We find the money your operation is losing. Then we stop it.
ThompsonMTech is an operational loss recovery firm. AMOLIS, our platform for industrial manufacturers, reads the data your plant already collects, finds the losses hiding inside it — yield drift, micro-stoppages, energy variance — and explains them in plain language. Read-only. No new hardware. No dashboards to babysit.
Your historian already has the answer. Nobody's reading it in time.
Every unplanned stoppage, every yield slip, every slow energy creep leaves a trail in the data your plant already collects. That trail usually surfaces only after someone goes looking — hours or days after the loss happened, and after the cost is already booked.
Dashboards make this worse, not better. They show you that something moved. They don't tell you why, what it cost, or what to do next. So the data sits there, technically available, practically unread.
One system, six capabilities, zero new infrastructure.
AMOLIS connects read-only to the systems you already run, continuously watches for the patterns that precede loss, and tells your team exactly what changed, what it's costing you, and what to check first.
Autonomous Loss Recovery Engine
Continuously scans operating data for the signal patterns that precede downtime, scrap, and yield loss — and quantifies the cost before anyone has to ask.
Multi-Modal Sensor Fusion
Combines historian tags, OPC UA streams, and MQTT telemetry into one coherent picture of plant state — instead of forcing your team to reconcile five screens.
Confidence Integrity Layer
Every flagged event ships with a confidence score and the contributing signals behind it — so your team can trust an alert without re-deriving it by hand.
Behavioral Anomaly Detection
Learns the normal rhythm of each line or asset class, then flags the early-stage drift that precedes failure — before it shows up as an alarm.
Natural Language Ops Interface
Ask AMOLIS what happened on second shift, in plain English, and get a plain-language answer — not a query builder.
Cross-Plant Intelligence
Patterns found on one line inform every other line and every other plant on the platform — loss-pattern knowledge compounds instead of resetting per site.
Illustrative example — not a live customer result
Yield Drift — Line 4 Extruder
- Barrel zone 3 temperature drifted +3°C over the last 6 hours
- Feed rate variance increased 11% versus the trailing 30-day baseline
- Pattern matches early-stage heater band degradation seen previously on this line
Three steps. No control-system changes, ever.
Connect, read-only
AMOLIS connects to your existing historian or OT data source — GE Proficy, AVEVA PI, OPC UA, or MQTT — with read-only credentials. No agent touches control hardware. No write access is ever granted.
Detect and explain
AMOLIS watches continuously and, when it finds a loss pattern, surfaces the contributing signals and a plain-language explanation — not just a threshold alarm with a timestamp.
Act on one verified metric
We agree on a single success metric before day one. Every result is measured against that metric, in writing — not a bundle of vanity statistics chosen after the fact.
AMOLIS sits between your existing data sources and your team — never between your team and your control systems.
Technical detail for your engineering team
AMOLIS connects read-only against GE Proficy Historian, AVEVA/OSIsoft PI, OPC UA, and MQTT sources. No agent is installed on control hardware, and no write access is requested or granted at any point. During a pilot, data is processed within your own Azure tenant rather than copied out to a shared multi-tenant store. Your IT/OT team retains the ability to revoke read access at any time, for any reason, without needing to coordinate an uninstall.
- Authentication
- Scoped, least-privilege service credentials issued by your team — never a shared or standing admin account. Credentials can be rotated or revoked unilaterally on your side at any time.
- Network path
- Outbound read calls only, over your existing OPC UA / Web API / MQTT broker endpoints. No inbound connection is opened to your OT network, and no new firewall rule needs to stay open for AMOLIS beyond the read endpoint you already expose internally.
- Data in transit
- TLS-encrypted between the read source and the AMOLIS engine inside your tenant.
- Data at rest
- Stored within your Azure tenant for the pilot term, not replicated to ThompsonMTech-owned infrastructure.
- Audit trail
- Every read connection and every surfaced finding is logged with a timestamp, so your team can independently confirm AMOLIS never issued a write call.
- Offboarding
- At pilot end or exit, access is revoked on your side and any pilot-specific deployment artifacts are removed — there is nothing left running against your systems by default.
Built for plants that can't afford to guess.
Food & Meat Processing
Catches yield loss from over-trim, giveaway, and unplanned line stoppages before they erode margin on thin-margin proteins.
Automotive & Tier-1 Suppliers
Surfaces micro-stoppages and quality drift across stamping, welding, and assembly lines before they show up on a supplier scorecard.
Plastics & Polymers
Flags energy and material variance in extrusion and molding processes that standard SCADA alarms are tuned to ignore.
Oil & Gas — Process & Terminal
Identifies throughput loss and energy variance across process units and terminal operations without touching control systems.
| Capability | Threshold alarms & dashboards | AMOLIS |
|---|---|---|
| What you get when something changes | A timestamp and a number out of range | A plain-language explanation of contributing factors |
| Setup effort | Manual threshold tuning per tag, per line | Learns normal behavior per line automatically |
| Cost framing | You infer the cost yourself, after the fact | Estimated cost impact attached to each finding |
| Control system access | Varies by vendor | Read-only, always — no write path, ever |
| Cross-plant learning | Resets per site, per dashboard | Loss patterns found on one line inform every other line |
Built to be trusted by people who can't afford to be wrong.
ThompsonMTech is an operational loss recovery firm. We're proving that model first in industrial manufacturing, where a historian already exists and a finding can be checked against real data — but the standard below doesn't change by vertical.
