Artificial Intelligence
AI-powered assistants that read, interpret and make suggestions based on the data your ERP calculates.
AI suggests, you decide
In ERP REAL, AI is an assistant, not a decision-maker. The numbers come from the ERP's own calculations; the AI reads them, explains them in plain language and makes suggestions. Whether to act on them is up to you.
Never changes data on its own
The AI only reads and suggests; it never creates, changes or deletes a record by itself.
Numbers come from the ERP
Quantities, dates and ratios are calculated by the ERP. The AI does not produce new numbers; it interprets the calculated result.
Supporting figures are visible
The figures and data behind each interpretation are shown next to it, so every statement can be checked against the raw data.
Guarded against made-up numbers
A summary containing a number that is not in the input is rejected; values outside a plausible range are not accepted.
How It Works
Example Scenario: Stock Running Low
Some items have fallen below their critical stock level. I suggest opening a purchase request and reviewing the minimum stock settings. If you agree, you start the action yourself.
- 1The ERP detects itStock quantities are compared with the critical and minimum stock levels defined for each item. This calculation belongs to the ERP.
- 2The advisor flags itThe Warehouse / Stock advisor highlights these items as intervention points in its report and suggests an action.
- 3You decideOpening the request, postponing it or rejecting it is your call. The AI never creates the order itself.
- 4It is followed upThe next report checks, together with the indicators, whether the suggested action has made progress.
Management Advisors
Role-based AI advisors look at your business through a manager's eyes. The data package is calculated by code; the advisor only reads it, interprets it and returns a structured report with an action plan. Each new report follows up on the previous report's action plan.
What Is in the Report?
- Intervention points: prioritised as Critical, Important or Monitor
- Findings: the figure behind each finding is written down as evidence
- Action plan: concrete action, responsible role, duration (days) and a measurable success criterion
- Data quality notes: missing or inconsistent records are findings too
- Follow-up of the previous action plan: Progressed, Partly, Not progressed, Not measurable
Action Tracking and Indicators
- Put a report into tracking; keep the owner, target date, status and progress notes for each action
- The Indicators tab shows the change between reports; the advisor evaluates both together
- If a user marked an action 'done' but the indicator has not moved, the report says so plainly
- Departments whose detailed setup is not finished can be flagged; the advisor makes no firm judgement there
Advisor Roles
Production Manager
ActiveLooks at production as a whole from the outside: evaluates capacity, efficiency, delays and bottlenecks.
- Output, working time and hourly efficiency per department
- Downtime and its causes (including how well it is tracked)
- Queues, waiting and delayed work orders; bottlenecks
- Terminal record quality: cancellations and corrections
- Staff efficiency (by employee code, process-focused)
- Orders that are late or at risk of missing their delivery date
Planning Lead
ActiveEvaluates the planning flow from order to work order and the capacity loading.
- Time to open work orders and planning-date discipline
- Department capacity load and sequencing
- Rush jobs and plan changes
- Ability to commit to delivery dates
Sales Manager
ActiveEvaluates offer-to-order conversion and sales performance by customer and product.
- Offer → order conversion rate and time
- Revenue and trend by customer and product group
- Price, discount and profitability signals
- Impact of delivery performance on customers
Warehouse / Stock Officer
ActiveEvaluates stock accuracy, turnover, unit consistency and critical stock.
- Stock accuracy, count differences
- Unit and conversion consistency
- Critical and idle stock, turnover rate
- Pallet and location usage
Shipping / Logistics
Awaiting dataEvaluates shipment plans, packing and on-time delivery. The advisor is ready; it is waiting for enough shipment data to build up.
- Plan → packing → waybill lead times
- On-time delivery rate
- Vehicle/pallet utilisation
- Partial shipments and their reasons
Quality Manager
PlannedEvaluates quality risks from quality control, scrap, return and correction records.
Finance / Costing
PlannedEvaluates product costs, profitability, receivables and cash flow.
Purchasing Manager
PlannedEvaluates supplier performance, purchasing lead times and raw material security.
Human Resources
PlannedEvaluates absenteeism, shifts, competencies and training needs.
Important Information
- Reports containing monetary values (Sales Manager) are only available to authorised users
- Staff are sent to the advisor by employee code only; suggestions target the process, not the person
- The data package used is visible under every report
- The report is written in the user's language (Turkish, English, German, Russian)
Bill of Materials Suggestion from Similar Products
When setting up a bill of materials for a new product, 'Suggest from Similar Products' prepares a proposal from the BOMs of similar products. The AI summarises this proposal in plain language for the production planner: what to do and what to check.
- The summary uses only the data in the proposal
- If the summary contains a number that is not in the input, it is rejected and not shown: no made-up measurements
- The summary does not write to the database; you approve and save the BOM
Extracting Fields from Product Names
Product card names often carry more reliable information than the fields themselves. The AI extracts fields such as tonnage, length, width, weight per metre, colour and safety factor from the product name and offers them as a suggestion.
Example
- Returns suggestions only; it never writes to the database itself
- The output is bound to a schema and additionally validated by rules: is the colour in the system list, is the number in a plausible range
- Uncertain fields are flagged separately
- If AI is unavailable, rule-based parsing continues
Daily Briefing for Instant Requirements
The calculated result of the Instant Requirements Analysis is turned into a short briefing for the planner every morning: the most urgent issue first, then the bottleneck, and a concrete suggestion where there is one.
- The model does not produce numbers: quantities, dates and day counts are calculated by the ERP, the model only interprets them
- Supporting figures are shown next to the briefing: number of items short, critical, at risk and not evaluable
- Prepared once a day at the time set in Company Settings; can be updated during the day with 'Refresh' if needed
- If AI is unreachable, the briefing is built from the figures using a template; the page always opens
AI Usage and Cost Tracking
AI usage is transparent: every call is logged, and cost and connection status are monitored.
Every call logged
Feature, model, input/output tokens, duration, success status and user are recorded.
Estimated cost
Cost is estimated from token counts and the model price; the monthly total is tracked.
Budget warning
A monthly budget can be set; you are warned when usage approaches the chosen percentage and when it is exceeded.
Connection status
The connection status of the AI service is shown on screen and checked at regular intervals.