Eleflow Talk to us

For mining and mineral processing

The intelligence layer for your plant.

Eleflow watches your cameras and process data around the clock, writes the shift report before handover, and works through any question your engineers bring it. It runs on your own servers, where a local AI directs expert models trained on your plant’s data, so your data never leaves the site.

  • Runs on your site
  • Local AI, expert models
  • No data leaves the plant
Illustration Unheard: a plant drawn in its own readings.

Why Eleflow

Your plant already has the answers.

The camera saw the spill. The historian shows the transmitter that froze. The assays explain the recovery drop. What’s missing is someone with the time to read it all. Eleflow gives your plant’s data that attention, every minute of every shift, so your people act sooner and spend their time on the plant.

Illustration How Eleflow works: the intelligence layer over your plant's data, its three modes, and how it runs privately on your site.

  • Always on

    More eyes on the hazards.

    Exclusion zones watched every minute on the cameras you already have. Each event reaches a supervisor with the frames that show it, not buried in hours of footage.

  • Always on

    Problems caught while they’re small.

    A spill spreading across a pump bay, a screen deck starting to blind, tramp metal heading for the crusher: flagged as they start, not found at the next inspection.

  • Always on · Every shift

    Recovery protected, shift by shift.

    A pH step, an analyser reading high or a tail grade creeping up shows in the shift it starts, not when the month’s numbers come in. Fewer shifts run off target before anyone knows.

  • Every shift

    Your engineers’ time back.

    Shift reports, production by period, lab against analysers and the balance, done on schedule in your format. Your metallurgists spend their hours improving the plant, not compiling reports.

  • On demand

    Answers while they still matter.

    When recovery drops, ask why and get the working with its evidence while the plant can still act on it, not weeks later behind month-end.

  • All three

    What your plant learns, it keeps.

    Every analysis is a notebook with its working, kept on your servers. When rosters change or people move on, the next engineer starts from it, not from scratch.

Three modes

Always on. Every shift. On demand.

One layer over your cameras, historian, LIMS and SCADA. Every deployment is set up for its site: we connect your systems, train expert models on your data and build the skills your plant needs.

01 Always on

Eyes on the plant, every minute.

  • Restricted zones. Anyone entering an exclusion zone, flagged with the frames that show it.
  • Vibrating screens. How material spreads across the deck, and where it starts to blind.
  • Conveyors. Oversize rocks and tramp metal, caught on the belt.
  • Leaks and spills. Slurry on the floor, sized and flagged early, checked against the pump's own pressure.
  • Flotation. Froth flow and appearance, cell by cell.
  • Process tags. Drift and excursions in your historian, as they happen.
  • Set-point advice. When the feed changes upstream past a model's threshold, a suggested set point with the reason, for your control room to decide.
  • And more.
Synthetic data A restricted zone, a screen deck, a conveyor, a pump-bay spill and historian tags: what Eleflow flags in each, with its evidence.

02 Every shift

The routine work, done before anyone asks.

  • Shift reports. Start and end of shift, in your site's own format.
  • Lab against online. Assays checked against online analysers and meters, with the report written.
  • Production. Shift, day, week and month, on your site's clock.
  • Metallurgical balance. Reconciled every period, with every adjustment explained.
  • Data rounds. The historian, stock surveys, lab results, workshop logs and the shift inbox, read on schedule and written into the report.
  • Set-point check. Your plant's models run against the latest feed, and any set point worth changing goes into the report with its reason.
  • And more.
Synthetic data Scheduled jobs and what they write: the day shift's report, production by period, the online analyser checked against the lab, and the daily balance.

03 On demand

Ask the question. Get the working.

  • Balances and flowsheets. Mass balances that close, and flowsheets with their recycles.
  • Concentrator. Comminution, cyclone partition curves and flotation kinetics.
  • QAQC. Duplicates, standards and control charts.
  • Chemistry. Water and leach chemistry by PHREEQC.
  • Documents. Standards, procedures and quotations compared, every change cited.
  • And more.
Synthetic data Questions asked in a chat and worked through in a notebook: a recovery drop traced to pH, a cyclone curve, CRM results, a water analysis and a revised standard.

Numbers you can stand behind.

  • Every number has its receipt.

    Each figure is traced to the tag, file and time it came from. Open any result and see how it was worked out.

  • It says what it cannot see.

    A camera blinded by dust is reported as not visible, never as clear. A missing unit becomes a question, not a guess.

  • Your plant's conventions.

    Your shift pattern, your production day, your units and your report format.

How we start

Set up for your site, not a template.

  1. Connect.

    Your cameras, PI, LIMS and SCADA, or the exports you already have.

  2. Set it up for your site.

    Expert models trained on your footage and data, and your zones, shift pattern, report formats and conventions.

  3. Start with one problem.

    A pilot on the question that matters most to your plant, then the next.

Private by design

Your data stays on your site.

Eleflow runs inside your network, on your own servers. The AI is local: an open-source language model on your hardware directs expert models trained on your plant’s data, with the tools and skills built for your site. Nothing goes to an outside AI service.

  • Local AI, not cloud AI. Open-source models served from your own GPU servers, with no calls to outside AI services and no silent fallbacks.
  • Expert models for your plant. Models trained and tuned on your own data, such as vision models on your camera footage. The language model picks the right one for each job and runs it with the right tools.
  • Nothing leaves the site. Set the network to closed and Eleflow talks only to your own model servers. Every outbound request is logged.
  • Private to each person. Every user works in an isolated workspace, and your models, files and results stay on your servers.
Your cameras, PI and SCADA, LIMS, exports and more connect to Eleflow inside your deployment, with your local AI and expert models; a gate at the edge sets network access to open, allowlisted or closed. YOUR DEPLOYMENT Cameras PI · SCADA LIMS Exports … Eleflow Local AI +expert models Gate open · allowlist · closed Your cameras, PI and SCADA, LIMS, exports and more connect to Eleflow inside your deployment, with your local models; a gate at the edge sets network access to open, allowlisted or closed. YOUR DEPLOYMENT Cameras PI · SCADA LIMS Exports … Eleflow Your Local Model(s) Gate open · allowlist · closed

Eleflow is built by Eleticle, a Brisbane company founded by George Bai, PhD, a mineral processing engineer turned data scientist.

About Eleticle

Contact

Talk to us about your plant.

Tell us what you want watched, reported or worked out, and where your data lives today. We reply within one business day.

Please do not send confidential plant data through this form. We use your details only to reply, as set out in the Privacy Policy.