INDUSTRIAL OPERATIONAL INTELLIGENCE

Find out what your plant is losing.
Fix it, and prove you did.

A crore-rupee machine goes down; nobody notices for an hour. A quality problem starts today, surfaces next week. By then the money is gone.

EdgeBits reads the meters and PLCs you already own, ranks what each loss is costing you, and records the fix. Your data stays on your server, and publishes to SAP, Oracle, or Tally.

Start with one line See how it works

Year one, typically: 25–35% less unplanned downtime, 5–10% off the energy bill. Industry figures, sourced. Put your own numbers in.

7+ Yrs IIoT engineering
by founding team
₹5,000 Hardware to start
(a Raspberry Pi 4, about $60)
DMZ-native 0 new firewall rules
Brownfield Ready, no rip-and-replace

START WHERE YOU SIT

I own the business “What does this save me, and do I own it?” Model your numbers I run the plant “What’s breaking, where, and what do I do about it?” Solutions by first pain I run maintenance “Does it fit the TPM program I already run?” The Six Big Losses I run IT / OT “Does this open a port, and who holds the data?” The architecture
CONNECT · ANALYSE · ACT

Three steps, and you own every one of them.

Most plants are stuck at step one. The money is in steps two and three.

Connect

Reads the meters, PLCs, CNCs and sensors already bolted to your floor, over the protocols they already speak. Nothing gets replaced, no rewiring, no production stop.

Proof: seven protocols in the catalog: Modbus, OPC-UA, Siemens S7, EtherNet/IP, MQTT, REST and IEC 60870-5-104. Runs on a ₹5,000 box. Need one that isn’t listed? It ships as an add-on block, not a platform rewrite.

Analyse

Not another dashboard to read. You get a ranked list of where money is leaking, each line with the reason attached: which machine, which shift, what changed just before it started.

Proof: every figure is computed by the platform, never written by a language model. You can trace any number back to the readings behind it.

Act

Alerts to the people who can act, setpoints back to the machine, production actuals into your ERP. Every change is stamped with who did it, when, and why, and can be reversed.

Proof: a tamper-evident record you can export whole and hand to an auditor, a customer, or a regulator.

Running a TPM program? See how this maps to the Six Big Losses. You can also ask your plant questions in plain English.

START HERE

The First Line.

One production line. Thirty days. Your meters, your server, your numbers — and at the end, a figure you can check against your own books rather than a slide we made.

Week one — connected

We read the machines already on that line, over the protocols they already speak. Nothing is rewired, nothing is replaced, and production does not stop for us.

Weeks two and three — the picture

Live view of that line, the losses ranked by what they cost, with the reason attached to each one. This is usually the week somebody says “I did not know it was doing that.”

Week four — the number

What the line actually cost you over the month and what is recoverable, next to the estimate you started with. If the two do not agree, we would rather you find out on one line than on ten.

Fixed-fee, and it comes off the licence if you go ahead. You keep the data either way, because it never leaves your server. Tell us which line you would put it on, or estimate the number yourself first.

EDGE TO LAST MILE

Normally this is three purchases.

Getting from a machine to a decision usually means buying an edge data layer from one vendor, a historian from another, and an intelligence tool from a third, then paying somebody to make the three of them talk. Two of those vendors will keep your data on their cloud.

The edge layerusually a separate product

Reads your machines where they stand, processes at source, and survives the network you actually have. Outbound-only, so plant IT never opens a port.

The historianusually a licence per tag

Three years of history on your own box: raw for 90 days, then rolled up. No per-tag tax, so adding a line is not a licence negotiation.

The intelligenceusually a cloud contract

Losses ranked by what they cost, the reason attached, the fix recorded. The part the other two leave to you, and the part that normally means shipping your plant data into somebody else’s tenancy.

One platform, edge to last mile, on infrastructure you choose: the plant floor, your private cloud, or hosted by us. Already run a historian? Keep it; we publish into AVEVA PI and reach Ignition over Sparkplug B. How it deploys.

OPERATIONAL INTELLIGENCE

The plant glows where the money leaks.

Not another dashboard to interpret. A map where each line is sized by what it costs you and coloured by how bad it is, then the reason, the fix, and proof it worked.

