Companies

Approxima

approxima.ai

Approxima provides AI agents that autonomously test, debug, heal, and merge software changes through CI/CD.

HQSan Francisco, California, United States
Employees1-50
Jobs checked 19h ago
Developer ToolsAI AgentB2B SaaSPre-Seed / Seed

About

Approxima builds AI agents that automate software maintenance by scoping work, writing code, validating fixes in sandboxes, and merging changes into production. It sells to software development teams using CI/CD workflows, differentiating through self-healing end-to-end tests that combine Playwright with real-browser validation.

Market

Approxima competes in AI-native software testing and developer infrastructure, spanning pull-request validation, end-to-end browser QA, and agentic software maintenance. It positions itself as a trust layer for AI-assisted development by running agents against each pull request in an isolated sandbox and validating changes before production. Its differentiation is a lightweight, self-healing layer on top of customers' existing Playwright tests—intended to preserve high fidelity while being more deterministic and faster than general browser-use agents, rather than forcing companies onto a fully proprietary testing platform.

Target Customers

Software engineering teams at companies shipping web applications, especially those maintaining Playwright-based end-to-end test suites and adopting AI-assisted development. Likely buyers are engineering or QA leaders seeking automated pull-request regression detection, reduced flaky-test maintenance, and safer validation before production; the company is positioned in enterprise software but does not disclose a firm-size threshold.

At a Glance

Problem

Approxima addresses the maintenance and reliability tax of end-to-end software testing, especially for teams adopting AI-assisted development and shipping changes frequently. Engineering teams must either spend substantial time maintaining extensive Playwright suites—updating selectors and debugging flaky failures—or move their tests to proprietary third-party AI QA platforms. The resulting pain is recurring engineering cost, noisy or slowed CI, and a tradeoff between control and convenience. The core use case is validating every pull request in a real browser without developers babysitting the test suite.

Product / Service

Approxima plugs into a company's CI/CD pipeline and runs AI agents on every pull request. The agents build the application in an isolated sandbox and test it in a real browser, while Approxima operates as a layer on top of Playwright that diagnoses flaky tests and heals them. Its wrapper is designed to be low-risk: if a customer runs out of credits or stops using the service, the underlying tests continue to work as before.

The product's benefit is to preserve Playwright's speed and ownership while reducing the maintenance burden that makes end-to-end testing expensive. Approxima's broader positioning is an AI trust layer for software development: agents can scope work, implement changes, validate fixes end-to-end in sandboxes, and potentially merge them into production.

Market

Approxima competes in AI-assisted end-to-end testing, autonomous QA, and developer-infrastructure software. Its positioning is differentiated from fully proprietary AI testing platforms because it augments an existing Playwright suite rather than requiring a complete migration. Relevant adjacent competitors include mabl, whose agentic tester autonomously updates tests, and Momentic, an AI testing platform that builds, runs, and maintains tests as products change.

The company appears to be at an early commercial stage. Y Combinator lists it as an active Winter 2026 enterprise-software infrastructure company founded in 2025 with a two-person San Francisco team, while its public materials offer a demo and its GitHub organization describes an open-source agentic web-testing platform. The available evidence does not disclose customer counts, revenue, or commercial funding, so Approxima should be treated as pre-revenue or, at minimum, traction-undisclosed rather than as a company with demonstrated commercial scale.

Founders & Leadership

Ethan PronevFounder
Co-founder
Ashish SelvarajFounder
Co-founder

Funding History

2026-01
Seed (Crunchbase labels it Pre-Seed)$500K

Y Combinator

Recent News

2026-07-15
Approxima featured in Y Combinator’s enterprise-software directory

Y Combinator’s enterprise-software directory describes Approxima as building agents that scope work, write code, validate fixes end-to-end in sandbox environments, and merge changes to production.

2026-05-28
Approxima listed as a Y Combinator W26 company

A public Y Combinator portfolio list identifies Approxima as a W26 company and presents its positioning with the tagline “Your software should build itself.”

2026-03-10product
Approxima — AI Agents That Test Every Pull Request

Approxima’s product positioning centers on end-to-end tests that heal themselves. Its software operates as a layer on top of Playwright to debug flaky tests and automatically heal them.

Active Roles

0

No active roles right now.

Get notified when they post

Business Model

Approxima appears to monetize as a B2B enterprise-software platform for development teams, using a demo-led sales motion. Its product references credits, but no public price sheet or specific subscription tiers were found, so the exact pricing model is undisclosed.

Products

Autonomous AI agents that test every pull requestApproxima's self-healing Playwright layer and wrapperEnd-to-end test debugging, regression detection, and sandbox validationBroader autonomous software-maintenance agents that scope work, write code, validate fixes, and merge changes

Tech Stack

PlaywrightAI agents / agentic AIReal-browser automationIsolated sandbox executionCI/CD and GitHub pull-request integration

Competitors

Greptile
QA Wolf
Momentic
Qodo
Devin (Cognition AI)
Factory AI