We automate despatch paperwork, cut losses on the plant floor, and give you production and carbon numbers that stand up to an audit.
Three documents from one weight entry. A person confirms the weight. The system does the filing, and keeps the proof.
In brief
EigenSpark builds AI systems and the software around them for steel producers, mining companies and heavy manufacturers. We automate the despatch chain end to end, from the weighbridge through the transit pass, the invoice and the e-way bill. We use the data your plant already records to cut downtime, scrap and energy cost. And we assemble the carbon, royalty and safety numbers you have to report, with the evidence attached.
What we cover
Carbon, exports, mineral despatch and safety. We build the numbers and the proof behind them.
The starting position
Five systems that disagree, work that happens outside SAP, and no two sites set up the same way.
Weighbridge, energy meters, the historian, SAP and a paper register. An auditable figure needs every input traced back to where it was recorded.
A despatch takes five manual entries, a handwritten gate slip and a transit pass typed into a portal. The record ends up correct, and hours late.
Different states, portals, weighbridge vendors and local practice. A system built for one site fails at the second unless the rules are configuration.
Decades of judgement about a furnace or a seam, almost none of it written down. Capturing it is an interview problem with a deadline.
How an engagement runs
Assess the functions, build one workflow at one site, then hand it to your team.
A structured read of the functions you actually want changed, at the corporate office and at representative plants and mines. Process mapping of the high-volume workflows, a data and systems audit for each one, and every opportunity ranked by impact and feasibility.
A single workflow taken to production at one location. A dispatch module goes from kickoff to go-live in seven to eight weeks, including SAP and portal integration, user acceptance with the operators who will run it, and a parallel run against the manual process.
Tiered programmes so the organisation can specify, judge and run what gets built. Board and executive fluency, business tracks for plant, procurement, safety and commercial teams, and technical depth for IT and the analytics group.
Use cases
Mines and despatch, then plant operations and safety, then procurement, then the back office.
Trip creation, gate-in, weighbridge, gate-out and closure in one system. The transit pass, the invoice and the e-way bill come off the weight and file themselves.
Systems Integration & APIs →Tare and gross captured against a trip, overload blocked before any document is raised, and weight variance tracked by vehicle, transporter and route.
Data Engineering & Platforms →Registration, insurance, fitness and licence checked at trip creation and again at the gate, with cloned plates and swapped trailers flagged from the camera record.
Agentic AI →Incident reports, near-miss records and shift logs read together to find the conditions that keep repeating, with the DGMS clock tracked against each event.
AI & Data Strategy →Inspection logs, shift reports and permits checked against the written procedure, with each deviation linked to the clause it breaches and tracked until closed.
Continuous Compliance Monitor →Vibration, temperature, load and fault history read together to spot the pattern before a failure. Work orders drafted with the spares and vendor history attached.
Predictive Maintenance System →Bids read against your eligibility criteria, price format and specification, returned as a matrix with every deviation linked to the clause it fails.
Contract Risk Analyser →Specifications, datasheets and drawings compared against the standard or the last revision, returned as a difference report with the page and field recorded.
Document Intelligence Engine →Fuel, power, process and production data assembled per tonne and per consignment, with the source trail a verifier asks for and a calculation your team can change.
Data Engineering & Platforms →Consumption history, production plans and lead times read into a spares forecast, with slow-moving stock surfaced and reorder points a buyer can accept or change.
AI & Data Strategy →Rake arrivals, yard use and despatch reconciled across FOIS, SAP and the manual records, with demurrage exposure visible before you incur it.
Agentic AI →The judgement of retiring engineers moved into a searchable, cited record. Interviews, manuals and incident history become an assistant that shows its source.
Generative AI →The engineering half
Four layers of ordinary engineering. They decide whether any of it survives an audit.
We read what your hardware emits and stop there. Sensor networks and industrial vision on physical equipment are a specialist trade and we partner for them.
SAP holds the record and the government portals hold the permission. Most of the effort in a heavy industry project sits between them.
Web screens built for a weighbridge cabin and a security gate, with role-based access and an audit entry on every action.
