AI Intelligence Archive

TL;DR: We built an agent that investigates card fraud on a TigerGraph knowledge graph. It gathers...

An agent endpoint returns HTTP 200 in 99.95% of requests. The service is “up.” Meanwhile, some...

We're excited to team up with Kaggle for a brand new challenge! Running through October 11, the...

Swap a cheap model in behind an expensive one and the obvious way to check it is shadow traffic: send...

People keep asking me how you actually know an agent got better. I collected my answers to the most...

When you sit down to solve a complex puzzle or plan three moves ahead in chess, do you narrate every...

The fine-tune took an hour and a half on a single consumer GPU. Everything that made that hour and a...

Ten smaller LLMs played a chess round robin live: 30 of 37 games ended in repetition. Data, lessons, and the 'kick' rule for slow thinking models.

I once watched a 47-minute pipeline run die because of a typo in a date field. The researcher agent...

LiteLLM is an open-source AI gateway that provides a unified, OpenAI-compatible API for over 100...

Langflow is an open-source, low-code visual framework for building artificial intelligence (AI)...

A practical build note on capture, human review, replay, and knowing what a passing test actually proves.

Ten places I slow down, and the 95% of the diff I leave to the pipeline. Most of a diff deserves a glance. A small part deserves an hour.

Are you the kind of person who enjoys a good philosophical conversation but would rather not have to...

Let's begin with a very fundamental question whose answer is not so obvious. How to design software...

Kavach: building a fraud investigator that knows when a risk score is lying TigerGraph x...

When an LLM application starts handling real users, calling a model API is usually easy. The problems...

Apartments.com has integrated its listing experience with Google Gemini and ChatGPT, allowing people...

When people say we need an open standard for agent identity, I usually agree. The harder question is...

Representation Autoencoders (RAEs) enable diffusion models to operate in the feature spaces of pretrained visual encoders. However, many off-the-shelf encoders are not optimized for faithful reconstruction, discarding fine-grained visual d…

Our results show that contrastive learning outperforms a classification-based approach to authorship verification under the tested settings. We identify loss function, batch size, training duration, pre-trained model, input context length,…

A hands-on experience report on a structured, repeatable, AI-assisted development approach We...

Artificial intelligence offers an unprecedented opportunity to augment human capabilities, yet progress at the frontier has focused primarily on advancing model capabilities. We introduce StudentBench, a suite of AI teaching evaluations an…

Social navigation typically assumes a specified goal and focuses on reaching it while respecting social conventions, whereas robot group joining requires predicting where to join based on the group's real-time activity and formation. This …

We introduce Sharper Transductive Local Complexity (STLC), a localized complexity method for transductive learning under uniform sampling without replacement. The construction starts from a Bernstein-type concentration inequality for the s…

Fraud investigation is rarely a single-query problem. A suspicious transaction may look normal in...

Large language models (LLMs) are increasingly used in coding tasks, but their ability to reason about code execution remains unclear. Existing repository-level QA benchmarks mainly evaluate static code understanding and often rely on LLM-b…

We study the nonequilibrium dynamics of a minimal recurrent transformer with $N$ normalized tokens, $Q=K=I$, and a negative value map $V=-I$. Similarity-based attention selects nearby representations, while the negative value map drives to…

Reordering a set of mathematical rules without changing its meaning should preserve the correct answer, but must a model's internal representations stay invariant too? We investigate this question using synthetic multi-step function-compos…

We study minimal-norm interpolation and $\ell_2$-regularized logistic-loss minimization for binary classification by univariate two-layer ReLU networks. We give complete geometric characterizations of the optimal classifiers in function sp…

Everyone eventually wants to build their LLM eval set from real traffic. It's the right instinct —...

This work addresses continuous depression-severity score prediction from clinical interview transcripts under data scarcity. We propose a sequential low-rank adaptation (LoRA) protocol for cross-scale transfer: a Qwen3 backbone with a boun…

Personalizing exoskeleton assistance across operating conditions is constrained by the time and physical effort required to collect user feedback. We examined whether a user's preference landscape varies smoothly across operating condition…

Recursive state estimation often executes approximate numerical solutions inside a feedback loop, where highly accurate local steps do not guarantee better overall results. For a fixed linear Kalman model, we characterize when a correction…

Live replay of a dead-letter message is a control-plane act. Free inference should not own it. A...

Building an agentic fraud investigator on a 590,742-transaction TigerGraph, what a 0.963 AUC actually measured, and the number we had to withdraw.

Recent advances in large language models (LLMs) have enabled agents to tackle long-horizon tasks across diverse environments. To further improve agent performance, existing language world models typically predict environment observations, …

Latent world models that integrate a flow in a frozen self supervised latent space train stably and cheaply, yet silently lose the property manipulation depends on most: motion. The pretrained flow never moves the manipulated object; retra…

Vector Symbolic Algebras project data structures into a hyperdimensional vector space through the application of their vector algebras to randomly generated atomic vector symbols and fractional power encodings of real-valued data. Embeddin…

Collective responses depend on individual differences, contact opportunities, and accumulated experience. Learning their dynamics from aggregate counts requires connecting a population's response distribution to both current observations a…