01

See it

Every line drawn in proportion to the money leaking through it. Red where it hurts, green where it does not. One look tells you where to spend Monday: power, water, compressed air, scrap and throughput — on one map instead of five dashboards.

02

Understand it

Each loss carries what you actually used, what you should have used, and the gap between them in your own currency. Plus the reason, the evidence behind it, and correlations the platform found on its own. Every figure is computed, never written by a language model.

03

Fix it, and prove it

A recommended next step written for that loss on that line, not a boilerplate tip. Open it as a job, and the platform captures what it was costing before, follows it while the floor works on it, and measures what you got back. Next month you can show the fix held instead of arguing about it.

Underneath, that map is a contextualised knowledge graph of your operation: site, line, device, and what each one is costing. It is built from your own readings rather than a modelling project, and it deepens the longer it runs. There is a full-screen version for the floor too — the OEE kiosk runs on a wall-mounted TV, needs no login, and an operator can retune what it shows without calling anyone. Part of the Analytics add-on. Then it keeps score, by day, week, month, quarter and year, normalised per unit produced so a busy month cannot flatter you. When there is no tariff for a meter the loss is shown in units and the money is marked unknown, because an honest gap beats a confident zero. Ask it questions in plain English.

FROM NOISE TO WHAT MATTERS

Millions of readings. A handful of things worth your morning.

A plant generates facts by the million. Most alarm systems turn those into thousands of notifications a day, which is how operators learn to ignore them. The job is not more alerts. It is fewer, each one real.

Readingsmillions an hour

Every value from every machine, kept and queryable. Facts, not judgements. Nobody is meant to read these.

Alertshundreds a day

A condition you set was met, with severity and context attached. Useful, but on their own this is where alarm fatigue comes from.

Incidentstens a week

Related alerts correlated into one episode that has a state and an owner: open, being worked, resolved. Six alarms from one failing pump are one incident, not six interruptions. This is the list a shift lead can actually work through.

Aligned with ISA‑18.2, the alarm-management standard your process engineers already work to — so the shape is one they recognise rather than one we invented.

AI IN MANUFACTURING

“Why was line 3 worse last week?”

Ask in the words you would use in a shift meeting, and get the answer with its working attached — the readings, the window, what changed. The hard part was never the asking. It is defending the answer in the room afterwards.

The numbers are computed, not generated

Every figure comes from the platform doing the arithmetic on your readings. The model narrates; it never invents the number. One report you can sign. The other you cannot.

It can act, not just answer

Ask it to open a job on the worst loss and it does — under the same permissions as a person, on the same audit trail. An assistant that can only talk is a search box with better manners.

The model is yours to choose

Hosted for speed, or running on your own hardware so nothing leaves the site. Swap it later and nothing else changes — no rebuild, and no lock-in to one AI vendor.

Everyone asks in their own words and sees only their own scope — an owner across every plant, a manager on theirs, an operator on their line. See how asking works.

Reads from the equipment you already own

Modbus TCP/RTU OPC-UA EtherNet/IP (Allen-Bradley) Siemens S7 MQTT / REST / HTTP IEC 60870-5-104

Sends it wherever you need it

Sparkplug B (Ignition) SAP RFC & OData AVEVA PI REST / HTTP

Built on published standards your IT team already knows: ISA-95, MQTT, OPC-UA. Nothing here is proprietary. Audit + correlation via fasten (GitHub), our open-source audit trail. Every state change is auditable and immutable: who did what, when, where, why, and how. New blocks ship on demand, not as a platform rewrite. No proprietary protocols, no vendor lock-in.

Intermittent network? Not your problem.

Data queues locally at the edge, encrypted. Syncs automatically on recovery. Nothing lost, no manual intervention, no data gaps in your reports.

No raw sensor data to the cloud.

Processing happens at the source. Only refined, approved data leaves your network. Your sensors never talk to the cloud directly. You control what moves and when.

Intelligence at every layer, edge first.

Rules, aggregations, FFT, and alerts all run on the edge node. Cloud gets answers, not noise. Less bandwidth, faster response, lower cost.

THE PART THAT COMPOUNDS

It gets better the longer you run it.

Most plant software is worth the same on day 400 as it was on day one. This is built the other way.

