Depot-1

Logistics operations generate large amounts of information, but understanding what is actually driving performance is harder.

Our research explores how Depot-1 can reason across scorecards, workforce data, defects, disputes, fleet information and other operational signals to build a clearer picture of what is happening and why.

Key areas include:

• operational state understanding
• root-cause analysis
• prioritisation
• performance trends
• constraint awareness
• decision support

Not every issue deserves equal attention.

Our research with Depot-1 explores how a model can distinguish between symptoms and underlying causes, determine which problems matter most and identify the interventions most likely to improve an operation.

The aim is to move from “What happened?” to “Why did it happen?” and ultimately to “What should happen next?”

Operational decisions create measurable outcomes.

Depot-1 research explores how those outcomes can be used to improve future reasoning by connecting operational state, decision, intervention and outcome.

Over time, this creates the potential for models to develop a deeper understanding of which actions work, in which environments and under which conditions.

Modern logistics operations rarely exist inside one system.

Our research explores how Depot-1 can work across different data sources and operational tools while maintaining context across the wider operation.

This includes areas such as:

• scorecards and performance data
• workforce information
• vehicle and fleet data
• defects and disputes
• route and volume information
• financial and cost data
• internal operating processes

Depot-1 is being researched first as an intelligence system that supports operational decisions rather than replacing the people responsible for them.

The focus is on helping operators understand what deserves attention, what evidence supports a conclusion and where human judgement is still required.

As confidence and capability increase, this research can inform increasingly automated forms of operational decision-making.

A specialised model should be judged on the decisions that matter in the environment it is designed for.

Our evaluation work with Depot-1 focuses on realistic logistics scenarios rather than general-purpose benchmarks alone.

We are interested in measuring areas including:

• diagnosis accuracy
• root-cause reasoning
• intervention selection
• prioritisation
• confidence and uncertainty
• consistency across operations
• tool use
• performance against expert judgement

As the research develops, we expect these evaluations to become an important part of measuring progress between model generations.

Operational intelligence needs to be reliable when decisions affect real people and physical operations.

Our work with Depot-1 includes research into uncertainty, traceability, human oversight and the ability to recognise when the available evidence is insufficient.

Depot-1 is proprietary to Wehand and is not sold as a standalone product.

We deploy it in our own work with operators. Depot-1 reads the operation and identifies the priorities; Operator-1, our agentic model, carries out the work to deliver them.

Together they help operators diagnose performance, reach their targets and keep improving as the results come in.