A steel rolling line at work: a glowing billet travelling down the roller table, the crew at their stations

For factories in Egypt and the Gulf — metals, food, plastics, pharma, assembly

Predict every production risk. Act before it costs you.

Orders, machines, stock and cash on one live screen, in Arabic and English — so you know which order will be late, which machine stops next and where the margin goes, before the shift feels it.

Book a demoIs it for my plant?

The walkthrough is free · we reply within one working day

ManufactINS · Plant overview live
Plant overview

real product screens — sample data

Real footage of working plants · Mirian Managadze and ArtHouse Studio / Pexels · illustrative — not ManufactINS clients

Enterprise AI by Epsilon since 2015 · Egypt · Saudi Arabia · UAE · Iraq · USA

ISO/IEC 27001ISO 9001:2015ISO 21001:2018ITIDA-licensed
A ferro-alloys plant in Upper Egypt, a long conveyor rising into its furnace hall through the haze

01Is it for a plant like mine?

Built for plants in Egypt and the Gulf. From one line to a group of plants.

The sample plants it comes with are in 6th of October, Sadat City, Borg El Arab, 10th of Ramadan, Ain Sokhna, Beni Suef and Riyadh — and it runs on the rules, wages and language your plant runs on.

  • Arabic, all the way through

    Every screen, report and the Android app in Arabic and English, right to left. Ask it by voice in Egyptian, Saudi or Emirati Arabic.

  • Egypt’s rules, built in

    Labour Law 14/2025 overtime and rest-day pay, the official holidays, CAPMAS manufacturing wages, the Egyptian pound, and import lead times through Alexandria.

  • Works on the floor

    Operators record jobs on a phone or tablet. With no signal, entries wait on the device and go in order when it is back — each recorded once.

  • One line or a group of plants

    Start on one line or one plant; a group sees each plant on its own and all of them together.

See it on a sample plant in your sector, with your own questions.

One order, from the RFQ to the cash collected

Ferro-alloys plant near Edfu, Upper Egypt · Photo: Marc Ryckaert / Wikimedia Commons, CC BY 4.0 · cropped and graded, illustrative — not a ManufactINS client

02Does it know my process?

Pick your sector. See what your plant gets.

The process modules a plant of your kind gets — from weld maps to batch records — on top of the full RFQ-to-delivery suite.

Metals & fabrication

Discrete & assembly

Process & consumer

manufactins//os — process modules · metal fabrication & welding

sector

Metal fabrication & welding

The process modules this sector gets, added to the plant on top of the full RFQ-to-handover suite:

  • Weld Map & NDTEvery joint by WPS and welder, with VT/NDT status and the repair cycle.
  • Coating & PaintSurface prep, booth windows and DFT by zone — the defects driving rework.
See the full RFQ-to-handover suite

03What does it fix?

Six things that eat a factory’s margin. And what changes in each.

  • Paper and WhatsApp on the floor

    TodayJob cards on paper, progress in WhatsApp groups, the real status known a day late.

    With ManufactINSOperators record each step on a phone or tablet, supervisors approve quantities and hours, and the Document Reader turns the paper you still get into records.

    Live Operations
  • Orders that slip

    TodayYou learn an order is late when the customer calls.

    With ManufactINSOrder Feasibility checks materials, capacity and import lead times before you promise a date; each order has one workspace from RFQ to collection, scheduled on the machines shift by shift.

    Order Feasibility
  • Scrap and rework

    TodayDefects are found at final inspection, after the batch is made.

    With ManufactINSEach critical characteristic on its control chart, NCRs raised against the order, and every lot traced from the supplier to the customer.

    Quality
  • Machines that stop without warning

    TodayBreakdowns at 2 a.m., parts ordered after the failure.

    With ManufactINSRemaining life per machine, the repair booked in a planned window with the parts reserved, and every downtime report from the floor routed to maintenance.

    Predictive Maintenance
  • Margin you cannot see

    TodayYou know the plant’s margin at year end — not each order’s or product’s.

    With ManufactINSStandard against actual cost for every product and order; a product that loses money always raises a margin review.

    Profitability
  • Cash stuck in stock and unpaid invoices

    TodayStock piles up, invoices wait, and the bank line carries it.

    With ManufactINSWorking capital from what finance recorded — days of receivables and payables, and the collections to chase first — with cash planning on the same figures.

    Working Capital

Every figure explains how it is calculated, and every recommendation names its owner — nothing changes on the line without a person’s approval.

Which one costs you most? Show us in a demo.

04What does it cost me not to act?

Put in your plant’s figures. See what a week of waiting costs.

Three leaks every plant has — scrap, unplanned stops and cash held up — worked out from your own figures, with the formula under every number.