01

The plant remembers

Each machine carries a live model of itself — a digital twin: its state, its history, how much life is left in the part that is wearing. Each line carries the same for a whole order: what it produced, at what quality, for what energy. It does not only map your machines and what they cost. It accumulates. What went wrong on this line before, what fixed it, what that fix was worth, what changed since. A plant you have run for a year is easier to run than one switched on last week, because the system knows the difference.

02

Your machine data meets your orders

Reconcile against SAP or the ERP you already run — production orders, cost centres, quality specs. The money stops being per line and becomes cost per order, per SKU, per customer. That is the number your finance director will argue with you about.

03

It sees the failure coming

Models run on the box itself rather than a cloud round trip — bearing wear, motor-current drift, the slow slide before a stoppage. Predictive maintenance on hardware you already own, with nothing leaving the plant to get it.

Historian integrations for FactoryTalk and Wonderware, native SAP RFC/BAPI, and alert-to-incident correlation round it out. Tell us what your stack looks like and we will show you the parts that matter to you.

7+ years in industrial IoT

The engineers building EdgeBits have spent 7+ years shipping industrial IoT software: contributing to open-source projects, building OT protocol integrations, and deploying systems in real factory environments. Deep familiarity with what actually breaks on the shop floor.

IT ↔ OT, by protocol

Bridges Modbus, OPC-UA, EtherNet/IP, S7, Sparkplug B on the OT side to SAP RFC/OData, REST, MQTT, and warehouses on the IT side. Works with ABB, Schneider Electric, Siemens, and Allen-Bradley equipment out of the box. No custom integration projects.

Domain-agnostic by construction

The same engine costs power, water, compressed air, scrap and throughput. Energy is where most plants start, not a separate product — so the second thing you meter needs no new licence and no new tool.

One platform, three parts

One platform, packaged as three independently deployable parts: the runtime on the floor, the layer that governs the fleet (the control plane), and the layer the answers are computed from (the data plane).

Each can be deployed and sold independently. Start with Edge Node, add Manager and Analytics as you scale.

connect

Edge Node

the edge runtime — any machine, on-prem, no VPN

Factory Floor · Per-Node License
manage

Edge Manager

fleet, topology and deployment across sites — the control plane

Run the fleet Cloud or On-Prem · SaaS
analyse

Edge Analytics

OEE, energy, history and the answers on top — the data plane

Reports & history Cloud or On-Prem · Add-On
Edge Nodes ── HTTPS (mgmt) ──→ Edge Manager + Edge Nodes ── pushes readings ──→ Analytics

Which one you start with depends on what hurts — see them against a real line, or read how they deploy on your network.

CENTRALISED MANAGEMENT

Ten plants on one screen.
Nothing new for IT to approve.

Deploy a change to every site from one place, and see which plant needs attention without logging into any of them. Each plant calls out to reach the fleet, so adding the tenth site is the same work as the first.

FACTORY · ON-PREMISES FLEET MANAGEMENT Gurugram Edge Node Faridabad Edge Node Noida Edge Node Edge Manager SaaS or self-hosted outbound HTTPS · no inbound port · DMZ-native
  • No VPN concentrators
  • No inbound firewall rules
  • No IT escalation per site
1 Dashboard
Multi-site Fleet topology
0 VPN licenses
See Edge Manager DMZ & security

STOP RENTING YOUR OWN DATA

Take back your data.

Meter vendors, historian vendors, MES-as-a-service — they all end up in the same place. Your data on their server, their dashboard, their schema, and a bill every month for the privilege of looking at what your own machines produced.

THE VENDOR PATTERN — TODAY

Your data lives on someone else’s server

  • Meter or PLC phones home to the vendor cloud
  • You see their dashboard, their schema, their features
  • Monthly fee — per tag, per user, per site
  • Your operations team can only look, not touch
  • Export is whatever they choose to support
  • Stop paying → your data goes with them
THE EDGEBITS PATTERN

Your data on infrastructure you choose

  • Meter → Edge Node → your TimescaleDB on your box
  • Your dashboards, your schema, your extensions
  • Fixed price. No per-tag tax. No per-user tax.
  • Full SQL access, REST API, and export
  • Publish to Power BI, Grafana, SAP, or your own tool
  • Stop paying → your data stays with you

Modbus TCP/RTU · Siemens S7 · OPC-UA · MQTT · REST connections, on-node storage on TimescaleDB, REST egress — all shipping today, packaged for Debian / RPM. Bring your meter, we’ll bring the box.

