How investors and founders use AI in fundraising

Three investors sat on one stage at Climb26 at the start of July to answer the same question for founders: how do you raise money in the age of AI? They each gave a different answer.

From having it integrated into every step of the process, using it to give feedback to the founders they decline, or running a light-touch check of a company’s digital footprint, they all used it at varying levels.

What followed was a thought-provoking discussion around the levels of usage in the process. As Sarah Barber of Jenson Ventures put it: “We don’t just use past data to look at future companies, because past data alone rarely reveals the future diamonds.” That’s not an argument against AI in fundraising, but it does raise the point of being wary about solely making decisions based on the past data that the AI is scraping from.

The notion of investors using AI for screening is becoming a more prominent topic of conversation, having been discussed at earlier points in the year at Startup Magazine’s Hustle Awards and the Accelerate UK Venture Conference. This article aims to give you additional insight into some ways investors are currently using AI, how that influences what they expect from founders and ultimately how founders should be using it in response.

Full disclosure before we begin: we’ve built the AI for fundraising

ThatRound is a platform that uses AI to find founders their investor fit, and we collect the full picture from a startup at onboarding, the team, the traction, the terms, the context, precisely because we think a deck on its own or an online scrape misses too much. AI is only a part of what we offer, as we’re bringing the early-stage fundraising market into one place. We make it easier for startups to reach the relevant angel networks, groups and syndicates, and we send investors quality deal flow straight to their inbox.

So while we could use this article to pitch our platform, that’s not what it’s for. What follows is what investors have been saying at the events we’ve been to this year, so you get a clearer picture of how they’re personally using AI, separate to our platform.

‍

How are investors using AI?

A couple of the uses of AI that come up repeatedly in panel discussions are using it for initial screening and providing feedback.

Screening and filtering decks

With a large volume of decks coming in, using AI to screen and filter decks comes as no surprise. Jenson Ventures alone sees around 2,500 decks a year, averaging seven decks a day. No small team reads that properly by hand, so the initial screening is increasingly automated.

But using just decks alone to filter has gaps, and the panel was clear about them. “If you’re only analysing a pitch deck using AI, you are almost certainly missing something,” Barber warned. A deck is a compressed version of a company, and startups don’t always put everything that matters in it. The context behind a pivot, the traction that hasn’t reached a slide, the depth of the team: the nuance often lives outside the deck, and a filter that only reads the deck can’t score what isn’t there. This is why investors choose differently on the level of decision making they give AI in their screening process.

Providing feedback at scale

Others use AI to help manage the volume of decks they see. As Guy Remond from EHE Venture Studio put it: “We had around 400 pitch decks last year and could only choose eight companies. AI now lets us give every founder feedback. You just can’t do that with a small team.”

So it’s not just filtering: AI can be how a small team gives feedback to a mass of opportunities it could never respond to by hand. It’s also worth noting what this is a fix for: a pipeline of 400 applications for eight places is likely to be made up of companies that were never going to be a fit. Feedback at scale softens the problem for founders, but it doesn’t lessen the inbound for the investor.

‍

How AI changes what investors are expecting to see

A higher bar for the pitch deck

The panel discussed how the standard for pitch decks should now be higher precisely because an AI-styled deck is so easy to produce. But they also pointed out that a purely AI-generated deck is obvious the moment it’s opened. No story, no clear problem and solution, no data behind the claims. The tools raised the average, so investors look harder for the things the average can’t include. And the design bar rose with everything else: a deck that looks like it was thrown together in Microsoft Paint now reads as a signal you don’t take your own raise seriously.

Further along the process: POC and market research

Building a proof of concept and testing the market is now cheap enough to do yourself, so investors expect you to arrive with validated market research and a POC that shows what you’re building and why you’re raising. Tech Nation’s latest report found 30% of founders say their business would not exist without AI [2], and when that many companies can get that far that fast, turning up with an idea alone no longer stands out.

Rounds have got smaller too. With the cost of software engineering down, founders simply don’t need to ask for as much funding for engineering, which changes what a sensible ask looks like at pre-seed and seed.

‍

Added importance on the digital presence for founders

The screening doesn’t stop at your deck. At Startup Magazine’s Hustle Awards, Lawrence Rosenberg, Founder and CEO of Rosenberg.Media, gave a keynote on exactly this. He’s spoken to multiple investors who are using AI for basic due diligence long before a first call: “Investors are already using AI to form an initial view.”

