Until now, generative AI has been akin to fairy dust — just sprinkle it in your pitch, and investors are all too ready to open their wallets wide.

This is changing. 

Q4 of 2023 saw a 55.3% decline in AI funding compared to Q1, mirrored by fewer deals. Experts like Matthew Marwick from the Intelligence Unit at CB Insights warn that while the AI funding will likely pick up steam again in the following quarters, the bloated investment rounds typical for 2021 and 2022 are now off the table.

The dust after the initial AI boom has somewhat settled, exposing the first challenges. Many AI companies that raised easy rounds during the funding boom didn’t achieve the commercial success investors expected.

A mere mention of generative AI is no longer enough to raise massive rounds. Investors now seek strategic defensibility and differentiation in your stack against incumbents and new entrants flooding the market. They want long-term market value

The question is, in a world where everyone pitches AI, how do you cut through the crowd and attract investment?

The differentiation and competitive moat key to AI funding success

Foundational models have been instrumental to the progress of the space. The problem arises when startups begin to pitch the same generic underlying models as their core value and competitive moat.

This approach is deeply flawed. The investor view is that AI and LLMs possess no inherent moat or competitive advantage — just like anything that is available to essentially everyone. If your company isn’t explicitly centred on researching, developing, and distributing foundational models, your moat and product differentiation must lie elsewhere.

AI should act as an amplifier of a pre-existing product or strategy differentiator. It is a supporting infrastructure for AI-first and AI-enabled startups that magnifies the initial value you deliver to the user — making the process cheaper, faster, more precise, and enhancing quality. But it doesn’t create value out of thin air. 

Cool AI features mean nothing if they don’t help to meet the needs of a specific persona, elevate user experience, or address specific use cases. This “AI for the sake of AI’s” approach most companies have mistakenly adopted has spawned a phenomenon termed “AI tourism,” where users, not witnessing substantial value addition to their lives or workflows, don’t stick with the product. 

Lesson learned: presenting an interesting case for investors will mean diving deep into the tangible use cases and value your AI technology will create. 

Companies with vertical data moats will lead the next wave of funding

Vertical AI — software customised for a specific industry and its problems — is gaining momentum among investors and end-users. Index Ventures partner Paris Heymann called vertical AI “the next logical iteration of vertical SaaS” that might act as an industry’s game-changer.

By delivering high-quality results on specific use cases without generating irrelevant BS, such tailored models increase time-to-value and user retention. This ability will make vertically integrated apps key in driving differentiation.

Unique data sets, customised go-to-market plans, and the capacity to deeply integrate into user day-to-day processes make them hard to replicate and promise immense potential. 

Looking at the investment landscape, we already see this trend playing out massively in data-heavy industries like legal and healthcare — but it will surely play out in other sectors as well. 

The core mistakes of pitching AI — and how to avoid them 

So, what does it mean for founders looking to raise money for AI solutions today? And how do you make sure your pitch lands in just the right spot? In 2023, we at Waveup worked on over 20 successful (and not so much) AI fundraisers, witnessing firsthand the difference between AI winners and losers. 

Here are the mistakes we see companies make when pitching AI and how to fix them:

  • Not representing your team’s skill set correctly. Your team’s structure must reflect the AI-first nature of your company and have deeply skilled AI scientists or people with prior experience in building AI applications. 
  • Not demonstrating vertical-specific GTM models using AI. Commercialisation is a big factor now in AI, so investors want to see a clear GTM motion with a pronounced distribution moat. 
  • Not knowing the cost. Too many companies have no clue about the inner workings of their economics, but this cluelessness isn’t sustainable in the AI space, where building, training, and deploying AI models can cost an arm and a leg. At a time when VCs put capital efficiency on a pedestal, you’re expected to understand your cost structure, needed resources, and gross margins in and out. 
  • Not knowing what you are building.  Be ready to walk the talk and prove you know the underlying tech. Here are some things you will need to explain to investors:
  1. Your model architecture and the techniques you use to improve it
  2. The process of collecting and growing your datasets
  3. The ways you measure your models’ accuracy and performance (benchmarks, performance baseline)
  4. Accuracy requirements for your model to be valuable for the user
  5. The strengths and limitations of your model

So… what’s next?

The AI race is underway, and we are all part of it. A prevailing ambiguity surrounds the AI application market, and while investors don’t place their bets just yet, they do, however, eye every AI opportunity that comes their way more thoroughly than ever before.

For early AI companies, defensibility and durability become the order of the day to attract funding. The casual mention of AI in your pitch deck no longer suffices to instigate serious investor engagement — you must know your technology in and out, understand its value for the end customer, have a preexisting moat, and show proof that you can execute on the highest level. Do this, and every VC door will open for you.

Source link