OpsentinelOpsentinel
SpecGuardOpsMindInsight EngineAbout
Live Ops · Pattern Detection

The customer who keeps changing specs.
The machine that keeps slipping.

See the patterns before they cost you a customer.

These patterns are obvious in hindsight. Heads-down running the floor, you can't see them until lunch. OpsMind reads everything you already produce — sensors, shift notes, schedules, QuickBooks exports — and tells you in real time.

Start with 3 free analyses →Schedule a 30-min walkthrough

What is this?

OpsMind watches the data your shop already produces — telemetry exports, shift notes, schedule CSVs, QuickBooks data — and surfaces patterns you'd otherwise miss. Customers who quietly slip on delivery. Machines that drift. Job types that consistently lose money.

It runs continuously and emails a daily digest, with severity-graded alerts when something needs attention sooner. Every alert links back to the underlying rows so your team can verify the pattern themselves — no “the model said so” black box.

Built for

  • Job shops and contract manufacturers, 10–200 employees, $5M–$100M revenue
  • Production managers tracking 30+ active jobs/month across multiple customers
  • Older shops still on Excel + paper logs (no sensors required to start)
  • CNC operations with machine telemetry / MES exports they aren't currently mining
  • Shops doing aerospace, defense, oil-and-gas, medical, or semiconductor work

Probably not for you if

  • Single-machine shops with 1–2 customers — the patterns aren't hiding
  • Teams already on a polished MES + ERP stack with AI-driven anomaly detection
  • High-mix prototype-only operations where every job is genuinely one-off
  • One-product OEMs running single-SKU production at high volume

Walk through a week with OpsMind

1Connect a data source
opsentinel.io/settings/integrations
Demo

OpsMind · Connect a data source

Scheduled CSV drop

● Synced

Folder

supabase://drops/job-schedule/

Cadence

Every 15 minutes

Last sync

2 minutes ago · 1,432 rows

Status

Healthy · 14 days uptime

Or drop a one-off CSV:

Drag a telemetry CSV or shift-notes export here

Schedule a CSV drop folder, or paste a one-off file. No ERP integration needed.
2Live operational dashboard
opsentinel.io/dashboard
Demo

OpsMind · Live Job Map

13 active · 2 alerts open
Intake2 jobs
Material3 jobs
Fabrication4 jobs
Quality2 jobs
Parts1 jobs
Loading1 jobs
Acme · CNC-03 · running hot
Every active job on a stage you defined. Severity overlays surface what's drifting.
3Daily digest in your inbox
opsentinel.io/inbox/opsmind-daily-digest
Demo

From: digest@opsentinel.io · Friday 4:02 PM

OpsMind daily · 3 patterns this week

high

Customer Acme has slipped delivery on 4 of last 6 jobs (vs 1 in 6 across other customers). Avg slip 8 days.

high

CNC-03 has run hot on 3 of those 4 Acme jobs. Bottleneck on Acme parts specifically.

medium

Aluminum-housing job type averaging 22% over budget across last 8 jobs. Recert + tooling-change time the biggest contributors.

Every alert links back to the underlying rows · Same data → same patterns

Severity-graded. Acme is slipping. CNC-03 is the cause. Decide before lunch.
4Drill into any alert
opsentinel.io/opsmind/alerts/acme-delivery-slip
Demo
High · ACMEPattern detected 9 days ago

Acme Aerospace delivery slip — 4 of last 6 jobs late, avg 8 days.

Avg days late · Acme jobs · last 12 weeks

Q1 baselineNow: +43%

Co-occurring

CNC-03 hot-runs (3 of 4)

Underlying rows

View 6 jobs · 1,847 rows

Trend over time, co-occurring patterns, link back to the rows. Verify in seconds.

A specific scenario

Friday afternoon.