Read-only by design
AMOLIS never writes back to a control system. No agent is installed on control hardware, and no action is taken without a human reviewing it first. The pilot is structured so your OT team never has to take that risk on faith.
One metric, agreed in writing
Before day one, we agree on a single success metric specific to your operation — not a dashboard full of statistics chosen after the fact. Either side can exit at 30 or 60 days, no cause required.
Derisked, not just promised
We're actively pursuing SOC 2 compliance, carry a plan for technology E&O and cyber liability coverage, and offer third-party source-code escrow as a standard pilot term — so a 90-day engagement doesn't depend on any one person showing up tomorrow.
Built by someone who's read the data for a living.
ThompsonMTech is an operational loss recovery firm built by EJ Thompson, a data scientist and analyst with a background at NewRez and TD Bank, now full-time on AMOLIS. The thesis is simple: most operations generate more data than anyone has time to read, and the losses hiding in that data are real, specific, and recoverable — if something is actually watching.
AMOLIS is where that thesis is being proven first: built for plants that already have a historian and already know something is being lost. It's not another dashboard. It's the read of your data that nobody on your team has hours in the day to do manually, delivered continuously and in plain language.
- Based
- Spartanburg, South Carolina
- Background
- Applied data science & analytics — NewRez, TD Bank
- Stage
- Solo-founder, full-time on AMOLIS, building out an advisory bench
- Current vertical
- Mid-size industrial manufacturing — food & meat, automotive, plastics, oil & gas
"We're not asking anyone to bet on a one-person company. We're asking for 90 days, read-only, on one line, to see if AMOLIS finds something your team didn't already know." EJ Thompson, Founder
The questions plant and ops leaders actually ask.
What happens to our data during and after the pilot?
Data stays inside your own environment for the duration of the pilot. ThompsonMTech connects read-only and does not copy production data out to run AMOLIS. Exact retention and deletion terms are spelled out in the pilot agreement before anything is connected.
What if AMOLIS gets it wrong — false positives?
AMOLIS produces recommendations and explanations, not control actions — a person on your team always reviews a finding before anything changes. False-positive rate is a fair thing to put inside the single success metric we agree on together before day one.
Why trust a small, focused team with this?
Lead with the structure, not a sales pitch: a 90-day pilot, free of charge, read-only, with a written exit at 30 or 60 days and no cause required. SOC 2 work, technology E&O and cyber coverage, and source-code escrow are part of how the engagement is derisked regardless of team size.
What does this cost after the pilot?
Pricing isn't set ahead of the pilot on purpose. The goal of the 90 days is to agree on what AMOLIS actually finds and what that's worth to your operation — pricing follows from that conversation, not the other way around.
What's the integration effort on our side?
For a read-only pilot against GE Proficy, AVEVA PI, OPC UA, or MQTT, your IT/OT team typically needs to provision read credentials and confirm network access — no agent is installed on control hardware, and no control-system changes are required.
How is this different from historian analytics modules we already have?
Most historian analytics modules require your team to configure thresholds and build the models that define normal versus abnormal. AMOLIS is built to take on that configuration work itself and explain findings in plain language rather than requiring a dashboard analyst to interpret them.
Have other manufacturers already run this pilot?
AMOLIS pilots are opening now with a small number of manufacturers. The pilot structure — free, 90 days, read-only, exit at 30 or 60 days — is built so you don't have to take being early on faith.
What if we don't know what success metric to pick?
That's normal, and it's the first thing we work through together before day one. We typically start from whatever your team already tracks — downtime minutes, scrap or giveaway rate, time-to-detection on a known failure mode — and agree on one specific, measurable version of it rather than starting from a blank page.
Do we need a dedicated person on our side during the pilot?
No new headcount is required. Your IT/OT team spends a small amount of time up front provisioning read credentials and confirming network access, and whoever owns the chosen line reviews findings as they come in — typically a few minutes at a time, not a dedicated role.
What we're proposing, in plain language.
| Duration | 90 days |
| Cost | Free of charge |
| Deployment | Read-only — no write access to any client system, ever |
| Data residency | Stays inside your own environment for the full pilot |
| Exit | Either party may exit at 30 or 60 days, no cause required |
| Intellectual property | ThompsonMTech owns AMOLIS; you own your data; no joint IP created |
| Success metric | One specific metric, agreed in writing before day one |
| In exchange | Logo use and a co-approved case study, only if the pilot succeeds |
What 90 days looks like, in three phases
Connect & baseline
Read-only credentials provisioned, connection verified, and AMOLIS establishes a baseline of normal behavior for the chosen line or equipment class.
Detect & calibrate
AMOLIS surfaces findings continuously. Your team reviews each one and gives feedback, which sharpens confidence scoring specific to your operation.
Review against the metric
We sit down and compare results against the single success metric agreed on day one — and decide together what, if anything, comes next.
What we'll need from your team to start
- One named point of contact on your operations or IT/OT side
- Read-only credentials to the historian or OT data source you'd like to pilot on
- Confirmation of network access for a read-only connection
- A signed pilot agreement defining the single success metric
Find your loss. Stop your loss.
One 90-day, read-only pilot on a single line or equipment class. One metric we agree on together, in writing, before day one.
EJ Thompson · Founder, ThompsonMTech · Spartanburg, SC · ejthompson@thompsonmtech.com