What this class of system is built to move

Typical targets for this class of system — not a client result and not a promise; yours is measured in the pilot.

EGP million
%
Currency
  • Scrap and rework

    %

    EGP 8Ma year

    sales × (1 − margin) × scrap rate — what it cost to make the output that failed first time

  • Unplanned stops

    %

    EGP 4Ma year

    sales × stop time × margin — the margin on output that was never made

  • Cash held in stock and unpaid invoices

    at cost

    days

    at the invoiced price

    days

    Central Bank of Egypt policy rates, 29 Oct 2026: deposit 20%, lending 21% — set your own

    %

    EGP 14.8Ma year

    (sales ÷ 365 × receivable days + cost of sales ÷ 365 × stock days) × cost of money — stock at what it cost to make, not its selling price; carrying EGP 74M held all year

At the example plant, every week of waiting costs about

EGP 515k

Change any figure to your plant’s

Stop the leak — book a live demoFree walkthrough · reply within one working day

Arithmetic on your figures only — not a forecast of savings and not a client result. The starting values are an example plant; put in yours. It is worked out in your browser and reaches us only if you send it with your demo request.

06How does it catch a loss?

Connect, reconcile, predict, act, verify.

One observable chain from your source systems to an accountable action — with a person in the loop wherever it is contractual, financial or safety-critical.

01

On the demo plant · CNC-04 · sample data

CNC-04’s vibration and spindle-load readings arrive alongside its maintenance history.

07Rather watch?

ManufactINS in five short films. The most important first.

Start with the whole picture: why plants lose output, quality and margin, in figures from Siemens, McKinsey, ASQ and Egypt’s own market. Then one problem a film.

Film 1 of 5The whole picture

5 films · most important first

What “The whole picture” says, with its sources
  1. Every plant runs on a plan. Every shift, something breaks it.
  2. $1.4 trillion is lost every year to unplanned downtime by the world’s 500 largest companies — 11% of their revenue. (Siemens, The True Cost of Downtime 2024)
  3. $2.3 million an hour — one hour of downtime in a large automotive plant; more than $600 every second. A large plant still loses 27 hours a month to stoppages. (Siemens, 2024)
  4. 60% is where most plants’ OEE really sits. World-class is 85% — the gap is capacity you have already paid for. (OEE.com, Vorne)
  5. 15–20% of sales revenue is the true cost of poor quality at many manufacturers: scrap, rework, returns and recalls. (ASQ, Cost of Quality)
  6. Every 3.7 years, a supply disruption that lasts a month or longer; over a decade they cost almost 45% of a year’s profit. (McKinsey Global Institute, 2020)
  7. In Egypt, the dollar went from EGP 15.77 in 2022 to about 49.5 in March 2024 — every imported machine, spare part and raw material repriced, and interest rates up 6 points in a single day. (Ahram Online; CNBC / LSEG; Central Bank of Egypt)
  8. 68% of the data a business has is never put to work. The data exists — it’s just everywhere. So problems surface after the shift: after the scrap, the stop and the spend. (Seagate & IDC, Rethink Data, 2020)
  9. ManufactINS — the intelligent operating system for manufacturing. Production, machines, quality, materials and decisions: one AI platform, in English and Arabic.
  10. See every plant, line and order on one screen. Catch the failure before the line stops. Know what to do first — and what it’s worth.
  11. Tell it what changed — in English or Egyptian Arabic. Track every part into the final product, with fewer tool changes. Trace every lot from the mill to the customer.
  12. Catch the defect at the first piece, not the hundredth. Reports your board will actually read.
  13. One platform. Every plant. Every decision. It connects the MES, historian, SPC and maintenance systems you already run. Book your demo — manufactins.epsilonaiglobal.com.

Product screens in the film show demonstration data.

08How fast does it go live — and what does it need from us?

Start with one line or one workshop. On the files you already keep.

How fast depends on the scope: the go-live plan, with its dates, is part of the written proposal after the walkthrough, and a pilot on one line or one workshop is the quickest start. You do not wait for an integration to see your own figures.

  1. 01

    A live walkthrough

    On a sample plant in your sector, with your questions. Bring your plant manager and your finance lead.

  2. 02

    Try it yourselves

    Individual, time-limited sign-ins. Upload your own exports to a sample plant and every figure follows — cleared when you sign out.

  3. 03

    A written proposal

    The scope you choose, its price and the go-live plan with its dates — in writing, once the scope is agreed.

  4. 04

    A pilot on one line or workshop

    The quickest start. We connect the feeds agreed for your scope, agree the measures first — downtime, OEE, scrap, on-time delivery — and measure them.

  5. 05

    Roll out

    More lines and plants — in the cloud, on your own servers, or both.

