If you’re a startup reading this you’ve probably pitched a VC and then waited a really long time for feedback. When you do receive feedback, it may be in response to fundamental errors in your application that send you back to step 1. 

But now there’s another way. Early-stage VC firm Ada Ventures has launched an AI pitch deck review tool dubbed “AdaGPT“. 

The tool offers founders instant feedback on their decks, creating an anonymous, free-to-access sounding board for any founder to use before approaching investors. 

It’s the brainchild of  Michael Tefula, who’s been hired as Principal and Head of Product.

I met with Michael and Ada Ventures Partner Check Warner to learn more.

AdaGPT can assess whether a pitch deck aligns with Ada’s investment thesis, flag information gaps, and suggest potential improvements. 

Once assessed, all uploaded documents are immediately deleted, allowing founders to amend their decks before they are shared with the Ada team. Founders are encouraged to trial the tool and provide feedback to help further refine the offering.

A polymath leveraging operational expertise to redefine VC

Michael brings with him a wealth of experience, having held a senior investment role at Downing Ventures and participated in the Atomico Angels programme. He has also spent time in operator roles, including CFO at impact-investing platform CIRCA5000. 

Michael is a Chartered Accountant, published author, and has previously collaborated with the Ada Ventures team as part of their market-leading Ada Angels programme and as the firm’s “Investor-in-Residence”; helping the team to source deals. 

Over the last three years, Michael has made over 20 angel investments in early-stage startups — including Cino, who I published an interview with last week. He is also a board advisor for the Oxford Seed Fund. 

Warner characterises Michael as a polymath and “a real student of VC.” 

“He’s always been incredibly curious about the future of venture capital. How can it be done better? Which is quite exciting for us to have embedded within our team.”

She detailed: 

“A core value at Ada has always been about how we can rewire VC? How can we make VC a better product for founders?  And what’s lovely about having Michael in the team is he brings that cage rattling.”

Alongside his responsibilities as an investor, Michael will also be taking on the role of Head of Product for the firm. He shared:

“I joined fintech company CIRCA5000 at a Seed stage, and that experience was incredible because I got to work with founders very, very closely in building a meaningful product.“ 

The UK VC market is underserved by VCs with operational experience. The average in the US is up to 50 per cent and as high as 60 per cent for top-tier VCs. The UK struggles with a mere 8 per cent.

One of the things I took away from that experience was the need for speed, whether that’s in shipping products or just dealing with your investor and the people you’re working with.

I learnt that many investors are very slow in responding to founders, for instance. This also highlighted the need for more empathy for founders.”

During this time at CIRCA5000, beta-GPT 3 emerged, and it unlocked a new enthusiasm for tech in Tefula, who learnt to write code and automate various bits of his work. 

As a self-taught software developer, Michael will design and build new tools to aid the fund’s investment strategy and support founders.

Why is AdaGPT needed? 

Ada Ventures receives over 1600 press pitches a year. The firm Ada already has a detailed plain language investment thesis available to founders and has also published a Seed investment framework.

But as Warner explains:

“We cannot give human feedback to all 1600 applicants. So we want this product to share useful feedback.”

The product is allowing us to  scale ourselves and be more accessible and available to founders.”

Can gen AI detect your authentic self? 

One thing that piqued my curiosity is where AdaGPT would stifle applicant authenticity — we all know the difference between an email or pitch from the heart and one written using generative AI. Both journalists and investors value authenticity over a slick press pitch, hands down.

Warner asserts: 

“One of our company values is to be human, and connecting with founders is a core part of making investments at Ada. And that is something that we will always hold on to.

So when it comes to the product,  we will enhance our process of getting to the heart, the human, and the mission of the company. 

We want to spend more time thinking about how startups can service real customers and patient groups, not writing documents and doing paperwork.”

Streamlining not ranking

Michael stresses that AdaGPT is not numerically ranking or rating startups, noting.“LLMs are, by nature, probabilistic. And they’re also trained on large amounts of data from the internet.

“They also have some considerable biases in terms of how they behave. So if we would try and have a large language model score a company, it would take away from, you know, our ability actually to step in and say, this is what we as a team, from a human perspective, having met the company, having spent time with the founders.

We want to leave ranking and numbers to the human elements and only use LLMs where we think they can perform well.”

The importance of humility 

Warner shared that startups are considered to have several characteristics, including ambition, productivity, pace, and how quickly companies can ship things. But Ada Ventures also—unusually—looks at humility. 

So of course, I wanted to know, how do you measure that? 

Warner explained, “We try to understand if they see the world from only their point of view. Do they understand the end users and the customers and how they feel? Do they listen to us and reflect on our feedback, or are they just railroading through what they want to do? Other funds may not have that in their framework.”

“Genuinely, we think humble founders are founders who listen to their customers. They’ll listen to their employees. They’ll take on board feedback from investors and other people. So that’s one of the characteristics we use. 

Commerciality and fundraising are super important to us in terms of how we evaluate founders, as are grit and resilience.”

Ada Ventures also considers founder trajectories over time to gain a broader understanding of the founders and their backgrounds. 

Michael notes that such a focus on founder trajectory highlights the limitations of what you can do with predictive machine learning. 

“Data may say a founder from a particular university will be a great founder. But it fails to capture the context around their life trajectory. This is where we think that understanding a founder’s background and story is a crucial advantage in identifying the best founders. “

Warner stressed: 

“We think a lot about the dangers and pitfalls of automated data-driven sourcing tools that are out there, which are very reductive, historic, and backwards-facing. So, internally, we’re thinking about how to ensure bias is reduced.

How do we make sure that inclusivity is included? And how can we weigh up these characteristics to ensure we don’t fall into the trap of just pattern matching on what’s happened before? Because that’s what’s led us to where we are in VC today.”

The company has conducted its first pilots with investees and prosüective startups and received valuable anonymised feedback. 

Michael recounts:

“One really interesting bit of feedback was that it’s too fast. When founders often try to get meaningful feedback from investors, it takes a very long time to hear back from them, so this feedback was interesting.

We also want to ensure that the quality of the feedback is not shallow. So, for example, if AdaGPT is looking at competition, can it actually highlight how a particular startup is differentiated in the context of the competition from the competition side?

We’ve baked in those feedback elements, but it’s early days. We’re looking to get more feedback.“  

I was curious if this could inadvertently gatekeep great founders. We’ve all heard the stories of friends applying for jobs that they are qualified for — and know the right people — but fail to get past the first hurdle because of a recruitment site that uses GenAI.

Warner admits:

“That’s what we’re really nervous about. That’s why this will not be any part of our investment process. It sits separate from that. So it’s more just tooling.”

AdaGPT is just the first in a series of products that will expand access to Ada Ventures knowledge and services to a broader group of founders.

Warner stresses, “If you’re a founder building a company that broadens access to work, money or skills – do get in touch with Michael.”

Try out AdaGPT (beta) today.

Lead image: Ada Ventures: Check Warner, co-founder and General Partner, Michael Tefula, Principal and Head of Product, Matt Penneycard, co-founder and General Partner.