Tech News
We charge $10k a week to delete AI-generated code
A company replaced a $120k/year low-code platform with AI-assisted 'vibe coding', where AI generates code from JIRA tickets and senior engineers review. Another platform is on track for similar replacement.
sqlite-utils 4.0, now with database schema migrations
This morning I released sqlite-utils 4.0 , the 124th release of that project and the first major version bump since 3.0 in November 2020. In addition to some small but significant breaking changes.
NVIDIA Vera CPU Boosts AI Factory Throughput to Accelerate Agentic Workloads
Agentic systems turn model reasoning into action through multi-step workflows that combine inference, tool use, code execution, retrieval, orchestration.
Build an AI-powered AWS support companion with Amazon Bedrock AgentCore
In this post, you build an AWS Support Companion using Amazon Bedrock AgentCore. The agent uses Strands Agents as the orchestration framework and connects to AWS services through the Model Context Protocol (MCP).
Expanding Managed Agents in Gemini API: background tasks, remote MCP and more
Managed agents feature bundle launch.
GitHub Repos
open-multi-agent/open-multi-agent
TypeScript multi-agent orchestration framework.
# TypeScript# agent framework# agent orchestrationhiyouga/LlamaFactory
Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)
# Python# agent# aistablyai/orca
Orca is the ADE for working with a fleet of parallel agents.
# TypeScript# ade# agent ideResearch Papers
Data Analysis in the Wild: Benchmarking Large Language Models Against Real-World Data Complexities
This paper presents DataGovBench, a benchmark using real government data to test LLMs on complex data analysis tasks, finding that current models still struggle with multi-table reasoning and insight discovery.
Improving LLM-Generated Process Model Quality Through Reinforcement Learning: The Role of Reward Function Design
This paper investigates how to design reward functions for reinforcement learning to improve the quality of BPMN process models generated by LLMs, finding that equal reward weighting works best and that design choices interact with model architecture.
RoboTALES: Learning Reasoning-Guided Robot Policies via Task-Aligned Simulated Futures
Pretrained video generative models are promising backbones for visuomotor control, but their imagined futures often drift from task intent and are not reliably action-conditional. As a result, these models can be difficult to use for planning or policy extraction.