2 months to refresh a brand, website, and product for a $5M Seed TTS startup

Gan.ai is a San Francisco company that makes AI video and voice tools. Its core product lets a team record one video and personalise it for hundreds of viewers, changing a name or a company in the speaker’s own voice and lip movements.

Founded by Suvrat Bhooshan, who worked at Facebook AI Research, Gan.ai raised a $5.25M seed round led by Surge in 2023. In 2024 we refreshed its brand, redesigned the website in eight weeks for a Product Hunt launch, and redesigned more than 50 product screens.

Client

Gan.ai

Industry

Generative AI, video and voice

Services

Brand refresh · Website design · Product design · Icon set · Motion design

Timeline

2024 to 2025

Gan.ai had grown from one product into a set of tools: personalised video, text to speech, voice cloning, lipsync, AI landing pages and a Chrome recorder. The site and the app had to explain all of them, quickly.

01

Many tools, one story

Marketing teams, creators and enterprises each come for something different. The site needed one clear story with room for each.

02

Hard to picture

A video that changes for every viewer is easier to show than to describe, so the key ideas needed visuals, not paragraphs.

03

A launch date

The new website had to be ready for a Product Hunt launch eight weeks out.

The new site leads with the outcome, The End of Generic Video Messages, and sets out the tools in one bento: lipsync and voice cloning, AI landing pages, the Chrome recorder, personalisation and integrations.

We animated the bento so each feature shows itself at work, in the light, violet-accented look we carried into the product.

Further down, each section explains one idea with one visual: a name dropped into a sentence, a video landing page made for one viewer, a contact list arriving from a CRM or a CSV file.

Visitors can try it on the page by typing their name to generate a preview, and How it works comes down to five steps, from recording the video to sharing personalised copies at scale.

A set of icons drawn as small violet tiles ties the website to the product, one for each tool.

In the product, we redesigned more than 50 screens across Gan.ai’s studio: text to speech, webcam recording, avatars and personalisation variables.

Every tool shares the same light layout, rounded cards and one black main action, so moving between tools feels like using one app.

Lipsync and video generation follow one pattern: upload a video and an audio file, pick a model, set the length and how closely to follow the prompt, then preview the result.

Generated videos stay in a list next to the preview, so a team can compare takes before sharing one.

Voice agents get their own cards, each with a face, a name and a sample to play. Text to speech sits in one large field, and the plans page keeps the choice to two tiers.