Marissa Mayer is taking another swing at consumer software. The former Yahoo CEO and Google veteran has secured an $8 million seed round for Dazzle, a new AI venture that effectively rises from the ashes of her previous startup, Sunshine.

The financing values the company at $35 million post-money. It was led by Forerunner Ventures’ Kirsten Green, with a roster of Silicon Valley heavyweights filling out the cap table: Kleiner Perkins, Greycroft, Offline Ventures, Slow Ventures, and Bling Capital.

For Mayer, this is more than just a capital injection; it is a hard pivot. Sunshine, which launched in 2020 with a focus on contact management and photo sharing, struggled to find a mass audience. While the technology was competent, the product-market fit for a standalone contacts app never materialized.

From Sunshine to Dazzle

According to Mayer, the shift began internally during the summer of 2024. The team started prototyping AI tools to solve their own productivity bottlenecks and realized the side project had more potential than their core product. The premise was simple: most AI chatbots are too literal. They do exactly what you say, rather than understanding what you actually want.

That realization killed Sunshine and birthed Dazzle. Mayer framed the decision bluntly in recent statements:

“We realized this was something we were much more excited about. [It is an opportunity for] a much bigger impact.”

Dazzle’s goal is to build an interface that interprets user intent rather than just processing syntax. While the company is keeping specific features quiet, the broad strokes involve a platform that acts less like a command line and more like a human assistant.

The “Late Bloomer” Thesis

The investment from Forerunner is notable because Kirsten Green has historically avoided the infrastructure layer of AI, preferring to wait for consumer applications to mature. Green describes consumer AI as a “late bloomer”—a sector that has lagged behind enterprise solutions but is now ready for a UX overhaul.

The thesis here is that the underlying models (like GPT-4 or Claude) are becoming commodities. The value, Dazzle argues, will accrue to the interface that makes those models actually usable for normal people. Dazzle aims to solve three specific friction points:

  • Context: Understanding the history behind a request, not just the current prompt.
  • Tone: Moving away from the robotic, paragraph-heavy responses of current LLMs.
  • Habit: Creating a tool sticky enough to be used daily, not just for novelty demos.

Mayer’s relationship with Green appears to be the driving force behind the round. Mayer noted that Green “has a great sense for where people and platforms are going,” suggesting a shared belief that the current chat-box paradigm is a temporary bridge, not the final destination.

A crowded field and high stakes

Despite the pedigree, Mayer faces a steep climb. The “AI assistant” market is already saturated. Apple is integrating intelligence directly into the OS; Google and OpenAI are locked in an arms race for the dominant consumer interface; and countless startups are building “agents” for niche workflows.

Dazzle has not disclosed its business model—whether it will be a subscription service, an app, or an overlay on existing tools. It also faces the classic “wrapper” risk: if Dazzle relies on third-party models from OpenAI or Anthropic, it is vulnerable to those providers simply building Dazzle’s features directly into their own products.

For Mayer, the $35 million valuation is a modest, realistic reset. It isn’t a unicorn-style coronation; it’s the price of admission to try again. The capital buys her team time to prove that a veteran grasp of consumer product design can beat raw model power.