en.Wedoany.com Reported - AI coding startup Baz Technologies Inc. today launched a new platform, Baz Planner, which sits between developers and the codebase to catch software vulnerabilities before they enter the production workflow. The company also announced a $9 million extended seed round, bringing total funding to $17 million. The round was co-led by existing investors Battery Ventures and Boldstart Ventures, with participation from new investors AFG Partners and Disruptive VC.
Baz Planner debuted at the AI Engineer World's Fair in San Francisco. The platform is a new gateway that automatically routes each new idea through dynamic loops, which instantly detect and understand the root cause of new vulnerabilities, then proactively rewrite coding plans to eliminate them.
Baz Planner builds on the company's AI code review tool launched last year, which currently ranks first on the precision-weighted Code Review Bench. With this tool, development teams can manage and secure their AI code using AI agents designed to enforce coding standards across product, design, architecture, security, and reliability issues.
Unlike network-capable AI models that can autonomously scan millions of lines of code to find subtle defects, Baz Planner elevates code security by reviewing every specific change against current and expected new architecture. Any problematic changes are flagged before new code is saved.
The startup was founded by a group of former Palo Alto Networks Inc. engineers who helped expand the company's cloud application security business, making it one of the leading code-to-cloud security firms. The team applies four principles that shaped cloud development—observability, explainability, predictability, and reproducibility—to AI-generated code to prevent defects from creeping in.
When using Baz Planner, each new suggestion from the AI model is evaluated against a strict risk matrix that blocks unsafe paths and enforces defined boundaries. Code is only allowed to be delivered to production after strict risk mitigation measures are applied. In this way, early adopters have reduced downstream rework by over 65%, measured by the frequency of rollbacks and hotfixes required after merging.
Baz co-founder and CEO Guy Eisenkot said customers have been pushing the company to intervene earlier in the planning stage, beyond code review. "That's where the cost of eliminating vulnerabilities and defects is lowest," he explained. "Baz exists because customers reject the notion that AI-generated code means blindly accepting risk."
Baz Planner does not focus on style and syntax but analyzes how new code will impact runtime to catch errors, silent regressions, and other security defects. The tool employs four specialized agents working together to review each new code suggestion, including a spec reviewer agent that validates new code against product requirements, design, and expected behavior. An advanced security agent reasons across authorization and network boundaries, infrastructure, pipelines, and final application code to uncover any new vulnerabilities. A site reliability engineer agent correlates repository changes with production telemetry data to identify risks related to performance, reliability, and observability. Finally, a fixer agent applies and verifies each code change deemed "safe" in an isolated runtime environment.
"Guy and the Baz team built the code-to-cloud security playbook at Palo Alto Networks, and now they're applying the same rigor to AI-native engineering," said Barak Schoster, partner at Battery Ventures. "As development teams deploy a large number of coding agents, Baz is becoming the super control platform to orchestrate them, covering spec-driven development, UI review, security, quality, reliability, and planning, ensuring AI-generated code is delivered safely at scale."
