Unusual Ventures reposted this
🌊 The rising tide of enterprise AI SaaS adoption is almost here. So many startups have just had (or are having) an incredible Q2 2024. This comes at a fascinating moment. The frontier AI labs are having a "quiet" summer as they grapple with the limitations of deep learning. Analysts and investors who leaned in aggressively in 2023 are anxious at lack luster returns and high-profile acqui-hire outcomes - rightly so, AI has been a bloodbath for most big VC funds so far. Meanwhile, the enterprise buyers of AI are starting to figure out how to buy (and roll out) tools that have clear value and are giving them some solid productivity gains. They were asked to do more with less, they are giving it a go! And what's working is somewhat counterintuitive - here are some observations: 1️⃣ The extreme GTMs are working best. Either you have an incredible self-serve product that's being bought bottom up you have built a CXO relationship at a company with an AI slush fund. Partnerships with consulting firms is another path to power. Look at who's praising the tool in public - is it the end user or just the buyer? 2️⃣ While there are still products getting bought on demos, the companies that have focused hard on enabling outcomes have found their champions. Customers successfully activated 6 months ago are driving pipeline now. Get your case studies and referral requests lined up asap! 3️⃣ The use cases finally getting adoption are the same demos we all got excited about over 12-18 months ago. RAGs for support. Copilots. No-code LLM apps. Design generation. Language generation. This is how long it takes to form the right expectations, make the tool somewhat reliable and unlock budget. 4️⃣ Customers are not wasting time deciding if the SaaS tool they are piloting is a wrapper. They have now wasted many hours on trying to make LLMs work themselves or created sh*tty videos no one wanted to watch. They love products that get the job done - what models are being used is nearly irrelevant except if it bumps up against security postures they need to route around. 5️⃣ Service providers to companies have been on the leading edge of AI adoption are often pivotal to introducing traditional buyers to tools. While we declare that AI agents bring us the services-as-software revolution, we might not know yet what shape it will take. Huge congrats to the AI SaaS founders celebrating this summer and if you are still wandering through the idea maze - stick it out, it's going to be worth it.
Thanks for sharing this Sandhya. Exactly what we are seeing and helps cut through a lot of the noise being created by those not building businesses / not closely involved with any businesses. I’ve been saying for months to our customers — “we are in the business of driving outcomes, not selling software” — which takes nuance to understand — and thank you for spelling that out 💙
Dhiren Jani
1mo
I keep looking for SaaS vendors who can offer AI features for different categories of products (CRM, Security, Finance, Support, Supply Chain, BI and so on). Apart from the app assistant/copilot use case, is there any other that a SaaS vendor can offer?
Let's use the new term, "AiaS" for genuinely AI-driven service or platform to differenciate from SaaS. Personally I would not put AI or LLM application or tool in this category because of their lack of intectuall property.
Hassan Bari
1mo
Hey Sandhya Hegde How about your thoughts on FlashIntel AI driven GTM platform? Let's connect over a quick demo. It's a game changer in market.
I help Tech CEOs position their company for AI-led growth, scale their GTM & build teams | ex-Google, P&G
1moExcellent insights Sandhya Hegde Especially, on "extreme" GTM. I see bipolar scenarios. AI startups with either a direct path to a user (typically data engineer/developer) to improve their productivity or with direct relationship with a buyer (CTO, savvy functional CRO) with enough juice to implement a pragmatic use case (direct cost savings). Use cases where AI induced error correction is built in or not important. Bloom is off for esoteric use cases.