Proof of concept and pilot:A proof of concept proves one use case against agreed KPIs in about one month on average — the time depends on the use case and the KPIs chosen. A pilot runs the chosen scope on one line or one workshop, with success measures agreed up front; its length depends on that scope and is set in the written proposal. The pilot fee counts towards the licence if the plant goes ahead.

What it needs from your team

  • One person who owns the result — the plant manager or COO.
  • Someone who knows where the data lives — in planning or IT.
  • The exports you already keep — orders, bills of materials, stock, downtime. Excel is fine.
  • Two or three measures you want moved, agreed up front.

Where it runs — your choice

Fully managed in the cloud — the fastest way to get started, with updates handled for you.

Where the data lives
In the cloud, in the region agreed for your deployment
Who applies updates
Epsilon, continuously
Best when
You want the fastest start, with nothing to host

Your ERP and the systems you run

  • Excel / CSV exports — orders, stock, OEE, downtime, quality
  • ERP — Odoo, or any ERP through webhooks or its exports
  • MES — SAP DM, Siemens Opcenter, Tulip
  • Historian / SCADA — OSIsoft PI, Ignition, OPC-UA
  • CMMS — SAP PM, Maximo, Fiix
  • Email, WhatsApp Business, Google Drive, OneDrive & SharePoint

Examples of the systems ManufactINS connects to. The feeds for your plant are agreed on and connected during your pilot; the demo runs on sample data.

Ready to pilot it on one line or one workshop?

09Is my data safe?

Built to be trusted with your data

InsAI is delivered by Epsilon AI Global, which holds ISO 27001 (information security) and ISO 9001 (quality) management-system certifications, accredited through IAF / EGAC, alongside an ITIDA IT-activities license. The certified entity, scope and validity are stated on the certificates, which are available for verification; data residency and processing are configured per deployment and documented in a data-flow diagram provided for review.

  • Nothing contractual, financial, safety-critical or touching the product changes without a named person’s approval.

  • Every figure shows where it comes from and how it is calculated; sample data is always labelled as sample.

  • Each person signs in individually, sees what their role allows, and every change is in the activity log.

Where your data goes and who sees it: Data, security and AI

ManufactINS AI

AI built for the plant floor, not adapted to it.

ManufactINS runs on its own manufacturing intelligence: a proprietary plant model, its own decision engines, built-in manufacturing knowledge in English and Arabic, and ManufactINS’s own AI models, fine-tuned on manufacturing work and grounded in your plant data.

Our own plant model

Capacity, OEE, worked time, assemblies and CNC tool planning all come from one model of the plant. One bill of materials drives the ledger, the assemblies and the CNC plan, so every screen agrees.

Manufacturing knowledge, built in

Heat numbers, WPS, receiving notes, shifts and units, read the way plants write them: Egyptian handwriting and Arabic-Indic digits included.

Proprietary AI models, fine-tuned on manufacturing work

ManufactINS’s own models handle every task: reading delivery notes, certificates and work orders, handwritten shift sheets, the hard questions and the quick replies. Leading language models are brought in only where they improve the result, under the same checks.

Grounded and accountable

Every answer comes from your plant data and shows its source. When the data doesn’t hold the answer, it says so, and nothing changes until the person responsible confirms the recommendation.

How ManufactINS handles your data and AI

10Who else uses it — and what is the proof?

Early access — and a team with a decade of delivery.

ManufactINS is in early access: we are onboarding design-partner plants now, so the pilot is shaped around your plant and what you need shapes what is built next. We will not show you logos we have not earned — this is what you can check today.

  • Early access

    design-partner plants onboarding now

  • 17

    sample plants in the product, reconciled against Egyptian sources

  • Since 2015

    Epsilon AI Global, delivering applied AI

  • 100+

    enterprise and government clients across Epsilon’s work

The client count and the letters below describe Epsilon’s company-wide work — not this product’s own customers.

Be one of the first plants — start a pilot on yours.

Signed client recognition

Organizations that put their recognition in writing.

Signed recognition letters from clients across Epsilon’s projects. They reflect the team’s track record — not every organization is a client of this product.

More signed letters — and the projects behind them — are on the Epsilon showcase.

11Anything else?

Before you book a demo.

Fit, timing, data, integrations and cost — answered plainly.

Not here? Talk to us
01Is it for a plant like ours?

It is configured per sector — metals and fabrication, food and beverage, plastics and packaging, pharma, automotive components, electronics and appliances — and runs one line, one plant or a group of plants, in Arabic and English. It follows Egypt’s working rules: Labour Law 14/2025 overtime and rest-day pay, the official holidays, CAPMAS manufacturing wages and the Egyptian pound. The sample plants it comes with are in Egypt’s industrial cities and in Riyadh.

