Skip to main content
Bivrost 彼络

An ops AI coworker for machine shops.
One that does the work itself.

It runs on the equipment and maintenance data you already have. Reports write themselves, work orders get chased, and repair history answers in one question.

Bivrost Factory AI workbench: projects, conversation, and the skills / instructions / tasks panel
100 +
Factories served
10,000 +
Machines connected
6 yrs+
Of shop-floor data collected
Hexagon
Digitalization partner
Kanghui Medical
Medical devices
First Fluid Power
Machining · 200 CNCs
Jiehe Motor
Motor manufacturing
The challenge

Seeing your data
is not the same as getting answers.

Ten years of industrial IoT solved connectivity. It did not solve reading the data, asking it questions, or closing the loop. In most machine shops the hard part is not collection — it is everything after.

01

The system replaced paper. It did not replace the person.

The data is in a system, but daily logs, work orders and reports are still exported, typed in and rebuilt in Excel by hand. Not one step fewer, not one person freed.

02

Plenty of data, no knowledge base.

How this machine was fixed last time, which spare part went in, what to watch during service — two or three veterans are still the only ones who can tell you. The experience sits in their heads and on paper, never in a system.

03

You can see it. You cannot act on it.

The report tells you this machine broke and that one stopped. It does not raise the work order, assign the technician, or chase it to closure. Between the alert and the action, there is still a person.

Cloud foundation → AI layer

Five ops jobs. Bivrost Cloud keeps them in order, Factory AI keeps them moving.

Asset records, maintenance, spare parts, repair requests, reports — Bivrost Cloud already holds this data. Factory AI sits on top of it and turns the data into judgments and actions.

01 Asset records
Bivrost Cloud

A digital ID per machine, full lifecycle history, records that match the floor.

Factory AI

Flags gaps between the record and reality. Ask it anything: what this machine has cost in repairs, whether it is still worth fixing, which unit of the same model fails most.

02 Preventive maintenance
Bivrost Cloud

Calendar-based service plans; tasks and reminders generated when they fall due.

Factory AI

Predicts which machine needs service early, from run data and failure history. Calendar-based becomes condition-based, written up as a work order awaiting your approval.

03 Spare parts
Bivrost Cloud

Safety and maximum stock levels, automatic alerts, work orders linked to parts.

Factory AI

Predicts what to stock and how much from failure data and machine age, then drafts the purchase request.

04 Fault response
Bivrost Cloud

Scan to report, accept the job, sign it off — progress traceable in real time.

Factory AI

The moment a fault is reported it pulls up similar past cases: likely causes, repair steps, parts used. Then it assigns the job and chases it to closure.

05 Decisions from data
Bivrost Cloud

OEE / MTBF / repair cost on one definition — but the reports are still exported and rebuilt in Excel by hand.

Factory AI

Daily reports, weekly reports and analysis generated straight from your existing data. No imports, no exports, no Excel. Ask in plain language: why OEE dropped this week, which machine burns the most repair budget.

Bivrost Cloud already does the first four; Factory AI turns them from passive to active. The fifth — reporting — is still stuck in manual work almost everywhere, and it is the first thing Factory AI takes off your desk.

Factory AI · the ops workbench

The ops work is done before you get to it: reports written, work orders chased, repair history answered.

Machine alarms, inspections, maintenance, work orders and past repairs live in one workbench. AI works on the data you already have — it writes the reports, closes the loop, answers the faults.

Bivrost Factory AI workbench overview: project sidebar, conversation area, instructions / files / skills / scheduled tasks panel
Automatic weekly report · replaces manual Excel
Ops weekly report conversation: unit downtime, work-order status and a one-line takeaway, written by AI
Scenario 01

Reports write themselves

Daily reports, weekly reports and analysis are generated on the data you already have. No manual imports or exports, no Excel. You never build the same table twice.

Work-order loop · assigned and tracked automatically
Work-order panel: repair request, assignment and closure status tracked in real time
Scenario 02

Work orders chase themselves

From the repair request to the assignment to the sign-off, AI watches the progress of every work order. Between the alert and the action, there is no longer a person.

