Miju Labs

← All buyers · Canva (incl. Leonardo.ai)

C

Cameron Adams

Co-founder and Chief Product Officer, Canva · Canva (incl. Leonardo.ai)

medium confidenceTier A · Talk to firstImage, video & audio modelsUI

Canva co-founder and CPO, and the person who has publicly described Canva's design foundation model in the most data-revealing terms of anyone in this batch: trained on structured data, millions of designs, and the actual sequence of edits used to build them, with the team deliberately breaking designs so the model learns to recognise spacing and hierarchy errors. That is a description of an internal human-judgement data programme, said out loud.

Background

A designer-engineer by origin — he was the design half of Canva's founding trio — which is why his account of the model is about process rather than pixels. He wrote the Canva post announcing the Leonardo.Ai acquisition, describing Leonardo's 120 researchers, engineers and designers as a bolstering of Canva's AI products.

His public argument about the future of creative work is that when execution is automated, judgement and empathy matter more, and that what separates great designers is strategic thinking and audience understanding rather than technical execution. For a vendor selling expert human judgement, that is the thesis sentence.

What they run now

Canva's product as a whole, including Canva AI 2.0 and the design foundation model, plus the integration of Leonardo.Ai (acquired 2024) into Canva's AI stack.

What they buy

Inference from his own description: edit-trajectory data (sequences of designer actions), deliberately corrupted/negative design examples for error recognition, and evaluation sets scored on alignment, readability and brand consistency. Also implicitly, a mapping from subjective language ('make it feel premium') to concrete design attributes — a rubric-and-annotation problem. Canva has not named any external vendor and appears to have built this in-house so far.

Career

  1. 2012 – presentCo-founder and Chief Product Officer, Canva

On the record

what the design model learns from · 2025-10

Says the model was trained on structured data, millions of designs, and the actual sequence of edits used to build them — capturing how designers work, not just finished outputs. source

negative examples · 2025-10

Describes the team deliberately breaking designs in testing so the model learns to recognise spacing and hierarchy errors. source

how they evaluate · 2025-10

Says they evaluate against real-world usage patterns including alignment, readability and brand consistency. source

subjective prompts · 2025-10

Notes users prefer describing work in subjective terms such as making something feel premium, and expect the tool to understand that contextual language. source

value of judgement · 2025-10

Argues that as execution becomes automated, judgement and empathy become more important, and strategic thinking separates great designers. source

Interviews & talks

Why it matters here

Highest-signal target in this batch after Ideogram. Everything Julian sells — trajectories, negative examples, taste rubrics, designer preference — is something Adams has publicly said Canva already does internally. The sale is not 'you need this'; it is 'you need more of it than you can generate in-house, and from designers outside Canva's own template ecosystem'. The weakness of an in-house programme is exactly that its taste is Canva's own.

How to reach

Speculation: a CPO is hard to reach cold and the operational buyer is more likely Danny Wu or the Generative Experience Group. Use Adams's statements as the framing device in outreach to them: quote his 'sequence of edits' and 'deliberately break designs' descriptions back and offer external designer panels that de-bias the in-house taste. If approaching him directly, the angle is diversity of taste — Canva's templates encode Canva's aesthetic, and a model trained only on them inherits it.

What we could not establish
  • No verified LinkedIn or X handle recorded here — a widely-used personal handle exists but was not confirmed against a company or personal source in this pass.
  • Whether the 'deliberately breaking designs' work is done by Canva staff designers or contractors.
  • Whether Canva's creator fund covers consent for design-model training.

Why they are on the map

CPO; public face of Canva AI 2.0. source — evidence dated 2025-10

Decides whether the problem is funded.

How Canva (incl. Leonardo.ai) buys data

Design model trained on 'millions of professionally created templates' (Danny Wu, Jan 2026); $200M creator fund (2023) pays royalties to designers who consent to training use. Canva declined to say whether customer data trains the design model (Computerworld, Oct 2025). Hiring Head of AI Research to bridge Canva and Leonardo research; 'Generative Experience Group' runs design-generation ML.

AI products under Danny Wu (Head of AI Products); Generative Experience Group builds design-generation models; Head of AI Research reports to CTO Brendan Humphreys and GenAI lead Paul Thompson and works with Leonardo.ai's research team in Sydney.

Known vendors: Canva Creators (template royalties)

Sources

  1. The Rundown AI: Inside Canva AI 2.0 with CPO Cameron Adams (Oct 2025)
  2. Canva Newsroom: Welcome to Canva, Leonardo!
  3. Computerworld: Canva debuts foundational 'design' model