I spent most of January doing a structured sweep of the warehouse software market — talking to operators, mapping vendor categories, and trying to understand where the actual white space sits for seed-stage investment. This is what I found.
The short version is that the warehouse software market is layered in a way that creates distinct investment opportunities at different levels of the stack, and that most of the noise in the category conflates layers that have very different competitive dynamics. The longer version requires working through each layer in sequence.
The WMS layer: entrenched and not going anywhere
The warehouse management system market is dominated by a small number of vendors — Manhattan Associates, Blue Yonder, SAP EWM, and Oracle WMS at the enterprise tier; Korber, Infor, and HighJump at the mid-market. These products have been in production at large distribution centers for 10 to 25 years. The switching cost is not primarily software licensing; it's the embedded institutional knowledge about how each facility's physical layout, labor categories, and carrier relationships are encoded in the WMS configuration. A large 3PL moving from one enterprise WMS to another is typically looking at an 18-to-24-month migration project that will consume the attention of its best operations engineers.
This creates a structural reality that seed-stage investors need to accept: you are not going to fund a company that replaces a Manhattan Associates installation at a 400,000 square foot distribution center in Hamburg. The WMS replacement market exists, but it moves in generational cycles — roughly ten to twelve years between major platform transitions at the enterprise tier. The last major wave was the migration from on-premise to cloud WMS, which is still incomplete at many large European operators. The next wave, whenever it comes, will likely be driven by a combination of AI-native decision surfaces and robotics integration requirements that current WMS architectures handle poorly.
We are not saying WMS replacement is impossible. We are saying it is a decade-plus sales cycle that isn't a seed-stage bet. The investment opportunity in warehouse software for a fund of Kvistlund's stage and check size is not in replacing WMS — it's in building above, around, and inside the WMS layer.
The automation software layer: hardware-adjacent and commoditizing
The warehouse automation wave of the past decade — goods-to-person systems, autonomous mobile robots, sorter technology — created a secondary software market for managing the hardware. Warehouse execution systems, robot fleet management software, conveyor control systems: these are products that sit between the WMS and the physical automation equipment, translating abstract order fulfillment requests from the WMS into specific task instructions for specific machines.
This layer has interesting software dynamics but is increasingly being captured by hardware vendors. Ocado Technology, AutoStore, and Geek+ have each built proprietary software that controls their hardware and communicates with third-party WMS through standardized interfaces. As the hardware vendors scale, they are less interested in selling open execution software that runs on competitors' equipment. The trend line for independent warehouse execution software vendors is toward acquisition by hardware companies or consolidation into the cloud WMS tier.
There are genuine exceptions — companies building hardware-agnostic execution software that can orchestrate mixed fleets across multiple robotics vendors are solving a real problem for large operators who have deployed multiple generations of automation technology. But the competitive dynamics require building both depth of integrations and speed of integration, which is engineering-intensive in a way that requires sustained capital to maintain. This is a Series B-and-beyond bet, not a seed bet.
The intelligence layer: where the seed opportunity actually is
Above the WMS and execution layer sits a set of analytical and decision-support capabilities that existing WMS products handle poorly or not at all. These are the categories where we see genuine seed-stage opportunity in warehouse software.
The first is labor planning and scheduling. Warehouse labor is typically 60 to 70% of variable operating cost at a distribution center. Current WMS products have basic labor management modules, but their scheduling algorithms are fundamentally reactive — they allocate available labor against anticipated volume without optimizing for individual worker performance characteristics, fatigue patterns, task-type proficiency, or the learning curves of new hires. ML-based labor intelligence products that learn worker performance models from historical pick, pack, and move data can produce material improvements in throughput per labor hour without adding headcount. Several companies are building in this direction, and the early traction signals in Europe are meaningful.
The second is slotting optimization. Slotting — the placement of SKUs in specific warehouse locations to minimize travel time and maximize pick efficiency — is one of the highest-leverage operational levers available to a warehouse operator, and most operators re-slot their facilities only a few times per year using heuristic rules. A product that continuously optimizes slotting based on real-time demand patterns, seasonal variation, and co-purchase data can reduce travel time per pick by 15 to 25% in complex pick-and-pack environments. The incumbents have slotting optimization modules, but they're static tools that require a trained analyst to configure and run. A product that makes continuous slotting a background process rather than a periodic project has a genuine wedge against the incumbent tools.
The third — and the one I spend the most time on — is exception management and exception prediction. Warehouse operations generate a constant stream of exceptions: delayed inbound shipments that need priority slotting, mis-picks that need rework, carrier capacity shortfalls that need contingency routing, labor absences that need same-day schedule revision. Current WMS products surface these exceptions reactively; they alert you when the exception has already occurred. A system that predicts high-probability exceptions 4 to 12 hours in advance — flagging inbound shipments at elevated risk of delay based on carrier performance history and current route weather, or predicting which shift-day combinations are at risk of labor shortage based on attendance pattern models — changes the operator's posture from reactive firefighting to proactive management. That posture change has real economic value in a cost structure as tight as warehouse operations.
The integration problem and what it means for investment
The practical barrier for warehouse intelligence products is the integration surface. Any software that reads WMS data needs to connect to a WMS that is almost certainly running a version that is years behind the current release and has a data schema that has been customized by 18 months of implementation consultants. The companies doing this successfully in our current pipeline have all solved the integration problem not through brute-force per-customer engineering but through a lightweight extraction layer that normalizes WMS event streams into a standard schema, independent of the underlying WMS version and customization history.
That extraction layer is itself a durable competitive asset. A company that has built extraction connectors for the ten most common enterprise WMS configurations in European distribution has an onboarding advantage that compounds with each additional installation. New customers can be onboarded in days rather than months. That speed-to-value is a direct selling point against incumbent analytics tools that require months of implementation work before producing a single insight. The integration surface, counterintuitively, is often where the defensibility is — not in the ML model, which can be replicated, but in the extraction and normalization work that makes the model trainable across a heterogeneous installed base.
Where we're looking in 2023
Given this map, our current sourcing focus in warehouse software is on companies building intelligence-layer products — labor, slotting, exception management — that are designed from day one to work above an existing WMS rather than replace it. The ideal profile is a founding team with WMS implementation or warehouse operations experience, a product that can demonstrate value in a proof-of-concept with a single WMS integration, and a go-to-market motion that enters through an operations or engineering buyer relationship rather than the procurement committee that controls WMS contracts.
We expect to make two investments in this category in 2023. The pipeline is active. If you are building in this direction, we'd like to talk.