We held a first close on Kvistlund Fund I in March 2021, at approximately 60% of target. I want to write down what that process revealed — not the mechanics of VC fundraising, which have been documented elsewhere, but the specific things we learned about LP appetite for a European logistics AI thesis that I didn't fully anticipate going in.
The fundraise took longer than I expected and produced a sharper fund strategy than I'd had at the outset. That's the honest summary. The longer version is below.
What we were selling and why it was a hard sell
The fund concept at launch was: seed-stage, AI-native logistics software, European geographies, operator-background team. The pitch combined three things that European institutional LPs in 2020 and early 2021 each treated as distinct risk categories. Seed stage meant high failure rate per company. AI-native meant technical bets that required understanding the model quality layer of supply chain software, not just GTM. European logistics meant an unfashionable vertical without the headline SaaS multiples that were driving LP enthusiasm elsewhere.
We got every version of "why not US logistics tech?" in the first ten conversations. The honest answer — that European operators buy from European vendors, that cross-border freight complexity in the EU creates a category of problem that North American logistics doesn't have, and that Nordic operators are better early-adopter customers for this type of product than the US mid-market — took fifteen minutes to explain and required LPs to accept a thesis about European logistics market structure that they hadn't previously formed opinions on. Most hadn't. They were comparing us against US-focused climate tech funds and enterprise SaaS funds where the category validation was already established.
The conversations that converted were the ones where we could put a specific company in front of an LP early in the process. Abstract thesis about European logistics AI is hard to evaluate. A specific seed-stage company with a Nordic 3PL as a paying customer, a founding team with 12 combined years of operations engineering experience, and a product that demonstrably reduced carrier selection time by 40% — that's evaluable. We made three investments before first close specifically to have those conversations available.
What LP conversations taught us about the thesis
The pushback we received most consistently was not about logistics or Europe but about the AI layer. Several family office LPs who had been early backers of enterprise SaaS funds were skeptical about whether machine learning would actually replace rule-based systems in logistics operations, or whether it would end up as an optimization layer on top of existing ERP and TMS infrastructure. This is a real question, and it forced us to be much more precise about what "AI-native" means in practice.
The distinction we landed on — and have held consistently since — is between AI as a feature and AI as the core decision surface. The former describes most of what the established TMS and ERP vendors were shipping: anomaly detection modules, predictive maintenance alerts, demand forecasting widgets bolted onto existing rule-based systems. The latter describes companies where the prediction model is the product — where the output of the ML system directly drives procurement decisions, routing choices, or inventory positioning, without a rule-based intermediary translating model confidence scores into operator actions.
This distinction mattered for LP conversations because it changed the competitive moat question. If AI is a feature, every incumbent can add it. If the model is the decision surface, a startup with a superior model that improves with each new customer's data has a compounding advantage that incumbents can't replicate just by licensing a model from a third party.
The LP base we ended up with
Fund I closed with a mix of family offices with manufacturing and distribution backgrounds, one Nordic institutional fund-of-funds, and several individual LPs who were current or former logistics executives. That composition was not accidental. By the second half of the fundraise, we had stopped spending time on LPs who lacked direct experience with supply chain operations. The conversations were more productive when the LP could evaluate portfolio companies directly rather than relying on our description of the market.
We are not saying that institutional LP capital from generalist fund-of-funds is less valuable — it isn't. We are saying that for a thesis-stage fund in a technical vertical, LP selection matters more than LP quantity. An LP who has spent 20 years running procurement at a European FMCG company will notice things in a portfolio company's product that a generalist investor will miss. That quality of LP feedback has been a genuine asset in our first deployment period.
The institutional fund-of-funds that came in at first close did so specifically because of the Nordic angle. Their LP base had significant exposure to Scandinavian corporates, several of whom had supply chain digitization programs that created alignment between our portfolio and their LP relationships. That context mattered for how quickly they were able to evaluate the thesis.
How the fundraise changed the investment strategy
We entered the fundraise expecting to write seed checks across a broad set of logistics AI categories: visibility, routing, forecasting, procurement, warehouse automation software. The LP conversations produced enough friction around the breadth of that scope that we narrowed considerably before first close.
The narrowing was operational as much as strategic. Fund I at target size — €22M — is not large enough to maintain meaningful positions across twelve-plus category bets. A seed fund of that size generates real returns if four to six of its investments become category leaders. That means the selection has to be more concentrated than the original thesis implied. We cut the number of initial target categories from seven to four, focusing on freight visibility, ML-based TMS, demand-side forecasting, and warehouse intelligence. Each of those categories has enough depth to absorb multiple investments across different market segments.
The fundraise also clarified our follow-on strategy. Several LPs asked specifically about pro-rata rights and follow-on reserves. We had not initially reserved capital for follow-on, expecting that later-stage investors would take Series A positions without meaningful Kvistlund participation. The LP feedback pushed us to maintain approximately 25% of committed capital as follow-on reserve — not because we plan to lead Series As, but because maintaining pro-rata positions in the best-performing companies matters for fund economics at this scale more than it would at a larger fund.
What the first deployment period confirmed
By the time of first close, we had made three investments and were in final diligence on two more. Each of those early investments has since provided some evidence about which parts of the thesis were right. The freight visibility company has grown ARR nearly 3x since seed. The ML-TMS company hit its first €1M ARR milestone faster than its founders' own model projected. The demand forecasting startup ran into enterprise procurement cycles longer than we'd anticipated — which validated the LP concern about enterprise adoption velocity, though the product quality signals remain strong.
The pattern that stands out from the first deployment period is how much the operator backgrounds of the founding teams matter for navigating those enterprise cycles. Companies where the founders have specific buyer relationships — not just category expertise, but named contacts at specific logistics operators who trust the founding team from prior working relationships — have moved through pilots to production contracts faster than companies without those relationships, even when the product quality is comparable. That's not a thesis change. It was in the original thesis. But it's been confirmed clearly enough that it will remain a hard requirement in our diligence checklist for Fund II.