# Stefano Corazza

Markdown copy of https://realalpha.app/people/stefano-corazza/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

Real Alpha has cut 8 moments of Stefano Corazza (Head of AI Research, Canva) talking, from an interview on The TWIML AI Podcast published Nov 10, 2026.

- Role: Head of AI Research, Canva

## What Stefano Corazza said

- [Canva's data makes smaller design models more economical](https://realalpha.app/s/canvas-data-makes-smaller-design-models-more-economical/)
  The TWIML AI Podcast · Nov 10, 2026 · 3 min
  > “you can fine tune or train a model that is two orders of magnitude fewer parameters if you have the data and you can beat the bigger model.” — Stefano Corazza · [5:53](https://realalpha.app/v/heE80uxOmus/?t=353)
- [Detailed scene descriptions help smaller models create better designs](https://realalpha.app/s/detailed-scene-descriptions-help-smaller-models-create/)
  The TWIML AI Podcast · Nov 10, 2026 · 5 min
  > “we underestimated how detailed of a scene description you can give to a diffusion model” — Stefano Corazza · [16:47](https://realalpha.app/v/heE80uxOmus/?t=1007)
- [Owning the model stack creates more room to cut inference costs](https://realalpha.app/s/owning-the-model-stack-creates-more-room-to-cut-inference/)
  The TWIML AI Podcast · Nov 10, 2026 · 4 min
  > “you can only make it really efficient if you own the full stack” — Stefano Corazza · [41:10](https://realalpha.app/v/heE80uxOmus/?t=2470)
- [Layered designs let people move elements before regenerating pixels](https://realalpha.app/s/layered-designs-let-people-move-elements-before-regenerating/)
  The TWIML AI Podcast · Nov 10, 2026 · 4 min
  > “And so it's a lot easier for a designer to just move that element to the right and then regenerate.” — Stefano Corazza · [13:34](https://realalpha.app/v/heE80uxOmus/?t=814)
- [Design editing needs benchmarks beyond single-shot generation](https://realalpha.app/s/design-editing-needs-benchmarks-beyond-single-shot/)
  The TWIML AI Podcast · Nov 10, 2026 · 4 min
  > “And you have benchmarks for image editing, which is great, but design editing is different than just image editing.” — Stefano Corazza · [23:33](https://realalpha.app/v/heE80uxOmus/?t=1413)
- [Style mismatches can be fixed without retraining from scratch](https://realalpha.app/s/style-mismatches-can-be-fixed-without-retraining-from/)
  The TWIML AI Podcast · Nov 10, 2026 · 3 min
  > “the cartoon images generated with craft initially were a little bit too realistic and too 3D looking.” — Stefano Corazza · [28:37](https://realalpha.app/v/heE80uxOmus/?t=1717)
- [Brush revises designs through tools and visual feedback](https://realalpha.app/s/brush-revises-designs-through-tools-and-visual-feedback/)
  The TWIML AI Podcast · Nov 10, 2026 · 5 min
  > “the faster you make that iterative loop, the more effective everything is” — Stefano Corazza · [35:15](https://realalpha.app/v/heE80uxOmus/?t=2115)
- [A simpler agent harness lets the model use familiar code paths](https://realalpha.app/s/a-simpler-agent-harness-lets-the-model-use-familiar-code/)
  The TWIML AI Podcast · Nov 10, 2026 · 4 min
  > “trying to leverage how good these LLMs are at writing code and at the same time simplifying the harness in the most basic possible way” — Stefano Corazza · [37:08](https://realalpha.app/v/heE80uxOmus/?t=2228)

## The episode

- [AI Design Still Feels Like a Slot Machine](https://realalpha.app/v/heE80uxOmus/) · The TWIML AI Podcast · Nov 10, 2026