02Can we sign up and try it ourselves?

Not self-serve. You request a guided demo — the walkthrough is free — the Epsilon team walks you through the platform on sample data, then arranges individual, time-limited access to explore it — ideally on one of your own plants’ production data. Submitting the form never creates an account.

03Where does our data live?

Where your policy requires: fully managed cloud, inside your own network on-premises, or hybrid — sensitive data on-site, compute in the cloud. Residency and processing are set per deployment and documented in a data-flow diagram for your review.

04Which of our systems does it connect to?

The systems you already run — MES (SAP DM, Siemens Opcenter, Tulip), historian / SCADA (OSIsoft PI, Ignition, OPC-UA), maintenance (SAP PM, Maximo, Fiix) and spreadsheet exports. The feeds for your plant are agreed on and connected during your pilot; the demo runs on sample data, and we never present a sample integration as a live connection.

05Is ManufactINS running in plants today?

ManufactINS is in early access: we are onboarding design-partner plants now. The screens you see are the real product on sample data. Epsilon’s wider delivery record — 100+ clients since 2015 — is across all its work, not this product alone.

06How long until it runs on our plant?

It depends on the scope. The go-live plan, with its dates, is part of the written proposal after the walkthrough, and a pilot on one line or one workshop is the quickest start. You do not wait for an integration to see your own figures: with your trial access you can upload your own exports to a sample plant and every figure follows. Some capabilities need a history of your own orders before they say anything useful; your demo maps exactly what applies to you.

07What does it need from our team?

A person who owns the result — the plant manager or COO; someone who knows where the data lives, in planning or IT; the exports you already keep — orders, bills of materials, stock and downtime, and Excel is fine; and two or three measures you want moved, agreed before the pilot starts.

08Does the AI change things on the line by itself?

Only reversible steps, inside limits you set. Anything contractual, financial, safety-critical or touching the product waits for a named person to approve it — and every recommendation shows its evidence and owner.

09Does it work in Arabic — and with no signal on the floor?

Yes — the platform, its reports, the Android app and this site run fully in Arabic and English, right-to-left included, and you can ask it by voice in Egyptian, Saudi or Emirati Arabic. On the floor, a job step or stage entry made with no signal waits on the phone or tablet and is sent in order when the connection is back — recorded once.

10What do a proof of concept and a pilot look like?

A proof of concept proves one use case against agreed KPIs in about one month on average — the time depends on the use case and the KPIs chosen. A pilot runs the chosen scope on one line or one workshop, with success measures agreed up front; its length depends on that scope and is set in the written proposal. The pilot fee counts towards the licence if the plant goes ahead. We connect the agreed feeds, agree the measures up front (for example, downtime, OEE or scrap) and measure them during the pilot — the result is yours to judge, not a number we promise in advance.

11What does it cost?

The walkthrough is free. The platform is priced on the scope you choose — the plants and lines, the modules, the users, and cloud or on your own servers — and quoted in writing after the walkthrough, once the scope is agreed, in the same proposal as the go-live plan. The pilot fee counts towards the licence if the plant goes ahead.

12What happens when I ask?

Book a live demo — or a pilot on your plant.

ManufactINS is in early access — we are taking on design-partner pilot plants now.

  1. 1You send this formIt does not create an account — it reaches the Epsilon team directly.
  2. 2The team replies within one working dayBy phone, WhatsApp or email — and pick two or three times for the walkthrough as soon as you send.
  3. 3A free live walkthrough on a plant in your sectorYour questions — with your plant manager and finance lead, if you like.
  4. 4Trial access, then a written proposalIndividual, time-limited sign-ins — ideally on your own plant’s data — then, if it fits, a written proposal for a proof of concept (about one month on average) or a pilot, with the scope, the price and the go-live plan with its dates.

A live walkthrough on a sample plant in your sector, with your questions.

Name, work email and company are required. The rest are optional.

The walkthrough is free, and we reply within one working day.

We use your details only to contact you about this demo request — never sold or shared. See our privacy notice.

13Our offices

Talk to the team nearest you.

We pick your nearest office automatically and show its email, phone and WhatsApp — tap any pin on the map to choose another.

manufactins//os — global offices International line
USAEgyptKSAUAEIraqEgyptKSAUAEIraqUSA

your nearest office

Global

International line

Your location is used only to pick the nearest office — it is not stored or sent anywhere; your network’s country is looked up without keeping its address.

The InsAI ecosystem

Explore the rest of the platform

Every product runs on the same InsAI engine — move to another operation, or see everything at Epsilon AI Global.

Everything from Epsilon AI Global ↗