Skip to content

How IBM tests AI apps and code with AIDA and AI Workspace

How an IBM leader built an app with IBM Bob, validated it with AIDA, and piloted AI Workspace for agentic QA across the SDLC.

Oct. 13, 2026
Author: Lizzie Stokes

Key takeaways

  • Speed vs. risk: Gartner predicts 40% of new business software will be built via AI-assisted development and “vibe coding” by 2028. While speed accelerates time-to-market, it introduces risks that traditional QA cycles are too slow to address.
  • Agentic QA for rapid prototyping: Tricentis Labs is developing AIDA to close this gap, giving teams a lightweight, autonomous way to catch common errors in AI-generated applications before deployment.
  • End-to-end enterprise governance: AI Workspace orchestrates quality and development agents across the SDLC, automatically translating business requirements into automated test cases and scaling governance alongside enterprise applications.

Mike Dillon is a Global Platform Leader at IBM. While deeply technical and experienced in enterprise architecture, he is not a traditional hands-on software developer. Yet this spring, Dillon paired his domain expertise with IBM Bob, IBM’s AI coding assistant, to engineer his first full-scale enterprise application: WatsonAuditX.

Designed to automate the documentation required for complex SAP audits, the app addresses a familiar operational roadblock. Having led numerous SAP compliance reviews, Dillon possessed the domain depth to know exactly where legacy auditing processes broke down and how an automated application could help.

Within a few months of AI-assisted engineering with Bob, WatsonAuditX was operational, capable of locating records and surfacing change materials for SAP teams mid-review. But as deployment neared, Dillon faced a familiar challenge: how to rapidly and reliably validate an application built without a dedicated QA team.

Faster testing for faster development

Dillon’s experience reflects a broader industry shift. Technical leaders and domain experts are increasingly adopting AI-assisted and “vibe-coded” development workflows, a trend Gartner predicts will account for 40% of new business software by 2028.

The primary driver is speed. Rather than enduring extended procurement cycles or waiting for overloaded developer queues, technical practitioners can build bespoke, domain-specific applications in days or weeks with AI coding assistants like IBM Bob.

However, rapid creation creates a quality problem. Although these applications frequently coordinate critical enterprise workflows, putting them through heavyweight, traditional QA cycles would negate the speed that made building them worthwhile. Mitigating this risk requires an intermediate solution: an autonomous, agentic QA tool capable of quickly identifying AI-generated errors and edge cases without requiring a pre-existing test suite.

Watch the webinar: AI transformation roundtable with IBM and TTC Global

Meet AIDA: Agentic QA for AI-built applications

At SAP Sapphire, Dillon connected with the Tricentis team and was introduced to a new AI agent, project-named AIDA, that specializes in testing AI-generated applications and surfaces critical risks autonomously. AIDA is one of the first innovations from Tricentis Labs, an innovation incubator that gives users early access to AI technology and a hand in shaping its development.

When Dillon pointed AIDA at WatsonAuditX, the agent immediately identified a blocking modal pop-up error that would have disrupted end users. Dillon fed the finding directly back into IBM Bob, which located and resolved the root cause across three separate files.

“Then I went back to AIDA, ran it again, and it was resolved,” said Dillon. “With Bob building and remediating the code, and AIDA validating the user experience, the loop was completely autonomous. For an agile application that I want to iterate on and deploy quickly, it’s a game changer. If this is my use case as one person at IBM, there are countless technical leaders across every industry doing the exact same thing.”

AIDA exemplifies the core mission of Tricentis Labs: build in public alongside industry leaders to prepare QA teams for the era of agentic testing and development.

Read more: Tricentis introduces new AI innovations to advance agentic enterprise software development

Managing risk as agile tools become business critical

As AI code generation accelerates, organizations face the risk of “application sprawl” — similar to the user-developed tools and shadow-IT macros of previous decades. A lightweight project created for a single team often expands across business units until revenue and compliance depend on it.

“Think of the sprawl,” Dillon warned. “Something a small team builds with AI proliferates across departments, and suddenly there’s critical operational revenue attached to it. How do you validate that quickly without halting momentum? You need an autonomous tool like AIDA to verify AI-written software before it enters formal enterprise pipelines.”

Agentic quality engineering at enterprise scale

Dillon’s rapid prototyping use case with IBM Bob and AIDA represents just one pillar of IBM’s broader quality engineering strategy. As the leader of IBM’s Tricentis platform, Dillon recognized the opportunity to bridge AI development and quality engineering, introducing Tricentis AI Workspace to pilot agentic testing across core enterprise workflows.

Watch the webinar: How to embed quality across the agentic SDLC

Dillon introduced AI Workspace to IBM to generate automated test cases directly from Jira user stories that had been coded by IBM Bob. Together with Bob, AI Workspace coordinates development and quality agents across the software delivery lifecycle:

  1. IBM Bob generates the application code to fulfill specific business logic directly from requirements (e.g., managing customer credit holds).
  2. AI Workspace analyzes the Jira requirements alongside the code produced by Bob to automatically generate comprehensive end-to-end test scenarios.
  3. The workflow executes autonomously, validating requirements and surfacing defects without manual QA intervention.

While currently being piloted in focused enterprise environments, this integration is nearing production readiness. Dillon’s broader roadmap for AI Workspace targets two major areas as it scales:

  • Automated backlog resolution: Triaging and remediating long-standing transactional queue backlogs.
  • Continuous SAP compliance: Providing built-in governance and auditability for SAP transactions, directly supporting IBM’s alignment with SAP Clean Core principles.

Conclusion

Quality engineering is evolving to match the speed of AI-driven software development. Through the combination of developer agents like IBM Bob and quality agents like AIDA, organizations gain rapid development spped coupled with lightweight safety checks for agile prototypes. As platforms like Tricentis AI Workspace advance toward full enterprise production, they establish the automated governance and end-to-end testing required to scale AI safely across mission-critical application portfolios.

Sign up for Tricentis Labs to explore AIDA and other emerging AI technologies from Tricentis

Learn more about Tricentis AI Solutions here

Author:

Lizzie Stokes

Writer, Tosca and SAP solutions

Date: Oct. 13, 2026
Author:

Lizzie Stokes

Date: Oct. 13, 2026

Lizzie Stokes

Writer, Tosca and SAP solutions

Lizzie is a writer covering test automation and SAP testing solutions for Tricentis.com. Throughout her career, she has reported on technologies such as commercial IoT, connectivity software, and satellite communications, with a focus on translating subject matter expertise into accessible, engaging content.

Recommended

You might also be interested in...