From Pixels to Test Suites: AI as the New Co-Worker in Design and Quality Assurance

Two professions that look very different on the surface, design and software testing, are going through the same change. In both, AI has moved from a curiosity to a co-worker that produces first drafts, repeats routine steps and never gets tired. This post looks at the numbers and at what skills help you stay valuable.

Design and images

Google said more than 5 billion images were created in the Gemini app in under a month after the Nano Banana image editor launched, as OfficeChai reported.

Other signals from the design world:

• Canva has over 265 million monthly users, and Dataconomy reported on how its growth was linked to its use of ChatGPT. Its AI site builder is reported to have 10 million or more users.
• Adobe's annualised recurring revenue hit $26.06 billion exiting the first quarter of fiscal 2026, according to Quartz. Firefly is reported at more than $250 million in ARR with video generations up eight times.
• In August 2025 Meta announced it would license Midjourney's "aesthetic technology" for its future models and products, according to Engadget.

The pattern: the big design platforms are building AI into every step, and even rival tech giants are licensing image models rather than building everything themselves.

Quality assurance and software testing

The World Quality Report 2025-26 from Capgemini surveyed more than 2,000 executives. Its headline findings:

• 89% are piloting generative AI in quality engineering.
• Only 15% have rolled it out company-wide.
• 50% say they lack AI skills, and the report lists data privacy risk (67%) and integration complexity (64%) as top barriers to scaling.
• The average productivity gain reported is 19%.

Playwright, the popular testing framework, now ships Planner, Generator and Healer agents that write and repair tests. That means the routine part of testing, writing the same kind of check again and again, is increasingly automated.

The gap is the opportunity

The distance between "piloting" (89%) and "scaled" (15%) is exactly where skilled people are needed. Companies have plenty of experiments and a shortage of people who can turn them into reliable processes.

Each step in that chain is a job: someone has to measure, someone has to protect data, someone has to connect systems, someone has to teach colleagues. None of these needs you to build a model from scratch.

What I would learn if I were a designer or tester

Both: write down your process. People who can describe a workflow clearly get the most from AI, because a clear description is exactly what an agent needs. For a free way to practise testing and security with real tools, see the free toolkit for testing, security and learning from the source.

Keep reading

• Shrinking Tasks vs Growing Roles: Which Jobs Are Under Pressure and Which Are Growing
• Two Years, Every Month a Headline: AI From Sora to GPT-5 and Nano Banana
• Free AI Toolkit (3/3): Test It, Secure It, Learn It From the Source

See the complete AI Era series

---