Zencoder just launched an AI that can replace days of QA work in two hours

ZencoderThe start of the artificial intelligence coding established by the Serial Owner Andrew File, was put into operation in public beta today MountainousAI-powered agent designed to automate the latest software test until the end. It can be critical, but often slow-step product releases can delay the day or weeks.

The new tool represents Zencoder’s most recent attempts to distinguish only the role of code generation, but all software development has competed in the market to automate workflows. Unlike existing AI coding tools, the Zentester, which is primarily aimed at the writing code, is targeting the inspection stage – it works as the software is intended before reaching customers.

“Verification is a missing link in the scaling of AI from experimental experiments,” said File Akluv was an exclusive interview with Venturebeat. CEO of project management company previously Goggles and sold Citrix for $ 2.25 billion Added in 2021: “Zentester does not create tests only – confirming the development of developers about the EU’s arising or human writing code will be made.”

The AI ​​coding market is coming to the announcement as quickly as the consolidation. Last month Zancoder Mechanical EngineeringAnother AI coding assistant with over 100,000 downloads. At the same time, Openai has reached an agreement to obtain windsurf for coding About $ 3 billion (Deal in May ended). The actions emphasize how companies are in a hurry to build a comprehensive AI development platform, not the point solutions.

Why the program test has become the largest roadblock in the AI-powered development

Mountainous Software solves in the program development: long opinion loops among developers and quality guaranteed groups. In a typical enterprise environment, the developers type the code and send to QA teams for testing, often wait a few days for feedback. Until then, developers, the cost of expensive, other projects were transferred to other projects.

“During a typical engineering, a developer sets a feature and sends a feature, a few days later,” FileV Venturebeat. “Then they passed on something else. In this context, it can extend simple adjustments for a weekly long period of weekly – a particularly painful-obvious in the transition and retreat.”

Early customer Club Solutions Group Dramatic progress with CEO Mike Cervino, “How did the QA take several days for how many days it took now, and now took a few days last 2 hours.”

Timing, especially the AI ​​coding tools generate an increasingly large amount of code. As the means GitHub Copilot and Cursor Have an accelerated code generation, and they also created new quality guarantees. FILEV, AI tools increase the generation of code generation with 10x, will increase the same as traditional QA processes.

Zentester AI agents hits the keys and fill out forms such as human tests

Unlike traditional test frames that require complex scripts from developers, Mountainous Operates in flat English language instructions. AI agent can interact with applications such as human users to navigate through the buttons, filling forms and software workflows by clicking.

The system combines them with existing test frames, including dramat and selenium rather than replacing them. “We do not like people who leave the items that are part of DNA completely,” File said. “We feel that the AI ​​must use existing processes and means.”

Zentester offers five basic allowance: Creating a comprehensive test set, creating a comprehensive test set during feature development, AI generated code, automated test operation and sustainable integrated pipeline to improve the quality of autonomous inspection.

The tool offers a wider platform that includes coding agents for the latest addition to a wider agent platform of Zencoder, a wider agent platform. Company “Repo grokking“Technology analyzes all code deposits to reduce a bug fixing pipeline.

Starting as Microsoft’s EU development tools market strengthens competition in the market GitHub Copilot and newer applicants Cursor The developer is revived for the mindset. The approach of specialty agents in the building for various stages of development, is primarily compared to opponents aimed at the generation of code.

“At this point, there are three powerful coordination products in the market: This is the cursor and wind,” File said at the last meeting. “It is difficult to compete for small companies and difficult.”

The company requires superior performance on industrial criteria, indicating 63% success rates Swe-bucket 30% of the tests and newer SWE-BENCH Multimodal Assessment – The results are the best performances of the double previous performance.

Note that the test automation for the end is the next step for the test automation AI coding tools, but a successful application requires a complex concept of logic and user work streams.

Which enterprise buyers should know before receiving AI test platforms

Zencoder’s approach Evaluating AI testing tools offers both opportunities and difficulties for enterprise customers. Company SOC 2 TYPE II, ISO 27001 and ISO 42001 Certificates address security and compatibility are important for large organizations.

However, FILEV ensures that the enterprise is careful. “According to the enterprises, we do not necessarily be able to change the change in the program,” he said. “The given thing is aI-expanded, where now they can quickly review the AI ​​code and reduce the amount of work that should be made by the next party in the pipeline.”

Company integration strategy – those who work in the existing development environment Visual studio code and Jetbrains IDes Instead of requiring platform keys – can apply to enterprises with built-in tools.

Race to automate software development until the idea is placed

Zentester’s launch positions will compete for a larger part of the workflow of program development as Zancoder, AI tools. The company extends to the full automation of the requirements for vision, production placement, fleeces accepts current restrictions.

“The next jump will be the requirements for production – everything” said fleev. “Now you can have this with natural language requirements, and then can you help them split it, build, build, review, review, confirm and produce?”

Zencoder, Zentester offers with three price types: a basic version for free, $ 39 for a user with a user of $ 19 business plan and premium support and compliance features.

Zentester offers a third opportunity for an industry that will replace the programmers or simply to be more productive or more productive. The question is something that can no longer be able to write the code and try to test the code.