Looking Back: How Last Year's Predictions Measured Up

By Mike Fridgen, Keith Rosema, Henry Huang & Jason Flateboe • Dec 11, 2024

Last year, we came together to share our predictions for the future of innovation in 2024. As we gear up for 2025, we wanted to reflect on and check in with our forecasts. How did the year play out, and what lessons can we carry forward as we explore the next wave of innovation? Here's a look at what we saw coming and how it unfolded.‍

Mike Fridgen: Open Source & Ethical AI

Last year's prediction:  
Mike predicted that the open-source space would be an interesting area to watch. In 2024, he believed more companies would develop tools and ecosystems around open-source generative AI models. Platforms and tools that enhance the accessibility, usability, and customization of open-source AI models would take off.

Did we get it right? Mostly. Open-source AI models, such as LLaMA 3.1, gained prominence and are now widely used in applications ranging from drug discovery to creative production. Over the past year, new tools supporting fine-tuning, customization, prompt engineering, and AI monitoring emerged. At MVL, we even began building a property and agent debugging toolkit. Despite this momentum, proprietary models still dominate many applications.

Last year's prediction: Mike also expected an increased focus on ethical, transparent, and responsible AI. He believed areas like bias mitigation, data privacy, and ethical AI frameworks would attract investment. During 2023, MVL had already made several bets in these areas, and he was optimistic that more would follow.

Did we get it right? Yes and no. Privacy-enhancing technologies, such as federated learning, advanced and enabled data processing without exposing sensitive information. Yet, clear AI standards and frameworks remain elusive.

Last year's prediction: Related to his earlier prediction, Mike believed more funding would emerge for startups that offer solutions for managing and protecting AI-generated intellectual property. This included software for tracking AI-generated content, verifying originality, and ensuring compliance with IP laws.

Did we get it right? No. Direct funding for startups exclusively addressing AI-IP management was limited. Instead, investments focused on adjacent areas like compliance, legal tech, and cybersecurity, indirectly supporting AI-IP management infrastructure.

Henry Huang: Prompt Evaluation

Last year's prediction:
Henry was banking on prompt evaluation emerging as the most significant area of GenAI investment in 2024. Managing prompts and the returned output is critical to the success of nearly every GenAI company, and he believed this presented an excellent opportunity for tools and platforms that generate, process, train, and test prompting systems.

Did we get it right?
Yes and no. The most significant investments in GenAI in 2024 were still the models themselves, with at least 12 state-of-the-art LLMs released, many of which are multi-modal. Other areas, including vertical applications of agentic capabilities, OS toolkits and SDKs, and content generation tools, also saw substantial growth. Despite this progress, prompt evaluation remains a barrier to more widespread adoption. To address this, MVL is incubating a company called Impromptu, led by Keith.

(If you want to learn more, contact Keith at keith@madronavl.com!)

Keith Rosema: AI at the Edge & Big Tech’s Revival

Last year's prediction:
Keith was excited about AI at the Edge. He predicted that 2024 would see a strong push toward AI at the Edge, especially in cars, phones, smart homes, manufacturing, and healthcare: "We will finally be able to tell our web cameras to spot wildlife, not the trash collector, or even have our doorbell cam send out the robot to chase the porch pirates."

Did we get it right?
Yes. AI detection has become a staple of webcams, automotive imaging, and lidar. In healthcare, AI is now embedded in artificial limbs, making them more effective at operating as their users need. While home-based watchdog robots haven’t arrived yet, UPS announced AI algorithms for fighting porch piracy!

Last year's prediction: Keith also believed we’d see the next generation of social media platforms. Vertically siloed social groups are growing, and Meta and X feel dated. He thought this presented an opportunity: the 2024 ecosystem could be built around creators and influencers, catering to how a new generation looks for and participates in entertainment.

Did we get it right? Not yet. While platforms like Bluesky and Noplace and tools like Creator.co are more creator-focused, there’s still room for significant development.

Last year's prediction: Keith also believed it would be a year when big tech revisited its relationship with the entrepreneurial world. In the wake of significant layoffs at large enterprises, Keith believed these organizations would look to startups for freshness and innovation.

Did we get it right? Absolutely. Nvidia acquired Madrona’s Octo AI for $250M in September, and AMD acquired Silo AI for $665M. Companies like Inflection (Pi), Perplexity, and Codeium charted new paths in AI chatbots, followed by larger players such as Salesforce and Servicenow.

Jason Flateboe: Hyper-Personalized User Experiences

Last year's prediction:
Jason expected to see significant strides in user experience in 2024. He predicted that hyper-personalized experiences would continue evolving and become the norm as customers increasingly expect customized interfaces and content driven by unique needs and preferences.

Did we get it right?
Yes? We are seeing more products that are highly personalized. Jason spent a good chunk of the past year as a product designer on Otto, an AI business travel agent spun out of MVL. It was exciting for him to see what a highly specialized agent can do. From the first moments of onboarding, Otto starts to understand a user’s travel preferences by ingesting the historical data from their calendar. Otto knows which airport you fly from, which airline and hotel brands you prefer, and even if you like aisle seats or window seats. Once Otto knows those preferences, it will never ask you again. It can automatically plan a trip for you and even book the flights and hotels—just say “book it,” and Otto completes the transaction.

Stay Tuned for 2025


Be on the lookout for our 2025 predictions dropping soon. Follow us on LinkedIn and subscribe to our newsletter below to learn about all things MVL. If you're a founder experimenting in these lanes or crafting something unique, we'd love to chat with you! Learn more about MVL by delving into our manifesto here.

Recent stories

View more stories

Let’s start a company together

We are with our founders from day one, for the long run.

Start a company with us