Talk about your meter vendor See what else it does

Start with whatever is hurting most.

Four common shapes below — the ones we hear on the first call. ERP integration, predictive maintenance, and brownfield connectivity each get their own landing on the Solutions page.

Energy Monitoring — on your server

Own your energy data. Read from any Modbus meter you already have, keep the readings on your infrastructure.

Meters → Your Server → Reports

Manufacturing Intelligence

OEE, cycle analysis, downtime classification, read directly from PLCs and CNC machines. Live per line.

PLC → Edge → Dashboard

Quality & Compliance

Continuous monitoring for GMP environments. Immutable audit trails for pharma, food, and auto.

Sensors → Validate → Audit

Multi-site fleet

One plant to ten without VPN. Edge Manager runs the fleet; each plant calls out over HTTPS, so nothing calls in.

N Plants → Manager → One Dashboard
See all seven solutions

ERP integration, predictive maintenance, and brownfield connectivity each have their own landing.

Everything the plant needs, running on the plant.

Shop floor to ERP. One platform, any deployment model.

Industry-standard connectors

If your equipment has a network port, there is almost certainly a connector for it — and a new one ships as an add-on, not a platform upgrade. See the full list.

Visual Pipeline Editor

Drag and drop to build a flow: read a machine, shape the reading, send it on. No code. Works on any device with a browser.

Deploy Anywhere

Behind corporate DMZ, on air-gapped intranet, or in Docker/K8s. Adapts to your IT policy, not the other way around.

Local buffer + backfill

Edge node buffers locally through network drops and power cuts. When the link returns, missing windows backfill into their correct slot. The daily roll-up has no holes.

Edge Manager — centralised management for the whole fleet

Remote-deploy pipelines, push config and watch health across every site, with no VPN to any of them. SaaS or self-hosted.

Edge Analytics — the reports your managers actually open

OEE per line, tag explorer, data quality monitor. Correlate with your ERP (SAP, Oracle, Tally, ERPNext): production orders, cost centers, quality specs. No must-have for inbound access to the plant.

Audit + lineage (fasten)

Every state change is sealed by fasten, our open-source audit trail: who did what, when, where, why, and how. Tamper-evident hash chain, exportable for any audit, immutable by design.

SAP & ERP Integration

Push production actuals to SAP (MES/ERP) via RFC/OData, Oracle/Dynamics via REST, or legacy ERPs via batch reports. Close the shop-floor-to-MES/ERP gap in real time.

Line 3 in Pune means the same thing as Line 3 in Chennai.

Every machine on every site is named in one consistent structure: plant, area, line, machine, reading. It sounds like bookkeeping. It is the difference between comparing two plants on Monday morning and commissioning a six-week mapping project first.

It also means a new tool just reads. Point Power BI at it, or a historian, or an AI assistant, and none of them need anything rewired on the floor or changed in the PLC. The structure is ISA-95, published and vendor-neutral, carried on a Unified Namespace. There is nothing proprietary to learn, and nothing that locks your data to us.

A library that keeps growing, not a rewrite every time.

Everything the platform can read, shape, decide and publish is a building block in one versioned library. Need a protocol or a destination that isn’t there yet? It arrives as a new block on a running system — nobody re-platforms, nobody re-tests the plant. That is how it is built, not a promise for a future release.

7 Connectors
machines it reads
6 Destinations
where it publishes
20 Presets
ready-made processing
6 Rule types
conditions to watch for
9 Actions
what happens when one fires

New blocks publish continuously, and what gets built next follows what customers actually ask for. If a specific connector or destination is on your critical path, tell us and we will sequence it.

Plant on a screen. Machine on a pin.

Anyone who has run Wonderware, FactoryTalk View or Ignition Perspective expects a screen where each machine is a shape whose colour tells you whether to walk over. You get the same thing here, driven by live readings instead of a separate HMI project. Plant managers read pictures, not tables. The spatial view lays every machine on your shop-floor plan with a live health pin (green / amber / red), driven by the same live readings everything else on the platform reads from. Hover for the why, click for live values + history.

