Exclusive: Gamma revamps presentation platform to make output look more human
SiliconANGLE Paul Gillin ● Covered by 2 sources
Gamma rebuilt its presentation app to make AI slides look less generic. It now plugs into company tools and data, so the output can feel more human and less templated.
Based on reporting by SiliconANGLE, Paul Gillin — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Gamma has pushed out a rebuilt version of its presentation platform, and the pitch is simple: make AI-made slides look less obviously AI-made. The new Gamma 5 pairs a redesigned editor with a stronger agent and a much larger set of visual styles, all aimed at fixing the same complaint users keep raising — repetitive layouts, generic imagery and the kind of design choices people now call “AI smell.”
Jon Noronha, Gamma’s co-founder and chief product officer, said that complaint has grown louder since early this year. The old product kept users inside a narrower set of design components. The new one gives the agent more freedom to write custom code, draw shapes and place elements on a canvas, which is a fancy way of saying it can do more of the messy visual work itself.
Gamma’s agent runs in a sandbox and can generate images, handle files, do web research and check its own output. The company also says it has automated checks for familiar presentation sins, like text falling off a slide or pictures getting cropped in awkward ways. Users can still step in and edit individual pieces directly, and they can work with the agent on an outline before any slides are built, tightening the argument, audience and visual style first.
The company is also leaning hard into workflow and context. Gamma says it has around 20 connectors for outside information, including Slack and transcription tools from Granola, Fireflies and Fathom, and it applies users’ existing permissions when those sources are pulled in. It also supports importing PowerPoint files, PDFs and other existing material, with Noronha saying an internal benchmark for visual accuracy has moved from roughly 30% early on to over 97%.
Under the hood, Gamma says it uses more than 20 AI models, including ones from Anthropic, OpenAI and Google, plus several image generators. It selects those models through internal evaluations and user experiments that track things like edits, shares and whether people become paying customers. The company, founded in 2020, says it has had about 110 million sign-ups for its free version and has raised $90 million from investors including Andreessen Horowitz and Accel. Noronha said the roadmap now includes more graphic design, video generation and workflows that can update sales presentations automatically when new customer data shows up.
My take — AI-written commentary, not fact-checked reporting
Gamma is chasing the right problem: not making prettier slop, but making AI output feel less like AI output. That’s the real market now, because nobody wants a deck that looks like it was assembled by a cheerful template robot with a caffeine habit. The next battle isn’t who can generate slides fastest; it’s who can make the result survive first contact with a human audience.
Read more about this at: SiliconANGLE
