MES on the plant floor, ERP in the office, no one agrees on OEE: Ask every line the same Arabic question, get the same governed answer
Bilytica's NL2SQL unifies MES, historian, SAP/Oracle ERP and quality data behind one semantic layer so plant managers, reliability engineers and quality leads ask OEE, yield, scrap, MTBF and Saudization questions in Arabic, with plant-scoped RLS and an OT-safe read layer that never touches control systems.
- NL2SQL accuracy
- 94.6%NL2SQL accuracy
- MES / historian / ERP connectors
- 12+MES / historian / ERP connectors
- unplanned downtime (median pilot)
- -18%unplanned downtime (median pilot)
- SAP HANA time-to-first-answer
- 3-4 wksSAP HANA time-to-first-answer
In Saudi Arabia
What a manufacturing reviewer checks before an NL2SQL bot goes live
Every question, Arabic or English, runs through the same governed semantic layer. This is the evidence the sector authority expects before the bot touches production numbers.
SASO, NDMO
Metric definitions used by the bot match the definitions الصناعة reporting already uses.
Semantic layer built from the existing governed metric catalogue, not invented per question.
Metric-to-source mapping document reviewed by the reporting owner.
SDAIA, PDPL
Row and column-level entitlements applied before an Arabic or English question can return a number.
Identity resolved and entitlements applied at the semantic layer, ahead of query generation.
Entitlement test log showing the same question answered differently per role.
NCA ECC
Read-only execution path, with SQL validated and cost-checked before it runs.
Query validator gate ahead of execution, no write path exposed to the bot.
SQL audit trail attached to every answer, reviewable line by line.
Where it runs
The warehouse connection and the semantic layer run inside Saudi Arabia; the bot never ships raw data outside that boundary to generate an answer.
Arabic, as it is spoken here
Arabic business questions, including mixed Arabic-English phrasing common in Saudi teams, resolve against the same catalogue without a translation step.
Free 30-day PoC on a sanitized slice of your MES + ERP
Recipe, BOM and formulation fields masked by default
Edge deployment for plants with limited connectivity
Saudization + NUSANED reporting out of the box
Questions your team actually asks, answered with governed SQL
In Arabic, English or a mix, the bot converts each question into SQL over a unified semantic layer.
OEE per line for the Jeddah plant this shift, ranked lowest first.
أعلى 10 أسباب توقف غير مخطط في مصنع الخرج خلال آخر 7 أيام
Compare scrap rate between shift A and shift C for SKU 44231, last month.
أي الموردين سجّلوا أعلى تأخيرات تسليم مع أثر على خط 3؟
Saudization ratio per plant vs plan, month over month, with headcount drivers.
أعطني MTBF للماكينات النقدية في مصنع الرياض آخر 90 يوم
Where teams use NL2SQL in nl2sql for saudi manufacturers
Plant manager Q&A on the floor
Plant managers ask OEE, MTBF, MTTR, yield and scrap by line, shift and SKU from a tablet on the shop floor, the bot joins MES + ERP behind the semantic layer and answers before the shift meeting ends.
Quality & supplier intelligence
Quality teams ask which supplier drove today's scrap spike or which SKU is trending toward spec drift, with cited SPC tables and a link to open the CAPA in SAP/Oracle QM.
Saudization & NUSANED reporting
HR and plant leads ask 'نسبة السعودة في مصنع الخرج مقابل خطة MOF لهذا الربع؟'، the bot pulls from SAP HR + Muqeem, reconciles with the plan, and generates the NUSANED-ready view.
Certified semantic layer + complete enterprise security layer
We never wire the bot straight to your warehouse, every question runs through a governed semantic layer and a security stack that respects every row and column.
Certified semantic layer
- Certified: OEE (availability × performance × quality), MTBF, MTTR, first-pass yield, scrap %, OTIF.
- Semantic layer over lakehouse (Databricks / Snowflake) with dbt models per plant + Saudization templates.
- Connectors: MES (Wonderware, Ignition), historians (PI, Aveva), ERP (SAP S/4HANA, Oracle EBS/Fusion), QM.
- Graceful failure: if a shift's data hasn't loaded, the bot says so, it never inflates an OEE from partial data.
Complete enterprise security layer
- OT/IT-safe read layer, the bot NEVER touches control systems, PLC or SCADA.
- Plant, line and shift RLS with delegated maintainer roles for external contractors.
- IP protection: recipe, BOM, formulation and yield-secret fields masked by default.
- Immutable audit for MOF Saudization, NUSANED, external auditor and internal QA review.
- Edge-capable, plant answers questions offline; audit trail syncs when connectivity returns.
Questions Saudi leaders ask before rollout
Our SAP S/4HANA + PI historian schema is a monster. Real integration timeline?
Integration effort differs by stack, SAP S/4HANA, Oracle EBS/Fusion, and combined SAP + PI + Wonderware each need their own validation path, including plant-side checks. We give a fixed-scope integration plan after the PoC, so procurement sees a real scope, not a guess.
Can it answer across multiple plants without exposing peer data?
Yes. Row-level security enforces plant, line and shift boundaries at the query engine. Regional managers see aggregated peer benchmarks; plant managers see only their own detail. External contractors see only the assets they maintain, nothing else.
What about connectivity, some of our plants are in remote industrial cities.
The semantic layer can run at the edge with periodic sync to the central lakehouse. The NL2SQL bot answers plant-scoped questions locally, and syncs audit trails when connectivity returns. Tested at plants in Jubail and Ras al-Khair.
Can it produce Saudization / NUSANED reports directly?
Yes. Saudization ratio, nationality mix, wage-protection status and NUSANED KPIs are pre-built certified metrics. Ask in Arabic, get the export-ready sheet, with an audit log HR can hand to MOF and MHRSD.
See your first OEE / downtime / Saudization question answered on YOUR sanitized data, 20-minute working session, no PoC fees.
No fees, no procurement commitment, no data leaving Saudi Arabia. You walk out of the session with an accuracy evaluation report your team owns, whether you buy or not.