Announcing Model GPT: Generative AI for enterprise software delivery

The new tool scales generative AI throughout DevOps and CI/CD, providing visibility, optimal test generation, pipeline integration and cross-team collaboration

Bray, Co. Wicklow, Ireland, May 4 2022 - Curiosity Software, veterans of continuous software delivery, today released a preview of “Model GPT”. The tool implements generative AI at scale during enterprise software development, unlocking AI’s transformative capabilities for “agile” ways of working, CI/CD, DevOps, and more.

Where early explorations of Chat GPT for software delivery have remained poorly understood, speculative and confined to ad hoc tasks, Model GPT provides the transparency, control and integration needed to implement AI at enterprise scale.

It automatically analyses scenarios and web pages provided by a “human in the loop”, converting them into flowcharts of the desired system. Model-based test generation then creates and maintains optimised tests, moving in seconds from text prompts to models, tests and data.

This AI-augmented, quality-centric approach avoids perennial bottlenecks and quality issues in agile software delivery.

Software developers, engineers, testers and business analysts can collaborate from logically precise models, using AI to map out how complex systems should work. Engineers can in turn develop faster, without rework caused by misunderstood requirements.

The system and requirements are furthermore tested continuously to avoid defect creation and perpetuation. Optimised test generation tests as rigorously as possible in-sprint. The models further explain what’s been tested, why, and when it’s “enough” for a release. Testers can retain the transparency, governance and control needed to release enterprise software, even when high-performing AIs generate tests.

Curiosity’s Test Modeller further provides the open integrations needed to scale generative AI throughout CI/CD. Model GPT maintains AI-augmented artifacts in existing tools, including requirements, user stories, test cases and data. A closed feedback loop updates exported artifacts as the models change, avoiding technical debt, quality issues and maintenance bottlenecks.

With Model GPT, a “human in the loop” provides feedback to the generative AI, regenerating and perfecting its models based on new prompts:

Whereas early applications of Chat GPT lack feedback loops and memory, Model GPT can self-analyse its flowcharts. The “three amigos” of software delivery can collaboratively perfect models for evolving systems, using AI-augmented flowcharts to refine requirements, develop rapidly, and continuously generate tests.

James Walker, Curiosity’s Chief Technology Officer, describes how:

“We’re seeing generative AI perform common development tasks faster than ever. Yet, moving faster without quality will only take you to a worse result, sooner.”

He continues:

“Adding modelling provides visibility, transparency and control. Software teams can understand what an AI is proposing for development, what it’s testing, and why. They can provide feedback to ensure requirements’ quality before code is written, while testing rigorously. Measurable and demonstrable risk metrics further retain quality control during AI-augmented development.”

Learn more about Model GPT and book a meeting with a member of our team.

Want to understand how generative AI will impact software delivery? We're bringing three testing veterans together for a live roundtable discussion. Watch on demand!

Press Contact