Oct 17, 2026 Technology Pulse

🧠 Amazon Acquires DuckLabs and Integrates DuckDB in Amazon Aurora PostgreSQL

Amazon has acquired DuckLabs, the company behind the analytical database DuckDB, with plans to expand the database's development and reach while maintaining its open-source status. Co-founders Hannes Mühleisen and Mark Raasveldt will continue to lead the team under AWS. Following the acquisition, Amazon Aurora PostgreSQL is receiving its first feature powered by DuckDB. Read full analysis →

📰 Hillclimb: Open-source AlphaEvolve-style search with Claude Code and Codex

Rebase Energy published Hillclimb, an open-source project implementing AlphaEvolve-style search with Claude Code and Codex. The repository brings evolutionary search techniques to AI coding assistants, enabling automated search and optimization across both platforms. Read full analysis →

🧠 Claude CRISPR-like sequence may be a parallel discovery of unpublished research

A CRISPR-like sequence associated with Claude may represent a parallel discovery of previously unpublished research. The finding has sparked a priority dispute concerning the identification of the viral DNA pattern. The issue centers on whether the sequence emerged independently or mirrored earlier unreleased scientific findings. Read full analysis →

📰 I built a Claude Code terminal on the NDS

Developer Logan Carrillo created a Claude Code terminal client designed to run directly on the Nintendo DS. The project is hosted on GitHub as ds-claude-terminal, providing a portable handheld interface for the terminal assistant. Read full analysis →

🛠️ How the Claude Code team does code review

The Claude Code team at Anthropic published an overview detailing how its engineers handle code review across internal workflows. The team outlined its approach to solving review challenges and evaluating software changes across the development cycle. Read full analysis →

🧠 90% of US working hours are spent on tasks not covered well by AI benchmarks

Mercor published a labor index benchmark map showing that 90% of US working hours are spent on tasks not covered well by AI benchmarks. The findings highlight a substantial gap between the tasks evaluated by current artificial intelligence test suites and the everyday duties of American workers. Consequently, the vast majority of working hours across the United States remain outside the scope of today's primary evaluation frameworks. Read full analysis →

📰 Claude Builds an Internal Customer Service App

The Automated Operator detailed a practical workflow demonstrating how Claude built an internal customer service application. The project showcases using the AI model directly to construct specialized support software for operational teams. The implementation highlights expanding organizational use cases for model-driven development in automating internal business workflows. Read full analysis →

🧠 Four evaluation engineering lessons learned from building agent plugins

Measuring an agentic system is the primary challenge rather than constructing it, according to Google Cloud's evaluation engineering lessons from building agent plugins. Evaluation engineering provides the critical foundation necessary for building reliable AI systems. Setting up systematic, repeatable tests to measure an agent's true behavior determines whether teams can ship tools they can trust. Read full analysis →

🧠 Mark Wahlberg wants to take on the longevity-startup crowd with an affordable gym concept

Mark Wahlberg is building an affordable gym concept aimed at competing with the longevity startup crowd. Having previously invested money in a gym chain that was kicked off the NYSE, Wahlberg is now developing his own brand. The new concept promises to offer cold plunges, blood panels, and protein muffins without requiring a trust fund. Read full analysis →

📰 Parallax (YC S26) – A live reviewer running alongside each coding agent

Conifer launched Parallax, a live reviewer designed to run directly alongside each coding agent. The tool, part of the YC S26 batch, operates in real time to provide continuous review as coding agents work. Read full analysis →

🔎 More Interesting News

• MCP servers are more than just tools — HN AI Agents