• Dotika
  • Posts
  • Adobe Photoshop adds Google's Nano Banana AI

Adobe Photoshop adds Google's Nano Banana AI

ALSO : TwinMind Ear-3 sets new voice AI records

News of the day

1. Adobe Photoshop integrates Google's advanced image AI, 'Nano Banana' (Gemini 2.5 Flash Image), as an optional tool, enhancing creative workflows for users.Read more

2. TwinMind Introduces Ear-3 Model: A New Voice AI Model that Sets New Industry Records in Accuracy, Speaker Labeling, Languages and Price.  Read more

3. Robotics startups are experiencing a boom, driven by a decade of innovation, falling hardware costs, and market maturation, attracting significant venture capital investment.  Read more

4. Thinking Machines Lab investigates why large language models give inconsistent answers even at temperature 0  Read more

Our take

Hi Dotikers!

Should we still build our own models, or become master integrators?
Our view, clear and straightforward: companies like Adobe need a hybrid strategy. Invest in in-house model bricks where control and compliance are vital, and plug in the best external models without hesitation whenever they deliver a quality leap.

This week proves the point: Photoshop is integrating Google’s image AI, nicknamed Nano Banana, officially Gemini 2.5 Flash Image, as an optional tool for more consistent edits alongside Generative Fill, with rollout planned for September. According to The Decoder, its quality often surpasses Adobe’s own Firefly models.

Why is this the right move? On one side, Adobe must protect its defensible advantages: legal safety of content, traceability through Content Credentials, cost governance, and alignment with Creative Cloud. That’s exactly what Firefly promises, “commercially safe” usage built into pro workflows.

On the other side, the performance frontier shifts fast, and Nano Banana already brings an editing consistency that wins over the market. Apple is following the same orchestration logic: Apple Intelligence relies on homegrown models deeply integrated into the ecosystem, while also opening to partners. ChatGPT is already in the loop, and Craig Federighi explicitly said Google Gemini may be offered next, letting the user pick the best model per use case.

In AI, purism is expensive, pragmatism pays. Differentiation doesn’t come from “my model beats your model,” but from the ability to compose a multi-model stack, measured in quality, cost, and risk, and deliver it in an irresistible product experience. Build where it makes you unique and defensible. Plug in the rest.

M.

Tweet of the day

Reply

or to participate.