Tools

Tools

Using DSPy to Evaluate and Improve Datasette Agent's SQL System Prompts

An afternoon research task explored whether DSPy can help refine the system prompts Datasette Agent uses when generating read‑only SQL queries.

3 min read

Tools

Vercel AI Gateway rules: dynamic model routing against suppressed token loss

Vercel introduced the AI Gateway Rules feature, which can rewrite model routes in real time (for example anthropic/claude-fable-5 -> anthropic/claude-opus-5), thereby addressing production outages caused by suddenly deprecated models.

1 min read

Tools

Working with coding agents: understand the code so you can participate

At a recent AIE event Geoffrey Litt argued that people collaborating with coding agents must develop a deep mental model of the code to avoid accumulating "cognitive debt" as the agent makes larger, more complex changes.

2 min read

Tools

Paul Bakaus on 'Skill Engineering': Keeping Humans Central to AI-Driven Design

Paul Bakaus, creator of the open-source design skill Impeccable, argues that "skill engineering" can make AI agents more useful without displacing human designers.

6 min read

Tools

Meta quietly launches Pocket, an AI-powered app for creating short interactive 'gizmos'

Meta has released Pocket, an app that lets users generate and share small interactive apps and games—called "gizmos"—using AI prompts.

2 min read

Tools

OpenAI to Ship ‘Codex Micro’ Shortcut Keyboard on July 15

OpenAI has teased its first hardware product, the Codex Micro, a shortcut keyboard co-developed with Work Louder and due to launch on July 15.

2 min read

Tools

Z.ai launches ZCode: a free agentic IDE built for GLM-5.2

Z.ai (formerly Zhipu AI) released ZCode, a free desktop “Agentic Development Environment” tailored to its GLM-5.2 model, available for macOS, Windows and Linux with BYOK support and promotional usage bonuses through July 31.

7 min read

Tools

AI agent automation for dating: experiments with OpenClaw, benefits and privacy concerns

Some users are employing automated AI agents such as OpenClaw and Claude to automate dating outreach, generating large engagement but raising ethical and privacy questions.

3 min read

Tools

Introspection outlines practical patterns for self‑improving ‘autoresearch’ agent loops

Roland Gavrilescu, co‑founder and CEO of Introspection, described at the AI Engineer World’s Fair how autoresearch — an outer loop of agents that studies and improves a primary system — can be made reliable and deployable.

4 min read

Tools

Introducing a dry-run step for agentic deployments to reduce risk and cost

The provider's development team is introducing a 'dry-run' step for agentic (agent-based) deployments; the feature enables pre-deployment checks (for example node --check, tsc --noEmit, next build) and thereby reduces unnecessary costs and the risk of errors.

1 min read

Tools

Cursor builds forward‑deployed engineering teams to embed AI agents across the software lifecycle

Pauline Brunet, VP of Forward Deployed Engineering at Cursor, describes how the company’s FDE team helps customers implement long‑running AI agents across planning, coding, testing and deployment.

3 min read

Tools

When and How to Use Reinforcement Learning for Agents in the NeMo/Nemotron Ecosystem

Reinforcement learning (RL) is increasingly used to align and specialize open language models for agentic workflows, turning verifiable success criteria into training signals.

7 min read