EdgeBits floorplan view of an alloy wheel line — CNC-03 amber for spindle drift (78 wheels vs ~100 on the others), 380 wheels packed this shift, 1240 kWh on the energy meter, humidity sensor SH-01 offline EdgeBits floorplan view of an alloy wheel line — CNC-03 amber for spindle drift (78 wheels vs ~100 on the others), 380 wheels packed this shift, 1240 kWh on the energy meter, humidity sensor SH-01 offline
Green = running fine. Amber = drifting (CNC-03 is producing 78 wheels this shift vs ~100 on the others — a 22% gap, caught at the next shift change instead of four hours of lost output). Red = broken. Grey / striped = planned stop or sensor offline. No more walking the floor.

Three views ship together: Floorplan (where on the line is the red light), Data Wall (a kiosk-friendly tile gallery of every machine), Device Detail (make, model, photo, location, last 24 h). All driven by the topology you already have. Tell us which line you'd put it on first.

FOR THE TEAM THAT HAS TO APPROVE IT

Built for the review your IT team will run, not around it.

Nothing calls into your plant. No inbound port, no VPN concentrator, no agent phoning home on its own schedule — and air-gapped is a supported configuration rather than a workaround.

Mutual TLS, data residency, roles and audit, the protocol list, how it installs: all of it is answered on the page written for them. Send us your security questionnaire and we will fill it in.

From a meter on Friday to a dashboard on Monday.

Connect a meter on Friday, show a live dashboard on Monday. The first one takes an afternoon; the rest is you deciding what else to point it at.

1

Install & Onboard

Deploy on bare-metal Linux, in Docker, or on Kubernetes. Enter a one-time claim code, and the device registers itself via outbound HTTPS. No firewall changes, no IT escalation.

Your IT installs it once. You manage everything from a browser.

2

Build Pipelines Visually

Open the dashboard and drag the steps onto a canvas: read, shape, send. Configure each node with auto-generated forms — validation, hints, and dependency checks built in.

Modbus → Aggregate → Rules & alerts → SAP RFC
3

Monitor via Edge Manager (Cloud or On-Prem)

Each site checks in with Edge Manager continuously. Push deployments, pull health and readings, manage settings — all through Edge Manager. A secure on-demand session for live troubleshooting when you need it.

Edge Node ── outbound HTTPS ──→ Edge Manager
EdgeBits platform flow — sensor to ERP, historian, and BI EdgeBits platform flow — sensor to ERP, historian, and BI

From the blog

Data ownership · Energy Jul 2026

Take your energy data back from the meter vendor

Your meter has a Modbus port. You don't need the vendor's cloud. Honest math on what it costs to own your own numbers.

Read Article
CCTS · Steel Jul 2026

CCTS: Building the emissions trail at the source

A deadline, not a dashboard. What we shipped for CCTS: Modbus meter to sealed tCO₂e/t on the box, three deploy variants, audit-ready lineage.

Read Article
TPM · ROI Jul 2026

TPM ROI: what OEE percentage points are actually worth

Vendors quote 90% savings. The real year-1 number is 1.4% of revenue. Honest math with the citations open.

Read Article
Analytics · Assistant Jul 2026

Ask your plant in English

Owner, plant manager, operator: each asks the same assistant in the words they already use, and gets scoped answers with the audit trail attached.

Read Article
Architecture Jun 2026

The Unified Namespace, Edge to Last Mile

ISA-95 as a runtime, OPC UA as a feeder, and one MQTT bus on the edge that feeds MES, ERP, historian, BI, and AI. No point-to-point spaghetti.

Read Article
What & Why Mar 2026

Why Your Factory Needs an Edge Layer, and Why Now

Indian manufacturing is at an inflection point. Factories have PLCs and SCADA, but the data stays trapped on the shop floor.

Read Article
Workflow Mar 2026

From PLC Register to Live Dashboard: A Complete Workflow

Tag discovery, on-site processing, publishing to the cloud, and remote monitoring. Four steps from PLC register to live dashboard.

Read Article
Benefits Mar 2026

5 Ways Edge Computing Cuts Factory Costs

90% less cloud bandwidth, zero VPN licenses, offline resilience, and faster SAP integration. The ROI case for edge.