Repair knowledge base · experience that stays
Repair knowledge base Q&A: similar past faults, likely causes and the parts used
Scenario 03

Repair history, answered on the spot

How the same model was fixed last time, which spare part went in, what to watch during service — past work orders become a knowledge base. Ask and get an answer, without waiting for the one veteran who knows.

“The system replaced our paper forms, but the intelligence was thinner than we expected — a lot of the work is still manual: import, export, build the Excel. We have data, but it never became a knowledge base.”
— Maintenance lead, Jiehe Motor
Foundation · three layers

The AI can answer like this because of a data platform that has run for six years.

The gateway solves connectivity. Bivrost Cloud solves visibility. The IoG network makes multi-site deployment a single job. All three had to exist before AI could say anything useful about your factory.

Protocols in production FanucSiemensMitsubishiHeidenhainMakinoCitizenMurataLNCGSK
Layer 03 · Cloud

Bivrost Cloud

Cloud Platform

An IIoT device management platform built for the shop floor. OEE / output / alarm analysis, full equipment lifecycle, the repair-inspection-maintenance loop, and NC programs read and written from the cloud. One codebase for public cloud, private cloud and on-premise.

  • Equipment assets · spares · repair-inspection-maintenance loop
  • Live OEE on one definition, compared across machines
  • NC program transfer: 30 min → < 1 min
Layer 02 · Gateway

Bivrost Gateway

IoT Gateway

One point of entry for the machine shop. Up to 255 devices per gateway on the same LAN, sampled at millisecond resolution. HTTP / MQTT / MODBUS and a two-way API let every upstream system consume all machines through a single interface.

  • Adapters for mainstream CNC / PLC / robot protocols
  • Edge buffering · pre-processing scripts · remote control
  • Writes as well as reads: NC programs, tool offsets
Layer 01 · Network

IoG Network

Internet of Gateways

A distributed gateway network that spans networks. A software bus merges public internet, plant intranet, headquarters and satellite sites into one fabric: build against one API, deploy in many places, share data across nodes.

  • Pooled hardware resources, compute that moves between nodes
  • One scheduler over a distributed database
  • Deploy once across sites, consume the data anywhere
Architecture

The whole architecture, in one diagram.

Hard device communication is pushed down into the gateway and the edge. Visualization, analysis, workflow and collaboration sit in one workbench. Hard to deploy and hard to use are solved on two separate tracks.

Devices Gateway Edge / IoG Cloud Factory AI CNC · PLC · robots Cutting · bending Sensors · I/O modules Bivrost IoT Gateway 255 devices per unit millisecond sampling IoG Network software bus one API pooled resources cross-node sync Bivrost Cloud OEE · output · alarms equipment lifecycle program transfer public or private Factory AI Workbench natural language project context scheduled tasks Machining data layer · state / alarms / programs / tool / feed / speed / load / OEE / cycle time Connect once, consume anywhere

Roadmap
Ops is the first domain Factory AI has proven. The same foundation will expand domain by domain into production (MES) and supply chain (ERP).

Solutions

Built for the ops lead first. Then plants, integrators and builders.

Ops is the first domain Factory AI ships in. Every other role runs on the same foundation and enters where it matters to them.

For · maintenance & equipment

Ops & equipment leads

Fewer reports, less firefighting

The repair-inspection-maintenance loop is already running. Add a layer of AI on top and it takes over the work that eats your week, so you can go after the real problems.

  • From building the weekly report → AI writes it, and the hours come back to you
  • From waiting for the veteran → ask the knowledge base, anyone can handle it
  • From spotting the problem → work orders assigned and chased automatically
Ammunition for the budget conversation
Labour saved Hours near zero

The weekly report and the import-export routine stop costing hours

Downtime cut Measurably shorter

Faster diagnosis from the knowledge base, alerts before failures

Risk down The cost nobody prices

Veteran experience captured in the knowledge base, so it does not walk out the door

“We do not need a full MES. Put a layer of AI on the equipment data we already have, and start by earning back what these two jobs cost us.”

Same foundation · other roles
For · plant management

Machine shops

Fastest to live, visible results

No full MES required. The gateway is plug-and-play and the first dashboard is up in 30 minutes. Management, line leads and equipment staff each take what they need from one set of data.

