Demand, inventory and supply.
Connected by intelligence. Directed by you.
Assortment and on-shelf availability.
Cafés, restaurants and bakeries: production, freshness and daily orders.
Purchasing, lead times and working capital.
Production and deliveries aligned with demand.
Demand and replenishment across platforms.
Data, demand, forecasts, inventory and supply are connected. Second Intelligence helps explain the causes; people make the decisions.
From calculating requirements to an order your team can work with. Preparation, transfer and reconciliation are visible in one working process.
Prepare recommendations centrally, then open the order for an individual location.
Quantities, pack sizes and categories stay together. Expand, adjust and export the order.
The team transfers the order to its ERP and records when each step is complete.
From cleaning history and accounting for demand drivers to the right level of detail, manual adjustments and change history. Four screens show the different parts of this work.
Unusual spikes are highlighted in the history, so you can distinguish demand from distortions.
RPC AI compares sales with availability and reconstructs demand during stockouts.
Promotional periods are considered separately so a one-off uplift does not become regular demand.
Set a name and date range, then select all or specific products and locations.
The weather strip on the forecast chart helps compare demand changes with conditions in that period. How factors are applied depends on data and calculation settings.
Events stay linked to forecast periods, from individual days to seasonal changes.
Move between locations, categories and products. Switch between product → location and location → product views.
For infrequent demand, calculations can use an aggregated group and then allocate the result across the active assortment.
The model name is visible in the table. P50, P80 and P95 represent different forecast levels; the final choice remains under your control.
Adjust individual values or selected periods and compare them with the original forecast.
Find a product, category or location and expand the relevant level without losing the wider context.
The change panel shows when a forecast changed and which version is current.
English translation of the original RPC AI interface.
Machine learning finds patterns in demand history, seasonality, the calendar and weather. AutoML compares suitable models on past periods and selects a forecast based on error, stability and systematic over- or underestimation of demand.
Machine learning:global LightGBM through MLForecast, trained across groups of time series using lagged, rolling and external features.
Regular demand:AutoETS, AutoTheta, AutoARIMA and Seasonal Naive.
Intermittent demand:Croston Classic, ADIDA and TSB. Candidate models depend on demand type and configuration; an available model is not necessarily run for every item.
Models are tested on historical windows without access to future data. P50, P80 and P95 are forecast quantiles, not accuracy percentages.
Matrices define the rules, the plan shows requirements over time and analytics highlight where attention is needed.
Product status, coverage days, delivery days and delivery windows are visible in the relevant location cell.
Select products and locations, then apply the required settings to the whole selected area.
Product ranges can be aligned across matrices. The forecast uses each location’s active assortment.
Manage the production range by product and category.
Pack multiples, minimum batch sizes and shelf life are considered when preparing the plan.
Set an explicit allowance for stock remaining at the end of shelf life, instead of using one rule for every product.
The chart shows how stock changes between deliveries and against target and safety stock.
The plan accounts for delivery schedules, replenishment lead times, minimum orders and batch multiples.
Compare the recommendation with a manual adjustment and drill down to a location or product.
Review availability, days of stock, surplus and write-offs in one structure.
Dedicated sections cover forecast quality, cash flow and estimated implementation impact.
Period and object filters help you move from an overall indicator to the items that need work.
English translation of the original RPC AI interface.
One AI studies demand. Another helps explain what changed and where your decision is needed. Second Intelligence shows what needs attention, which data supports the conclusion and what to check before acting.
Each card describes a concrete situation, its business relevance and the next verification step.
Messages and history stay beside the finding. Bring AI into the discussion with @AI.
Preparing a decision or closing a card without action is a separate, deliberate choice.
The rationale, supporting data and verification method are collected in the same card.
Incomplete stock records, assortment differences and snapshot timing can change the meaning of a result. Those limitations should be visible.
Access to a card, investigation and actions are controlled separately. Preparing a draft does not change operational data.
You do not need to describe the whole table again. The scope of the answer is defined alongside the question.
“What matters now?”, “Explain this screen” and “Help with settings” let you begin with a specific task.
The assistant works within the user’s data access and permissions.
English translation of the original RPC AI interface.
An illustrative business case: how planning can release working capital and improve daily decisions.
inventory value
Target payback: within 2 months of deployment.
Illustrative estimates, not a measured customer result or a guaranteed outcome. The ranges reflect different business situations and are not a single financial calculation. Actual impact and payback depend on inventory mix, availability, margins and implementation.
Describe what you plan today and where the difficulty arises. Together, we will define the first working scenario, data and evaluation criteria.
Sales and stock, product and location master data, deliveries, lead times, promotions and constraints.
Your ERP records operations; RPC AI uses them for planning. Data can move through files or an integration. We agree the inputs and returned results for your system.
First, establish the current process and baseline metrics. Then compare the forecast and working plan with that baseline.
Standard data connection and integration included.
Every plan includes promotion effects and AI demand forecasting.
Six- and twelve-month subscriptions are paid upfront. All prices are in EUR.
From data and forecast validation to your team’s role.
You can start with a chosen category, channel or location. The first stage is defined by your task and data.
RPC AI uses data from operational systems for planning. Data exchange and returned results are agreed during setup.
We compare forecasts with actuals on a held-out period of history: future data is not used to build the forecast. Out-of-stock periods are considered separately. Metrics, horizon and the baseline planning method are agreed before evaluation.
Forecasting AI uses machine learning and automated model selection. Generative AI helps investigate causes and prepare a decision. The RPC AI core calculates inventory and supply within constraints; people confirm changes.
Sales history, stock, product and location master data, deliveries and lead times, promotions and constraints. Start by describing the process; we will define the required exports for your task.
When several supply-chain levels, warehouse transfers and the consequences of decisions across the whole network need to be modelled together. It is a choice based on process complexity, not a required upgrade from RPC AI.
In the demo, we will show the journey from data and forecasts to a recommendation and a human decision.
