penguinAI is a privacy-focused alternative to Character.ai that requires no self-hosting. It offers character-based interactions, allowing users to chat with various AI characters. The project is simple to use with no complex setup.

Through hundreds of conversations about AI, the author explores how AI acts as an accelerant for social change, work, and humanity. AI is not just a productivity tool but amplifies societal phenomena, making human traits more precious. The article also touches on policy responses in Pennsylvania.

The article introduces the Agentic CDP, a new type of Customer Data Platform built for the era of AI agents. It argues that traditional CDPs are obsolete because buyers now use agents that make decisions in milliseconds, demanding speed, hyper-personalization, and richer context (Golden Context). The Agentic CDP is embedded in the data foundation, powers Infinity Campaigns (always-on, AI-driven personalization), and is designed from the ground up for agents and humans to work together. Databricks' CustomerLake is presented as an implementation.

The author built a free alternative to Crunchbase using AI agents to research startups, inspired by grokipedia, with an admin dashboard and a feature to report inaccuracies.

In a two-week coding marathon, current AI agents plateau within 24 hours, while top humans keep improving over the full period. The study introduces a repeated sampling reference line for Elo scaling and shows that agents' gains largely match this baseline, while humans exhibit superlinear improvement.

This guide provides a complete walkthrough for deploying the Hermes autonomous AI agent using Docker, automatic HTTPS, and authentication via the Devopness platform. It covers server creation, application setup, persistent storage, deployment pipeline configuration, and testing, enabling the agent to run 24/7 with Slack and GitHub integration.

This tutorial provides a practical guide to using the xFormers toolkit for building fast, memory-efficient Transformer models on GPUs. It covers validating memory-efficient attention against standard attention, benchmarking speed and memory across sequence lengths, causal masking, packed variable-length sequences, grouped-query attention, custom ALiBi biases, and combining these into a trainable GPT-style model with SwiGLU layers and automatic mixed-precision training.

Introducing LifeSciBench, an expert-authored, expert-reviewed benchmark for evaluating how AI systems handle real-world life science research tasks and decisions.

We generated 12 landing pages with Kimi K2.7 Code and Claude Fable 5. Kimi cost 94% less and scored within a few points on every page. Open-source models are not only cheaper but genuinely competitive on quality, and the gap is closing fast.

The Agentic Resource Discovery (ARD) specification provides a discovery layer for AI agents to find tools, skills, and other agents dynamically, rather than relying on pre-installed configurations. Hugging Face has implemented a reference tool on the Hub.

Weaviate Cloud now offers free tiers across its entire product suite, including the managed database, Query Agent, and Engram, with no credit card required and no time limit, enabling users to build prototypes and use them indefinitely.

This video demonstrates how to build and deploy production-grade AI applications using the Claude Code and Cloudflare technology stack, covering key practices from development to operations.

Databricks launches Genie One, an agentic AI coworker that helps business teams orchestrate workflows and automate tasks. It leverages Genie Ontology to understand business context, supports structured and unstructured data, and can take actions across platforms. The announcement also includes Lake TAP architecture, Genie Agents, Genie App Builder, and more, with pay-as-you-go pricing.

Wolfram Language and Mathematica Version 15 launches with built-in AI Assistant, Symbolic Music, enhanced time series, categorical data computation, and many more core features. The article reflects on 38 years of development and the unique role of the Wolfram Language as a precision carrier in the AI era.

Today, we’re announcing a new API with Amazon Bedrock Guardrails. With this API, you can apply individual safeguards, also referred to as safety checks, at any point in your agentic AI applications without creating guardrail resources. In this post, we walk through how the InvokeGuardrailChecks API works and how to use it to build safe, multi-turn agentic AI applications.

Major enterprise platform companies are racing to build tools for a new class of users: artificial intelligence agents. One example of this was apparent today with the latest releases from Databricks Inc. The company unveiled a new architecture – Lake Transactional/Analytical Processing – that enables AI agents to access operational and analytics workloads on a primary copy of data that resides in a data lake.

Transfigure launches αLPHA, an AI that converts 2D images into 3D CAD STEP files, streamlining the creation of physical objects.

cwcode is a Go-based terminal coding agent leveraging DeepSeek V4 Pro, Qwen3.6-27B, and more. It offers file editing, sub-agents, semantic memory, and autonomous recovery. Key features: low cost (~$0.40/hour), high cache hit ratio (>85%), hash-anchored edits, checkpoint/rewind, and no SaaS lock-in.

Welcome to the 32nd edition of Ruby AI News! This edition features a fabled introduction to US government intervention, a concrete playbook for making Ruby and Rails impossible for AI coding agents to overlook, the road to RubyLLM 2.0, and much more.

NVIDIA XR AI is now available in public beta, giving developers a framework for building multimodal AI agents for AR glasses and XR devices. The platform integrates core capabilities for ingesting device signals, connecting to enterprise tools, supporting diverse AI models, and orchestrating agents, with applications already emerging in manufacturing, science, healthcare, design, and immersive learning.

Voice AI startup Bland announced a $50 million funding round to expand research, grow its engineering team, and scale its platform into more regulated industries. Unlike most voice AI systems that handle short scripted tasks, Bland builds its own voice models to manage long, nonlinear, high-stakes calls, such as guiding elderly patients through blood pressure readings. The company now handles over 3.5 million calls per week and has more than 250 enterprise customers. The Series C round was led by Dell Technologies Capital, bringing total funding to over $100 million.

Andrew Ng's new project, Open CoWorker, is an open-source desktop AI agent built on the aisuite library. It can chat, perform deep research, execute tasks on the user's computer, and supports scheduled automations. The agent requires users to bring their own API keys or run local models via Ollama.

Python developer Roman Imankulov used intuition and an AI agent to identify a malicious repository disguised as a recruitment test. The repo contained a backdoor hidden in npm installation scripts, which the AI flagged before any damage could occur. The incident highlights the rise of software supply chain social engineering attacks and the potential of AI for defensive code review.

Coherent broke ground today on an expanded manufacturing building in Sherman, Texas, to scale production of indium phosphide wafers and optical components essential for AI infrastructure. NVIDIA CEO Jensen Huang attended, highlighting the strategic partnership. A $50 million CHIPS Act grant supports the project, which is expected to create over 550 jobs and strengthen domestic semiconductor manufacturing.

Datasette 1.0a34 introduces tools to insert, edit, and delete rows directly within the Datasette interface, inspired by Datasette Agent.

Databricks introduces LTAP architecture to unify transactional and analytical databases for AI agents, removing the divide between operational and analytical systems.

I built a fail-closed execution gate for AI agents using KRONYQL, a forensic ledger for sovereign estates.

Qode is a terminal-based autonomous coding agent for long-running tasks and massive codebase development. It features a rich TUI, multiple LLM provider support, and an intelligent pipeline for coordinating multi-file changes from one-line fixes to cross-cutting transformations spanning hundreds of files. Its Modus Maximus mode aims to generate 50k-80k lines of code in a single shot through a four-phase pipeline with specialist agent profiles. Other capabilities include intent classification, dynamic escalation, four-tier memory, verification pipeline, self-correction, plugin and skill systems, and MCP client support.

Anthropic abruptly paused a planned billing change for the Claude Agent SDK that would have significantly increased costs for heavy users, allowing them to continue using existing generous subscription limits.

After a week-long test, the new Siri with Apple Intelligence degrades Spotlight search, requiring multiple taps for a simple web search, reminiscent of Google AI Overviews.