These tools scrape the internet, so what’s publicly out there about you and your company feeds that first impression. Your LinkedIn activity, your website, your coverage. If the AI comes back with thin or outdated information, that’s the version of you the investor meets first.

The practical move is to treat your digital presence as part of the raise. Keep the basics about your company current and correct wherever AI will look, keep your story consistent across channels, and show up where your investors actually are, which for most UK founders means posting on LinkedIn.

‍

Why the need for relevance is even stronger than before

More companies are competing for that first pass than ever. 2,489 UK companies raised their first-ever equity round in 2025, up 23.6% on the year [3], and more than 680 AI startups launched in 2025 alone [2]. With AI helping to automate cold outreach, investors are receiving more deal flow than ever.

Which brings us back to Barber’s other warning: “Many decks come through with clearly no research into who we are or what we invest in, so worth doing some homework into your potential investor to ensure a good fit.”

When an investor’s inbox is this full, relevance is the filter before the filter. You’re competing against hundreds of other decks, so you need to show why you’re relevant to this investor specifically. The fastest way to lose is to make them work out why you’re in their inbox at all. Sending decks to irrelevant investors is usually what AI-assisted outreach produces when it only has public information to work from. An investor’s actual thesis, stage appetite and preferences mostly aren’t published anywhere, so the applications can be way off the mark, and the investor’s pile keeps growing.

‍

What AI can’t fake

If AI has raised the floor, the question becomes what still separates companies at the top. The panel kept returning to the same things.

AI can’t fake your defensibility. “AI has democratised software engineering, so defensibility is a major must,” Remond said. Three protections came up: patents, regulation (a moat that can buy you around two years), and unique datasets that are genuinely hard to replicate. Alongside that, the human ingredients investors have always screened for: deep domain expertise and a strong commercial edge.

AI can’t fake your honesty. Claiming you have no competition comes up regularly as a big red flag on multiple panels. If nobody else is doing what you’re doing, it’s usually either not true or not that valuable. Pivots are fine, but obscuring your journey isn’t, because it comes out in the wash anyway.

And AI can’t fake your ability to pivot. When AI shifts the ground this fast, investors aren’t really backing your current product. They’re backing your ability to change direction when the ground moves. Nik Storonsky’s original idea for Revolut came out of his time on the Credit Suisse FX desk, and the business he ended up building looked nothing like where he started. We’ve heard the same theme on several investor panels this year, including at the Accelerate UK Venture Conference, and when separate rooms of investors keep landing on the same conclusion, it’s a point worth listening to.

‍

What founders should actually do with it

A few bits of practical advice from the panel on the day:

• Use AI to build the list, then rehearse. “I’d advise founders to build a shortlist of investors, use AI to narrow it to your top ten targeting funds genuinely interested in your niche, then work from the bottom up, saving your number one for last so you’re confident and pitch perfect by the time you reach them.” That was Remond’s playbook, and it treats early pitches as what they are: rehearsal
• Interrogate your own deck first. Role-play as an investor and pull your deck apart before anyone else gets the chance
• Check the investor fit before you send anything. A researched shortlist of 10 beats a spray of 100 every time

‍

AI is only as good as the context you feed it

The most useful framing from the whole panel wasn’t about tools at all. It’s a division of labour. AI does the volume work so humans can do the conviction work. Investors are using it to process more decks so they can spend real attention on fewer, better conversations. Founders should use it the same way. And what separates good volume work from bad is the context behind it.

It’s fair to ask why you’d need ThatRound for that when you could run your own investor research in Claude or ChatGPT. The honest answer comes back to Barber’s warning: a general AI tool can only see what’s public, and the context that actually decides fit mostly isn’t published anywhere. An investor’s thesis, stage appetite and preferences live in their head, not on their website. And a startup’s story, traction and terms rarely fit in a deck. So we collect that context from both sides directly. Startups give us the full picture at onboarding, and the angel groups, networks and syndicates on the platform tell us exactly what they want to see, with preferences refined over time. Investor fit is found based on what both sides actually said, not on what a scrape could find.

And that’s as far as the AI goes. It carries you up to the introduction, then hands over, because the conversation, the relationship and the decision to invest are conviction work, and conviction work stays human.

Feed the AI real context and the division of labour works: the volume gets processed properly, the right people end up in the room, and what happens next is down to you.

‍

References

1. How to Raise Money in the Age of AI | Climb26 (Stage 1), 1 July 2026
2. Tech Nation Report 2026 | Tech Nation
3. The Deal 2026 | Beauhurst
4. Lawrence Rosenberg Keynote | Startup Magazine Hustle Awards 2026