing models have achieved higher accuracy. However, these gains come at the cost of reduced interpretability, which is crucial for practitioners in educational settings to adopt new…

Most multilingual dysarthria-severity systems either train on a single aetiology-language pair or pool heterogeneous aetiologies into one label space. We test that pooling assumption with four matched HuBERT-base contrastive embedding mode…

A pretrained robot foundation policy may execute most of a long-horizon task yet repeatedly fail at a few critical subtasks. Collecting additional full-task demonstrations for supervised fine-tuning (SFT) requires operators to repeat behav…

Quick disclosure up front, because it matters for how much weight to put on anything below: this was...

Ten parts ago, ClinicLive worked but looked like the template it was born from. This series made the...

Introduction You build a dispatch command centre with ToolJet MCP for a regional broadband...

A US automotive SaaS product had an AI assistant that missed intent and took up to 30 seconds to answer over 120 GB of documents. What we rebuilt, wha

A medical model's benchmark score does not establish that the same conclusion holds under a different evaluation. This study tests whether claims about model ranking, score reliability and screening performance survive changes in cohort, p…

Lovable is nice. You describe a site, an agent writes it, you click publish. It also costs $25–50 a...

Métricas de efetividade de uso de IA — Spark Monitor Resumo da pesquisa sobre como medir...

How to create any sort of pipelines using DSCI and LLM

If you're rolling out Claude Code across a company, at some point you’ll eventually hit the same...

In 2024, AI coding agents (Claude, Cursor, Copilot, etc.) became capable enough to do real work....

How to embed your app in Claude, ChatGPT, Cursor, Copilot, and Gemini via MCP. Strategy, embedding depth, auth, build-vs-buy, and discovery for 2026.

MCP authentication: OAuth 2.1 + PKCE, Dynamic Client Registration, scope design per-resource per-verb per-sensitivity, audit logs, and SOC 2 mapping.

MCP client comparison: Claude, ChatGPT, Cursor, Microsoft Copilot, Gemini, Perplexity. Distribution, auth, audience, and discovery side-by-side.

MCP server cost in 2026: $100K–$300K read-only, $300K–$700K with actions, $1M+ agent-resident, plus $5K–$25K/month to maintain. Why using one is

This is a condensed version. The full write-up — the module decision table, the customized...

Every week, someone tells me our company shouldn't exist. The line is always some version of this:...

Your agent's permissions drift silently. A PR adds an MCP server with write scope, or allow: Bash(*)...

I did not need Jev to beat Claude or Kimi on a benchmark. I needed to know whether I could trust it...

Your SKILL.md is production config. Test it like one. You maintain agent skills — the...

Twice in this series, a client refused to connect to a working MCP server, and both refusals looked...

Hi, I'm Duke from Nexus, I’m proud to share a meaningful milestone in my professional journey: I...

What using a workflow engine is like: the front doors, the reusable pieces, the five claims Archon makes for itself, and the rough edges that come with it.

Like most developers who survived the 2000s, seeing the name LimeWire in 2026 usually triggers a mild...

Autonomous workloads under resource constraints develop independent incentive structures. The AI...

I ran a comparison prompt against Gemini to see how it described the CLI I maintain. It came back...

A claims-processing agent read a file, checked a policy, and issued a €1,180 transfer. The function...

Somewhere right now, an AI agent has a standing API key with more access than the intern you'd never...

Imagine dropping a small stone into a perfectly calm lake. You see the ripples moving outward. It...

For a long time, I used AI in software development for one primary purpose: Write the code. Give it...

GDG Basilicata and the IEEE Student Branch at the University of Basilicata, together with the...

The EU-funded MODELAIR project is developing AI-enabled 3D digital twins designed to predict urban...

What 22 Days of Building AI Systems Taught Me: Grounding, Evals, and Control I started...

My posts here so far have only been about the benchmark I'm working on. This time I want to introduce...

Most of the event-driven architecture I've built started the same way: a synchronous call chain that...

Nvidia shipped two native Rust tracks for writing CUDA kernels — cuda-oxide and cutile-rs. No FFI, no wrapper, compiler-enforced memory safety on the GPU. Here's what it does, what it doesn't, and why the moat conversation just changed.