COVINA, Calif., Sept. 15, 2026 (GLOBE NEWSWIRE) -- Applitools, the leader in Visual AI and test automation, today announced a major platform release establishing the deterministic governance layer for the Agentic Software Development Lifecycle (SDLC). By embedding Visual AI directly into AI coding workflows, Applitools removes a critical engineering bottleneck, delivering reproducible visual verification at AI development speed.
This capability directly addresses a rapidly compounding risk: as software engineering teams adopt AI coding agents,including Claude Code, Cursor, GitHub Copilot, and Cline, to author UI code 10x faster, recent telemetry from Faros AI shows median PR review times surging 5x and production incident risks more than tripling. Relying on vision-language models (VLMs) for code inspection creates a dangerous "Probabilistic Validation Gap” because AI agents cannot see how UI changes render across different browsers, devices, form factors, and operating systems. Without deterministic validation, teams risk undetected visual bugs reaching production while facing excessive token spend.
"AI can write code and write tests, but it can’t see how the rendered code impacts user experience across different devices, browsers, and operating systems—and that’s exactly where UI quality breaks down," said Anand Sundaram, CEO of Applitools. "Teams moving fast with AI run into review bottlenecks, incomplete test coverage, and visual bugs that only appear in production, costing time and threatening reliable production. The Applitools MCP server gives LLMs eyes across every form factor, meeting shift-left teams within their workflows to ship flawless experiences without trading quality for speed."
Closing the Validation Gap
In AI-assisted development, the app and its tests change faster than anyone can verify them. Asserting correctness element by element buries teams in code that breaks on the next iteration, while VLMs replace determinism for probabilistic guesswork. Applitools delivers an enforceable, reproducible quality signal through three key innovations launched today:
- Eyes Visual AI MCP Tools (@applitools/mcp): Connects deterministic Visual AI directly to AI coding agents like Claude Code, Cursor, Copilot, and Cline. Agents can now pinpoint exact visual differences and leverage page structure snapshots to correlate layout shifts directly to the underlying source code. Most notably, these tools empower agents to autonomously perform full baseline maintenance, from accepting and rejecting changes to dynamic data masking, slashing test maintenance overhead by up to 95% directly from the IDE.
- Figma Design Baselines: Integrates Figma design systems directly into SDK workflows, turning design frames into automated visual references. By automatically matching viewports and dimensions, the platform streamlines the design handoff process by enabling seamless comparison of design specs at scale against live implementations without requiring cumbersome manual code adjustments or brittle export scripts.
- NLP Test Steps for SDKs: Replaces fragile locator logic with plain-English execution within Playwright and JavaScript/TypeScript test suites. Powered by Applitools' proprietary Deterministic Language Model (DLM), this capability makes tests dramatically shorter, easier to author and review, and resilient to UI changes in complex, dynamic applications, all while protecting sensitive UI data and user information from public LLMs.
"Integrating Applitools MCP tools into our evolving AI-assisted development workflows is helping us move toward a more agentic approach to software development," said Artem Dvornichenko, an engineering leader at the leading European independent rail and travel platform. "As we continue this transformation, Applitools gives our developers greater confidence that AI-authored UI code—from individual components to complete web applications—aligns with our designs and meets our visual quality standards before reaching production."
Proven Enterprise Impact and Privacy
Empowering AI agents to perform automated visual validation and full baseline maintenance within agentic workflows and CI/CD pipelines, Applitools helps engineering teams deliver flawless user experiences without compromising speed or quality, cutting visual test maintenance by up to 95% while eliminating false positives and token overhead.
At the enterprise level, the Applitools platform is engineered for high-concurrency test automation pipelines and operates with rigorous data governance. Page snapshots, DOM trees, and rendered interface elements are processed securely via SOC 2 Type II certified infrastructure. Sensitive UI data and user information are never transmitted to public foundation models or used to train third-party AI systems, ensuring total compliance with enterprise security requirements.
Availability
The Applitools Eyes MCP Tools, Figma Design Baseline SDK integration, and Intent-Based NLP engine are available starting today, September 15, 2026, across the Applitools platform ecosystem, supporting JavaScript/TypeScript (@applitools/eyes-playwright), Python, Java, and .NET. Read the release notes.
Adam Carmi, Applitools CTO and Co-Founder, will demo these new features and how they bridge the Probabilistic Validation Gap in agentic workflows during a live webinar on September 24, 2026. Register for the webinar or request a personalized demonstration at applitools.com.
About Applitools
Applitools delivers deterministic, enterprise-grade AI test automation for modern software delivery teams. By combining a proprietary Deterministic Language Model (DLM) with mature Visual AI trained on billions of images, Applitools eliminates the "maintenance tax" of verifying AI-generated code by hand. Replacing thousands of fragile code assertions with single-line Visual AI checks, Applitools helps developers, SDETs, and engineering leaders release web, mobile, component, and document experiences across every device with maximum speed and uncompromising quality.
Media Contact
Joanna Schloss, CMO
Email: joanna.schloss@applitools.com
Phone: +1 650-680-1000
