It seems our VC friends over at A16z have finally caught up to something we’ve known all along – AI businesses are complicated, hard to set up, expensive to maintain and require a long view. In their recent article The New Business of AI (and How It’s Different From Traditional Software), the authors set the stage that AI is not the venture investment it was hyped up to be.
They are right!
AI is a bad investment, but we think we’ve figured out the way to make it a great investment.
Here at Verdant, we’ve been building AI applications and unlocking the value of data for enterprises and startups alike for many years, yes even before the iPhone! This experience has given us unique insights about the challenges of getting to revenue and creating a cash flow positive AI business. In fact, that’s exactly the reason why we’ve structured Verdant to be both a Startup Studio and Innovation lab.
Let’s take a look at some of the warnings to heed for investors looking to jump onto the hottest new emerging tech trend and how our model is poised to overcome those challenges.
Margins, margins, margins…
AI businesses take a lot more to get up and running, execute sales contracts and maintain than a typical software company. AI, by its nature, needs massive amounts of data to run, as well as computational power to run the complex mathematical algorithms that turn that data into applications. This adds up to extremely high cloud infrastructure costs paid by startups to AWS and Microsoft Azure, which have a negative impact on margins.
Verdant tackles these high computational load challenges in two ways. First, our long years of AI application development means we have lots of tricks up our sleeves. While many younger startups may lean on PyTorch or TensorFlow, AI’s most popular libraries, Verdant is able to pull from our extensive development toolkit built over decades and use mathematical insight to find methods with fewer compute cycles which translate into cheaper cloud costs. We leverage methods ranging from deep learning to reinforcement to early control systems, to game theory to genetic algorithms from the ’70s. Our text processing techniques are not using the same kind of math every time, we use parsing and our expertise to find out what really works without the same load.
Verdant has yet another trick up its sleeve when it comes to high cloud infrastructure costs that come with heavy compute. Our Innovation Lab partners directly with enterprises to help them turn their data into revenue positive gains through new products. In these cases, the enterprise will handle its own infrastructure costs and Verdant charges for the math.
Repeatability of Software Eludes AI
The partners at A16z have rightly discovered that AI startups should be structured as blended software and services businesses. We couldn’t agree more. They note that AI often lives in edge cases meaning that network effects can’t be exploited through the delivery of the same software code for multiple clients as most SaaS companies offer and thus require a “human in the loop”, which means thinking long term about the client relationship.
At Verdant, we have two sides to our business that act symbiotically to both service enterprise needs and leverage knowledge gained in the field to create new businesses with a customer in mind.
Building enterprise-grade AI products means we tackle three problems at the outset that haunt most AI startups:
- Access to data – Our corporate clients have massive amounts of data and Verdant helps turn that data into revenue-generating products for enterprise clients.
- Cost of data storage and processing – Verdant knows how to optimize for efficient data processing and those storage processing costs are already being covered by our innovation clients.
- Finding a market – Our “continuous innovation” is an interface between enterprises and entrepreneurs, enabling us to develop new IP and ideas founded in real corporate need.
The Long View
Structuring an AI business requires the internal expertise to build the right product but also a long term view of the client relationship that includes a services component. Our Innovation Lab is set up to service existing client needs, but also to create new ramps of products for future innovation and business opportunities.
For our studio investors, that means a suite of products built with markets in mind that go through a rigorous vetting process to come to life and supported by a team of experienced pros in the right disciplines. For Verdant this is key, if we don’t have the right team to take a product to market, we don’t take it to market until we do.
While AI investments may not yield the immediate 10x returns most VCs are looking for, we think our blended approach to AI development and corporate innovation is the answer to making AI investments attractive again.