Predicting the future: What top AI founders have to say about innovation in 2025

The age of AI innovation is here. Teams across industries are working to refine product strategies to leverage a “holistic AI approach” that meets the needs of their customers today and into the future, according to our 2024 Insights Report. But with such a rapidly changing space, it can be hard to know what is a passing trend, and what innovations and updates will quickly change the way products are built and teams work together. 

Regardless of uncertainty, the benefits of AI integration are clear: 

• 78% reported time savings for business and customers
• 60% reported cost savings for business and customers
• 68% reported improved customer productivity

In our research report, top AI leaders also describe the race to integrate AI. One respondent, a Head of Product, explains how “anyone who’s not leveraging AI will start to fall behind. It’s about harnessing it correctly though.” A Founder responds, “we have AI working to solve issues on several fronts—creating a revenue model that is profitable for us.”

At AssemblyAI, we’ve seen the use of our APIs grow nearly 300% just over the past year—with two-thirds of AI product leaders reporting that they are still in the preparation stage of implementing Speech AI and multimodality into their applications. 

When asked how they are implementing and innovating with AI, founders pointed to three main use cases: 

• Customer Experiences (68%): product recommendations, predictive text, virtual assistants, automated scheduling, smart home devices
• Speech Intelligence (59%): speech-to-text translation, sentiment analysis, call summarizations, intent recognition, tone and emotion detection, keyword extraction
• Marketing Processes (49%): predictive segmentation, content generation, email campaigns, automated social posting, real-time ad bidding, lead scoring

In our recent Assembly Required series, AssemblyAI founder and CEO Dylan Fox meets with other AI founders to discuss insights and learnings gleaned from building an AI-first company, as well as these leaders' unique take on what’s needed to truly unlock and capitalize on AI innovation. 

Here’s what they had to say:

In order to truly unlock innovation in AI, models need to be easy to integrate, use, and deploy 

While much of the conversation on AI models has been focused on their power, accuracy, and the opportunities they unlock, true innovation in AI cannot take place until barriers—to integration, user-friendliness, accessibility, etc.—are broken down or removed entirely.

Ben Firshman, Founder and CEO of the open-source developer community Replicate, describes how this process took place with LLMs. Llama, he explains “caught the imagination of the open-source community,” but its usability was hindered by licensing issues, so those building with it couldn’t create applications for commercial use. But when Meta released Llama 2, he continues, “that’s when [its use] really took off, because Llama 2 was much better and it was possible to use in products as well.” 

From there, companies could really start playing with and innovating with LLMs—and build them into production-ready applications and products. 

AI innovation is happening fast, and the speed continues to accelerate 

AI innovation is happening fast, with updates and improvements releasing over weeks, rather than months and years. For example, in a matter of two weeks, AssemblyAI pre-announced Slam-1, a one-of-a-kind promptable, production-ready speech language model that enables rapidly customizable speech-to-text and speech understanding tasks, and AI research company Anthropic announced Claude 3.7 Sonnet, the first hybrid reasoning model on the market. Industry-disrupting news like this happening within days of one another is becoming the norm, not the exception, when it comes to AI. 

Rather than feeling overwhelmed by the change, forward-thinking founders are loving the pace of innovation. Jason Boehmig, founder and CEO of AI-powered contract management software company Ironclad describes how energizing this fast rate of change can be: “Thinking back to nine months ago feels like almost as long ago as starting the company, which was ten years ago, like a totally different state of the world. And it's also giving me a ton of energy as a founder,” he describes. 

Edo Liberty, CEO and founder of vector database platform Pinecone, describes how even as an AI founder, you can be surprised by just how much better an AI model or application can become over the space of a few months.

“You build a new model or you build some new serving infrastructure, put it in production, and then you realize this thing's actually like super efficient. Okay, now we got to go get this thing to market. How do we think about that?” he describes. “Oftentimes you're surprised by just how much the rate of change is and how much better things are getting.”

While innovation comes from solid strategy, it also relies on foresight and a little bit of luck

Replicate’s Ben Firshman also touches on how both foresight and luck have to come together in order to build successful, innovative AI products. What users want to build with AI, and what AI is capable of supporting, may not always align—but when they do, the results are impactful.

He gives the example of AI avatar generators and ChatGPT: “People wanted to build image editors, they wanted to build AI avatar generators, they wanted to build generative games and all these kind of things just around the time of ChatGPT where there was all this interest,” Firshman explains. “And then it was just this sort of perfect storm of the supply and demand that just helped us grow.”