Defined once and used everywhere, so the plant report, the board pack and the compliance filing cannot disagree. Your team runs and owns it.
We also run the training that goes with it: SQL and data engineering, cloud, enterprise systems and cybersecurity, alongside the AI tracks. A team that cannot query its own data cannot specify a model either.
What we offer steel, mining and heavy industry teams
Most companies start with one workflow or one assessment, and use three of the four.
Where a multi-site programme starts.
Where a single workflow starts.
AI tracks, and the foundations under them.
The teams that own this work.
Working in a specific function? See how we help Supply Chain & Operations teams.
Security and controls
A person confirms, every document is filed against its trip, and the rules stay yours.
The operator reads the weight, the security officer verifies the load, the buyer accepts the recommendation. The system validates, generates and records; it does not release a truck or award a bid on its own.
Transit pass, tax invoice and e-way bill are stored against the trip that produced them and retrievable by anyone authorised, years later, without asking us.
Overload limits, validity rules, compliance checklists and portal parameters are managed in a console. A new state, a changed limit or an amended rule does not need a release.
Every action carries a user, a timestamp and a before-and-after. In a public sector engagement that record is what a CVC or CAG query is answered with, and it is designed for that from the first day.
FAQs
No. SAP stays the system of record. We automate the transactions that are being typed into it by hand, calling the standard interfaces for delivery creation, goods issue, billing, invoice release and IRN generation, so a despatch that took five manual steps becomes one. Each step is executed in sequence with error handling between them: if one fails, the system records which steps completed and lets the operator retry from that point or finish the trip manually.
Not to start. A first phase runs on the screens you already have, with the operator reading the weight off the existing indicator and entering it. Everything downstream is automatic. Reader and indicator integration, automatic weight capture and printing at the cabin are a second phase, and they can be added at any point after the software is live.
Not the hardware layer, and we will say so early. Sensor networks, PPE and conveyor vision, wearables and instrumentation are a specialist trade with specialist vendors. We build the software, the integration and the analytics on top, and we work alongside whoever supplies the hardware. The one place we do use cameras is number plate recognition at a gate, which is a software problem on a standard camera.
One system, configured per site. The workflow is common; the transit pass rules, the mineral portal, the weighbridge vendor and the local practice are not. Those live as configuration your team owns, which is what lets the second site go live in a fraction of the time the first took.
Seven to eight weeks from kickoff to go-live, on a scope covering the transporter portal, gate in and out, the weighbridge, closure, and the three external integrations. That includes a site visit in the first fortnight, user acceptance with your operators, and a parallel run against the manual process before it is switched off.
It is designed around them. Approvals stay with the officers who hold them, every action is logged with a user and a timestamp, and the system produces the evidence trail a vigilance or audit query asks for. Bid evaluation tooling is built to sit inside Integrity Pact conditions and existing tender guidance, and it recommends; a person awards.
No, and the parts that are not are usually what makes the AI work. SAP integration, the portal connections, the data model and the operator screens are ordinary engineering, and they are most of the effort. The same is true of the training: alongside the AI tracks we run SQL, data engineering, cloud, enterprise systems and cybersecurity.
In your tenancy, or on your own hardware where a plant or a mine requires it. Connectivity at a mine head is often the real constraint, so the operator-facing modules are built to keep working through a portal outage and to synchronise when the link returns. Where a hosted model is used, we agree in advance what is sent to it and what never leaves.
Related
The services this ships as, the product behind the tender work, and the teams that own it.
SAP, the state portals, the weighbridge and the e-way bill.
→ ServiceThe meter, production and dispatch data under every number.
→ ProductTender and offer scrutiny against the criteria you published.
→ FunctionThe function that owns dispatch, yard and inventory.
→ TrainingSQL, data engineering and enterprise systems for plant teams.
→ PillarThe engineering side, for systems that run a live operation.
→Send us one despatch SOP, one tender file, or a month of weighbridge logs. We call you within 48 hours and go through what can be automated end to end, what has to stay with an operator, and what a first module would take.
Talk to us