Adding AI to SaaS: Inside the AI Product Strategies of Figma, Cloudflare, GitHub and Ramp

"He was at a design tool startup called Diagram, making AI plugins for Figma, and one day Figma’s founder Dylan called him to join Figma itself to build out its AI features."

At SaaStr Annual’s AI Summit, we gathered an all-star panel of product leaders who have built some of the most widely-used AI features in production today. The speakers included Mario Rodriguez, Chief Product Officer at GitHub at GitHub, Diego Zaks, VP of Design at Ramp, Dane Knecht, SVP of Emerging Technologies at Cloudflare, and Vincent van der Meulen, Design Engineer at Figma, and Dani Grant, CEO at Jam.dev.

Each leader started by sharing their unique perspective on how they approached adding AI into their SaaS products:

How GitHub Built an AI Copilot to 1.8M Paying Users

For those unfamiliar with it, GitHub’s Copilot is an AI pair programmer and code completion tool developers utilize to write code to completion faster. Since its original inception in 2021, Copilot has evolved and grown to 77,000 organizations and almost 2 million users. GitHub focuses on creating developer-centric tools, drawing from GitHub’s long-standing commitment to the developer community. Its founders and many of the current leadership team all started as developers.

As Mario, GitHub’s CPO explained, “ Where we are today is by generating value not through technology but by a product,and the story of Copilot is one of being in the right place at the right time.  And the second reason of why GitHub won was because the founders had taste. And that taste was them being dev-centric, being developers themselves, and creating a tool for developers.”

How Ramp Reached $300M in 3 years and Uses AI to Save 25,000 Customers a Billion Dollars

Ramp’s VP of Design Diego Zaks has a different take on AI in SaaS. He views AI not as the product itself but as a means to create magical user experiences. “AI is not the product – the product is the product, and AI is one of the ways in which we make people’s lives easier,” Diego emphasized. Their guiding principle is simple: “Does it feel like magic?” They focus on making technology disappear: “We ask not how can we make the experience better or slightly faster, but how do we make it go away? That’s really where AI shines.”

In order to get to that, first, Ramp looks at the root in real user problems. They ask themselves, not how can AI make the experience a little bit better, or slightly faster, they ask themselves, how do we make it go away? For them, that’s really where AI shines. There’s a huge range between using AI to book a flight to go to SaaStr to just showing up at the airport and it’s taken care of for you. That’s the end-game for Ramp.

How Cloudflare Added AI to It’s Already Leading $30B Company

Dane Knecht, SVP of Emerging Technologies at Cloudflare approaches AI infrastructure by focusing on edge computing and democratizing access. “We want to be able to make it so that everybody can build applications the way Cloudflare builds them, where we don’t have to worry about scale,” Dane shared. They’ve retrofitted their existing Cloud infrastructure to support this vision: “In the past year, we’ve retrofitted our 300 cities, 500+ data centers with GPUs, over 170 of them have them today.” Their goal is to use to AI to further support the use and development of AI in SaaS and Cloud companies.

How Figma Integrated AI plugins in its Ecosystem

Vincent van der Meulen, Design Engineer at Figma prioritizes complementing designers rather than replacing them. His time at Figma actually started outside of Figma, as a fan. He was at a design tool startup called Diagram, making AI plugins for Figma, and one day Figma’s founder Dylan called him to join Figma itself to build out its AI features.

“When we decided to set out to make this bundle of AI features for designers, we really wanted to focus on complimenting designers and not replacing them,” Vincent explained. Their approach involves constant iteration and evaluation: “Once you have a prototype, it still takes a ton of time and a ton of living with the prototype to get it right.” They’ve built features like AI search, automated design prototyping, and smart layer naming, all while maintaining their commitment to quality through rigorous testing.

#1 Product Roadmapping in the Age of AI

The rapid pace of AI advancement requires a different approach to roadmapping. Here’s how these leaders handle it:

• GitHub uses “strategic roadmaps” focused on key bets and learning objectives rather than fixed deliverables per quarter. As Mario explains: “We try to plan out in a strategic way, features for a year, knowing that anything that we say four quarters from now will be completely untrue. And even with our customers, we have a lot of enterprise customers that want predictability. But in AI, you cannot have predictability, LLMs are not predictable. So to that end, what we try to do is make sure that we are achieving the right value in the product.”

• Cloudflare divides AI innovation into three horizons. Dane describes their approach: “We have a core product organization focusing on what we need to ship to customers next quarter… then we have another group focusing on growing TAM 12-18 months out… and then we have a research team which really focuses on the fundamental technology.”

• Ramp stays flexible with quarterly execution plans. Diego emphasizes their design-influenced approach: “Designers are very comfortable doing 99 things, finding 99 ways that something does not work before something makes sense… We’re very comfortable not really knowing where we’re going to end up 18 months from now.”

• Figma uses hackathons like “Maker Week” to potentially reshape their roadmap. Vincent shares: “Right now, the big project I’m working on and that I expect to work on for the next year actually came out of a hackathon like four weeks ago… you don’t have to have these plans set in stone.”

#2. Experimentation and Quality Control

A common challenge with AI products is determining when they’re “ready” to ship. The panel shared their approaches:

• GitHub invested in “COFFEE” (Compiler Offline Evaluation) to benchmark progress

• Figma built custom visual evaluation systems that match their product’s needs

• Ramp focuses on velocity and rapid iteration, believing that being right 52% of the time leads to winning

# 3. Building Effective AI Teams

The panel revealed different approaches to structuring AI teams:

• GitHub distributes AI capability across product teams rather than centralizing it. Mario explains: “There shouldn’t be just one Copilot team. Every team is a Copilot team, the PR team is a Copilot team, the Issues team is a Copilot team… all the companies center now on that Copilot.”

• Figma maintains a multidisciplinary approach. Vincent describes: “We have a mix – we’re starting to realize that we do need a fundamental team of ML engineers but then we do have a lot of AI product engineers designers who are specifically focused on AI and researchers who focus on doing a lot of experimental AI prototypes.”

• Ramp focuses heavily on internal tooling. Diego shares: “Our applied AI team is constantly plugged in to whatever is happening on AI.. They are the inspiration for everyone else in the company to say ‘this part of my job really sucks, I would like it to go away.’ And then they can say, ‘Oh, try this model.'”

• Cloudflare structures its AI efforts in three layers. As Dane explains: “We think about it in three different ways internally: operational AI, what tools are we providing to our employees, our product teams, and then offering it as a developer tool for everyone else.”

# 4. The Future of AI Products

Looking five years ahead, the panel shared their visions for each of its product in the future years:

• GitHub aims to become more “AI native.” Mario envisions: “These LLMs can actually understand natural language in a way that none of the other tools before could… if we really lean into that and extract the maximum value of natural language… the product looks completely different.”

• Ramp wants their product to disappear. Diego explains: “We actually measure engagement time and we want that to go down. So hopefully five years from now, nobody’s using Ramp, it just works… doing everything in the background.”

• Cloudflare hopes AI becomes invisible. Dane shares: “I kind of hope AI kind of disappears as a thing that we talk about in the forefront and go back to the business value that we’re creating… I can’t even imagine what they’ll be able to do, as far as enabling us to just do our jobs and live our lives better.”

• Figma envisions a “role collapse.” Vincent predicts: “Designers will be able to create software, engineers will be able to create designs, and all of these roles are going to start blending together.”