Erik Severinghaus is a serial entrepreneur who has submitted the highest peaks of the business world, selling four companies for hundreds of millions of dollars, and the physical world, successfully climbing Mt. Everest in 2018.
A year ago he identified the next peak on his list, joining Bloomfilter as co-CEO, with a mission of leveraging AI to redefine software development. In a nutshell, Bloomfilter uses AI drive process mining to help software development leaders understand what their teams are prioritizing, when it’s likely to be delivered, and how much it is costing.
We spoke with Erik about his experience building an AI-based company, as well as his insights and tips for founders building the next generation of generative AI startups.
1) Tell us about yourself and what you are working on at Bloomfilter
A year ago I joined Bloomfilter as co-CEO because I was so excited about the team and personally connected with the problem space we are attacking. I’m a fairly technical person (taught myself to code in 2nd grade) and even for me, it’s extremely difficult at an executive level to receive full transparency and observability into my development team.
So, in essence, I am building the tool I have always wanted. A platform to help me understand how resources are being prioritized, how projects are progressing, where are the risks and what are the challenges in building software.
2) How are you using AI in your solution?
Understanding the software development process is challenging, as it passes through a dozen different tools from ideation to deployment. We use AI to stitch together the process across multiple tools and to cleanse/augment the data from systems of record. We are also using Generative AI to improve our interface layer to our users, so we can make observations (and soon answer questions) in natural language.
Right now, we are using standard models and household names to drive forward our integration of AI across a number of different domains:
- Improving efficiency by stitching together multiple sources of unstructured data
- Improving the user experience my conversing with users in natural language
- Identifying process bottlenecks and risk points by having the models analyse the data to find issues and concerns
Step one for us is using our own tech team, many of whom have a background and experience in AI, alongside specialized researchers from Northwestern University. It’s obviously an area where we are continuing to invest, including potentially bringing on additional in house talent, and investigating fit-for-purpose models to augment our existing use of general purpose technology.
3) Do you think Generative AI is a buzzword?
No question about it, Generative AI is a buzzword! As with many hyped technologies, it may go through its trough of disillusionment, but the capabilities we are already seeing are far reaching and profound, changing the way that we produce and consume information.
We can compare this to other revolutions we have seen. The agricultural revolution moved us from a place of food scarcity to a reality in which most of us worry about consuming too much. Ditto the industrial revolution, and we are awash in a sea of mass-produced goods. During the past decade, the internet was already in the process of transforming society to an over-abundance of information, and AI will put that into overdrive. Content can now be produced en masse for effectively zero marginal cost which will further saturate the information overload that is all around us.
4) How do you think that Generative AI will change the way that new startups are built?
The cost of building new companies just keeps coming down, and I think Gen AI will continue to progress that trend. Lots of prototypes and MVPs in product, sales and marketing are going to have their first version done by AI.
As we build our startup, I am constantly thinking about the analogy of making sure I don’t invest in a lot of land-line infrastructure just before the cell-phone revolution. As we grow and scale, I don’t want to invest in a lot of business development reps to do basic customer outreach, I’d much rather invest in an architect to help me build AI automation around basic customer interactions.
AI still can’t write the type of thought-leading content articles we pride ourselves on, but I want to make sure our writing process incorporates AI to make it more efficient. A lot of the “doer” tasks that I would have hired people for previously can be automated, which leads to a smaller and more efficient organization.
I’m also really surprised at the direction of what AI is automating. If you had told me 2 years ago that I would probably be using the technology to automate away my customer interactions, but still be paying humans to do my accounting, I would have thought you were nuts! The problem spaces of what AI is able to solve in a practical way are evolving in a completely different direction than I would have anticipated.
5) What are your top tips for early stage founders building Generative AI based startups?
The only thing that might be growing faster than the demand for AI startups is the supply of them. Everyone is “AI” now in some form or fashion. But, the more things change, the more they stay the same. The purpose of a business is to create a happy customer, so as a founder you should focus on that rather than the buzzwords. And most importantly, remember that AI is a means to an end.
I have been building different companies with an AI or ML component for 15 years now. I have seen lots of different “cutting edge” algorithms and seen the space evolve in ways that are completely unpredictable. Everyone loves to “skate to where the puck is headed” but to use another sports truism, “predictions are hard, especially about the future.”
The one constant I have seen over and over again is that traction is king. Investors, and employees for that matter, want to understand if your company is solving real problems for real people, and because very few people are independently capable of validating the technology, they use your ability to get market traction as the key proxy. Solve problems for customers, show that they value those problems being solved, create traction from that, and the rest will take care of itself.
6) Tips for founders pitching investors GenAI ventures
Most of your investors won’t understand the nuances of your tech, but they will understand if you can get to traction.