You get the OpsMind daily digest in your inbox. It says:

  • HIGH — Customer Acme has slipped delivery on 4 of their last 6 jobs (vs. 1 in 6 across the rest of your customer base this quarter). Average slip: 8 days.
  • HIGH— CNC-03 has run hot on 3 of those 4 Acme jobs. It's the bottleneck on Acme's parts specifically.
  • MEDIUM — Aluminum-housing job type is averaging 22% over budget over the last 8 jobs. Material recert time and tooling-change time are the two biggest contributors.

You decide whether to call Acme, re-route their next job to CNC-04, or both — Monday morning, with data, instead of three months from now when a late delivery costs you the relationship.

What you get

Sensor mode for telemetry

10 anomaly categories run automatically on machine data — temperature spikes, vibration, spindle load, cycle time drift, scrap clusters.

Notes mode for everyone else

Drop in shift notes or a job log; OpsMind reads the prose and surfaces patterns: which customer slips, which machine runs hot, which job type loses money.

Daily digest, not 100 alerts

One email a day with just the things worth acting on. Severity-bucketed, actionable, with the operator note that triggered it.

How it actually works

Ingestion

Schedule a drop folder (Supabase Storage, S3, or Dropbox-style), or paste CSVs directly. Column-name normalization built in; you don't need a perfect schema. CSV / Excel / JSON / TXT / LOG / TSV all supported out of the box — no integration tickets.

Sensor mode + Notes mode

OpsMind classifies the upload by what's in it. Timestamped sensor columns get the anomaly path: rule-based threshold checks across 10 categories, plus an LLM consolidator that groups related signals into one alert per machine. Shift notes or a QuickBooks export get the notes path: an LLM reads the prose and surfaces customer behaviour, schedule slips, recurring failure modes. Same alert shape comes out either way.

Pattern detection

Statistical baselines per customer, per machine, and per job-type. Detect slips, drifts, and step changes — not anomaly-of-the-day, but trends-over-weeks. An alert fires when something meaningfully diverges from your shop's own baseline, not a global one — and when a row carries both a customer and a job-type, OpsMind picks whichever segment the reading is more anomalous against.

Severity grading

High / Medium / Low based on dollar impact and recurrence. A single late delivery is a Low; a customer slipping on 4 of 6 jobs is a High. Calibrated against your team's actual resolved-alert lifecycle decisions — so the system gets sharper as you use it.

Citation + determinism

Every alert links back to the underlying rows so you can verify the pattern yourself. Same data → same patterns; new patterns are explicitly flagged as new. Reproducible daily digests, not the same email-with-different-words every morning.

Under the hood

  • Customer scorecards — late rate, scrap rate, last note, health colour, ranked by activity
  • Machine scorecards — alert load by severity, last alert title, ranked by trouble
  • Alert calibration — "94% of resolved alerts were confirmed real" computed from your team's lifecycle decisions
  • Drop-in CSV / Excel / JSON / TXT / LOG / TSV support — no integration tickets
  • Drop-folder ingest from a scheduled CSV / Excel / JSON drop folder — no integration tickets to your IT team
  • Email-in for shift reports — forward shift summaries or daily exports to opsmind@; content lands in your dashboard within a minute or two (when configured)
  • Pattern badge in the alert UI distinguishes anomaly (sensor) from pattern (prose) so operators calibrate trust correctly

What it doesn't do (yet)

  • Real-time machine-bus integration (MTConnect, OPC-UA) — CSV drops only today.
  • Predictive maintenance — we name patterns; we don't predict bearing failure.
  • Auto-rerouting jobs — we surface what's happening; you decide.
  • Native ERP connectors (JobBoss, ProShop, E2, Epicor) — on request, schedule TBD.
  • Sub-hour latency — digests are daily; alerts are within 24 hours of the data drop.

See it on your own data.

Send a month of data — telemetry CSV, shift notes, an Excel schedule, whatever you have. We'll show you the customer behaviour patterns and machine slip signals OpsMind would have caught last month. Honest pilot, defended ROI number from your own data.

Start with 3 free analyses →Email us about a pilot

Pricing starts at $3,500/mo for OpsMind alone. See full pricing →

US-hosted · zero-data-retention LLM · NDA before any pilot · Security overview →