  • Live OEE / output / alarm dashboards
  • Program transfer replaces USB sticks, under 1 minute
  • Full equipment lifecycle, digitized
For · system integrators

Software integrators

Open interfaces, lower delivery cost

Bivrost standardizes machine-tool communication underneath and exposes one API above it. Integrate once, reuse across plants, and stop paying for every machine’s protocol.

  • Two-way API over HTTP / MQTT / MODBUS
  • Up to 255 devices per gateway, millisecond sampling
  • OEM / white-label options
For · equipment manufacturers

Machine builders

Ship it built in, skip the R&D

The gateway can be pre-installed in your machine or solution, so the product is digital out of the box. No communication stack to build, and a fast route into your customers’ data ecosystem.

  • Full protocol coverage for mainstream CNC / PLC / robots
  • On-premise deployment keeps data inside the plant
  • OEM / rebadge friendly
Customers

100+ manufacturers connect their factories with Bivrost.

From China’s big four automotive groups to precision medical devices, from construction machinery to university labs — six years in production.

Hexagon
Kanghui Medical
First Fluid Power
Schneider Electric
China Unicom
FAW-Volkswagen
Changan Automobile
China Lesso
Dongwu
Chengfeng
Weiben
Beijing Institute of Technology
Hualei Xuntuo
Hollysys
Anyuan Technology
Andataike
Ningmengdou
LiuGong
Gechuang
Wuhan Fingu Electronic
Pneumatic Research Institute
Zhejiang University
Panyu Polytechnic
Shengyuancheng
Xiangyu Medical
CASICloud
Northwestern Polytechnical University
Xuantian
Chongqing No.2 Machine Tool
Chongqing University
Ningbo Jingcheng Motor
Allead Precision
Jiangsu Chuangyuan
Zhongkang Powertrain
Nancal Technology
Customers

Six years running in factories like these.

Partner

Hexagon

Precision metrology equipment · RMB 1B+ scale

Machine tool brands are many and their protocols all differ. Adapting to each system in-house would have cost Hexagon heavily in money and delivery time. Bivrost pushed multi-brand machine communication down into a single gateway, so Hexagon’s data platform never pays for device protocols again.

Hexagon digitalization partner
What we delivered
A heavily customized machine-communication gateway that packs sensor collection, machine-tool collection and analytics into one device.
Business outcome
  • Device integration cost down sharply
  • Digitalization projects shipped faster
  • A long-term digitalization partnership

First Fluid Power

General equipment · machining
200 CNCs connected and monitored
30 min
< 1 min
Program update per machine

Management follows production from automatic roll-ups; line leads watch alarms and utilization rankings on the dashboard; cloud program management replaced the USB-stick routine.

Kanghui Medical

Medical devices
One OEE definition across mixed, ageing machines
Many brands
One OEE
Collection and enterprise integration

The shop runs many machine types, much of it old. A standardized Bivrost gateway collects across all of them and feeds the enterprise systems, making downtime and maintenance cost visible.

Jiehe Motor

Motor manufacturing
Equipment lifecycle, digitized
5 modules
1 dataset
Repairs · spares · inspection · service

A customized rollout of the Bivrost Cloud equipment management modules: scan-to-report repairs, low-stock alerts on spares, three-tier maintenance checklists — all on one data foundation. AI-written reports and the fault knowledge base are being piloted on this same data.

Automotive OEM

Vehicle manufacturing · China’s big four
33 plants on one connection layer
Many protocols
One pipeline
14 production bases worldwide

Equipment is spread across domestic and overseas bases, with many brands and different protocols. Bivrost gateways are plug-and-play, and a low-code PaaS lets management build cross-site dashboards without writing code.

Compressor parts maker

Precision machining · Midea supplier
Tool life managed across 300+ CNCs
Manual estimates
Data-driven
Tool wear: from gut feel to prediction

Running tools too long caused quality issues, changing them early wasted consumables, and fault response was manual. The Bivrost TMS module drives tool-change decisions from live machine data — quality, cost and response time improved together.

Next step

Connect every machine on your floor
to the decisions you make.

Give us 30 minutes. We will run your existing equipment data and produce the first automatic report live — without exporting a single Excel file.