With Fund II largely deployed across sixteen portfolio companies, this felt like the right moment to write an honest account of what five years of backing AI-native enterprise software has taught us — what our thesis got right, where the market surprised us, and how we think about the next phase. This isn't a victory lap. The fund is too early in its lifecycle for that kind of accounting. It's an effort to be transparent with the founders who trust us with their most important early relationships, and with the limited partners who make this work possible.
The core thesis — that enterprise AI adoption would run ahead of the infrastructure layer, creating durable investment opportunities at the primitive level — has proven directionally correct. The companies we backed that are doing best are infrastructure businesses: they sit in the critical path of workflows, they accumulate data gravity, and they have net revenue retention above 130%. The companies that have faced more headwinds are the ones we backed on market timing arguments rather than infrastructure arguments — they were right about the market evolving in a certain direction, but they didn't have the stickiness that infrastructure creates.
Where we were wrong
We underestimated how long enterprise procurement cycles would be compressed. In 2022 and 2023, AI urgency created procurement exceptions at large enterprises that we assumed were durable. CISOs approved pilots faster. Finance teams allocated AI budget outside the normal capital cycle. Boards pushed for AI roadmaps as a strategic imperative. We wrote a few investment memos that leaned on "procurement is faster now" as a structural assumption. That compression has partially reverted. Enterprises are more deliberate now, security reviews are back to normal length, and the strategic AI urgency is being filtered through normal capital allocation processes. Companies we backed with short sales cycle models had to adjust.
We backed too few pre-revenue companies. Our Fund II mandate was primarily Seed and early Series A. In practice, we migrated toward companies that had early revenue traction because the metrics were legible and the risk felt lower. Looking at the fund construction in retrospect, we would have benefited from more pre-seed bets where the technical advantage was clearer even if the commercial path was more nascent. The best outcomes in this category are disproportionately driven by technical moats built before the market is fully formed.
What we're carrying forward
The conviction we're taking into the next fund is that the AI infrastructure build is still in an early phase. The categories that will be most valuable in 2028 — human-in-the-loop workflow infrastructure, AI decision audit and governance, enterprise context management — are being built now, often by teams that are pre-revenue and working in the space between established categories. That's where we're spending most of our time, and it's where we believe the Ridgepoint investment approach is most differentiated.