AI Agents Reshape Software Testing, Infrastructure, and Kernel Debugging

AI agents are moving beyond chatbots into production infrastructure. This week, Y Combinator-backed TesterArmy launched an agentic testing platform, Baseten is finalizing a $1.5 billion funding round at up to $13 billion valuation, Databricks released Genie One, Cohesity introduced headless protection, and Claude AI helped fix a years-old AMD Linux display bug.

TesterArmy, founded by Oskar, Szymon, and Piotr, lets developers describe tests in natural language. The platform runs end-to-end checks before deployment and in production, supporting web and mobile apps. It integrates with GitHub, CI/CD pipelines, Slack, and Discord, and handles authentication including OAuth and OTPs.

Baseten's $1.5B bet on cheap inference

Baseten is finalizing a $1.5 billion funding round that values the company at up to $13 billion, according to The Next Web. The round is structured in two tiers: some investors buy in at an $11 billion valuation, others at $13 billion. The company is betting that AI's profits lie in cheap inference, not expensive training.

• Baseten's dual-tiered round allows early investors a lower entry price.
• The company focuses on inference infrastructure, competing with AWS, Azure, and GCP.
• Baseten's valuation has grown from $1.5 billion in 2024 to up to $13 billion in 2026.

Databricks and Cohesity add agent features

Databricks released Genie One, an AI agent coworker that can query data, generate reports, and automate workflows within the Databricks lakehouse. Cohesity introduced headless protection, an agentic approach to data backup and recovery that reduces manual intervention. Both products aim to reduce operational overhead for enterprise teams.

Claude AI, developed by Anthropic, helped fix a years-old AMD Radeon Linux display bug that caused laptop screens to freeze after periods of use. The bug, tracked back to 2017, affected numerous laptops. Claude Code analyzed the AMDGPU kernel driver and identified the problematic commit, leading to a proposed fix. The case highlights how AI agents are now assisting with low-level kernel debugging.

What comes next for agentic infrastructure

The convergence of agentic testing, inference infrastructure, and AI-assisted debugging signals a shift. Companies are betting that agents will handle routine tasks across the stack, from QA to kernel maintenance. TesterArmy's natural language testing, Baseten's inference focus, and Claude's kernel fix all point to a future where AI agents are embedded in production workflows, not just experimental tools.