Flexible or fixed automation - the architecture decision that comes first
This page frames the two classes and routes you to the right architecture. If you already know you are evaluating a fixed system, the direct comparisons against the NEO retrofit sit further down.
Fixed automation - what counts as fixed
Fixed automation ties goods to built infrastructure. A few architectures dominate the market, and they share one logic: first the system is built, then goods flow through it.
An AS/RS crane system runs a rail-guided crane through narrow aisles between tall racking and hands bins to conveyors. Shuttle systems use multiple vehicles per level, connected by vertical lifts, with more throughput than a crane system and more moving parts. Cube-storage stacks bins without aisles in a grid; robots drive across the top and lift bins from above. Conveyor links these systems to pack and dispatch zones, but is the connective element rather than a standalone goods-to-person concept.
Different as they are in detail, these forms share two traits: each needs a built installation, and each fixes its capacity before the first pick.
The most underrated point is capacity planning. A fixed system's throughput is set before it is built. Size it for the average, and you break during peak. Size it for peak, and you pay eleven months a year for idle infrastructure. This either-or cannot be resolved inside a fixed architecture.
AMR / flexible automation in the existing warehouse
AMR are the only architecture class that works without fixed infrastructure. That changes the entire deployment logic, not only the technology.
Autonomous mobile robots navigate the warehouse on their own, without rails and without installed guidance. Within the class, several approaches differ sharply in how much they automate (the AMR glossary entry covers the variants). The NEO approach is bin-to-person from existing shelving: a vertically reaching robot pulls individual bins straight from the shelf compartment and brings them to a stationary goods-to-person station, where the picker works without walking. The existing racking, the grown storage order, and the aisles stay untouched.
What the retrofit makes possible
AMR integrate into the existing racking without heavy construction. No racking teardown, no floor rebuild, no structural prerequisite beyond adequate aisle width - NEO operates from aisles as narrow as 85 cm and shelf heights up to 250 cm. Deployment is measured in weeks, not months: NEO goes live in 6-8 weeks from contract to first automated pick, because no structural work is needed.
Because no plant is built, the mandatory upfront investment disappears. AMR are the only class broadly available as an operating-cost model - at NEO as pay-per-pick, a fee per order actually picked, with no capex. That also shortens the approval path, since an operating expense skips the multi-year investment procedure.
Two properties no fixed architecture can match to the same degree. First, mixed operation from day one: part of the warehouse runs automated while the rest stays manual, and the boundary shifts gradually without stopping operations. Second, scaling during live operation: robots added within days, no construction and no lead time, and just as easily removed when demand falls.
Where AMR hit their limits
AMR need aisles to navigate. Their storage density therefore sits below cube-storage, which needs no aisles, and below shuttle systems in very tall halls. If lack of space is the real problem, that is a genuine drawback - no messaging removes the aisles.
There is a ceiling at the top, too. From around 10,000 order lines per hour, mobile systems hit physical limits, because only a limited number of robots can work the same aisle at once. For steadily very high volumes in that range, shuttle systems or hybrid architectures are the better fit.
The hybrid path - AMR as the entry before fixed investment
Fixed and flexible are not mutually exclusive. In practice, companies use AMR as a fast entry in the existing building and invest in fixed automation later, once volume and location certainty justify it.
The advantage of that order lies in the data. An AMR retrofit ties up no capital and builds operational experience within weeks. The pay-per-pick data from live operation give a solid basis for a later investment decision - real picks instead of projections. Starting with AMR lowers the risk: no purpose-built hall, no multi-million budget, and automated picking all the same.
For the seasonal peak, NEO resolves the capacity dilemma of fixed systems through collaborative picking: robots and human pickers work the same picking cycle. The fleet is sized for the base load and topped up with additional pickers during peak season, working right alongside the robots - no idle capacity out of season, no fixed peak sizing.
NEO in your existing shelving
NEO brings true goods-to-person into your existing shelf-based warehouse - no new building, no upfront investment. The robot fleet pulls bins from the shelf, the goods-to-person station presents them to the picker without walking, and NEO:os connects the whole to your existing WMS through standard interfaces. The platform overview shows how software, fleet, and station work together; the warehouse retrofit page covers the physical setup.