Read Article
Industry 4.0 → 5.0 Mar 2026

Industry 4.0 to 5.0: What Changes and What Stays

4.0 connected the machine. 5.0 connects the human. Here's how your edge infrastructure evolves without ripping and replacing.

Read Article
IT-OT Bridge Mar 2026

IT-to-OT Convergence: Bridging the Gap Without Breaking Production

SAP wants real-time data. The plant manager won't open firewall ports. Here's the architecture that satisfies both.

Read Article
Playbook Apr 2026

Brownfield to Greenfield: A 4-Phase Digitalization Playbook

Your factory has legacy PLCs and manual data entry. Here's how to turn it into a connected operation without ripping anything out.

Read Article

No logos to show you yet. Here is what we can prove.

References come out under NDA in a demo, not on a web page. The people building it are not anonymous though — EdgeBits is made by nerdAppLabs, a software company since 2012, and the engineers who build it are named on their team page.

Open-source community background

The founding team has 7+ years building industrial-edge platforms, deployed across factory floors processing sensor data from production equipment worldwide. That work built deep familiarity with OT protocols, brownfield environments, and what ERP systems actually need from the shop floor.

Open-source industrial IoT

Open standards, no lock-in

MQTT for the event bus. ISA-95 for the topic hierarchy. OPC-UA, Modbus, S7, Sparkplug B for the protocols. Standards plant IT teams already understand. No proprietary protocol. No vendor-only file format. No closed catalog.

ISA-95 · UNS MQTT · OPC-UA · Sparkplug B

In conversation across the floor

We are in active conversation with auto parts, pharma, steel, textile, and FMCG manufacturers in North India. If your plant runs Siemens, ABB, Schneider Electric, or Allen-Bradley equipment and your data still lives in spreadsheets, that's exactly who we're talking to. References available on request during a demo.

Talk to us →
“WE ALREADY HAVE SCADA”

Good. Keep it.

Not a SCADA replacement, and not a historian replacement. It reads alongside what you already run and takes the data where SCADA was never meant to go — your ERP, your reports, your auditor.

01

Your operators never notice it

It reads quietly in the background, on the same cables and the same screens your shift supervisor has used for ten years. Production counts reach your ERP on their own. Nothing on the floor changes, so there is no retraining and no window where output is at risk.

02

One line, owned outright

Replacing a SCADA platform with a certified integrator runs ₹15–20 lakh in India before a single dashboard is live. This starts on one line, and you own the licence rather than renting per tag. Connect a meter on Friday, show a dashboard on Monday.

03

Built for the review, not around it

A SCADA gateway usually wants an inbound firewall rule. This calls out from the plant and never in, so there is no inbound port to argue about and no three-month security review before you see data. The full architecture, for that team.

Own it. Don't rent it.

A sovereign deployment: perpetual licence, not a monthly rental. Your nodes, your server, your data. No per-tag fees, no data-volume charges.

Perpetual licence + AMC

Pricing on request

Buy your edge nodes outright and run them on your own hardware. You own the software, not a subscription that stops when you do.

  • No per-tag fees, no data-volume charges.
  • Optional AMC keeps you current: updates, security patches, priority support, and new protocol connectors as they ship.
  • Prove the value on a single node before you commit.
  • Fully self-hosted, air-gapped deployment for OT environments that require zero standing vendor access.
Start with one line

EdgeBits Analytics

Add-On

Built-in cross-site OEE dashboards, energy monitoring, predictive maintenance, and ERP-enriched reports. Also connects to Grafana, Power BI, or any BI tool via REST. Available as an add-on with any deployment.

Historian Lite Basic MES OEE Dashboards Energy Reports Predictive Maintenance SAP Enrichment Write back to SAP

Perpetual licence + optional AMC, tailored to your fleet. Talk to us for a quote, or model your own numbers first — the calculator shows its working and cites every multiplier.

Frequently asked questions

Real questions from plant managers and IT heads, answered without the jargon.

Talk to us

Tell us what you're trying to connect and we'll set up a call. No pitch deck, no pressure, just a conversation about your factory and whether EdgeBits fits.

  • 30-minute call with an industrial IoT engineer
  • No credit card, no commitment
  • We'll review your protocol and IT constraints upfront

Or reach us directly at

hello@edgebits.io