Guide hub
AI and learning
Spall's learning layer builds its reference points from a machine's own history instead of a spec sheet or a manually set threshold. The guides below cover the golden run, the best sustained stretch a machine has actually produced, why it beats a nameplate rate as a speed reference, and the ordinary ways that reference erodes over time, setpoint drift, operator overrides, tooling wear, and how to catch it in the cycle trace before months of speed are gone.
14 guides
- 01 What a Golden Run Is, and Why It Beats a Nameplate Rate The best sustained stretch a machine has actually run, as the speed reference. Why it beats a spec sheet number, and how drift away from it shows up. →
- 02 Why Your Best Run Disappears Setpoint drift, operator overrides, and tooling wear all erode a machine's best pace the same slow way. How to actually see it happening in the cycle trace. →
- 03
~8 min
What Spall's AI Learns From Your Machines, and What It Does With It A per-machine model forms automatically from telemetry, production, and downtime history already flowing in. What it learns, what it needs, and where it stays silent. → - 04
~8 min
Ask Your Factory: Twenty Questions a Plant Manager Asks, and What the Answer Cites Plain-language questions about downtime, quality, shifts, and money, answered from live data with the sources named. What it can answer, and what it says when it can't. → - 05
~8 min
How Spall Predicts a Failure: From the Machine's Own History, With the Count Behind It A failure-risk percentage built from a machine's own count of past breakdowns: MTBF, the odds of a failure this week, what it costs to be wrong, and when there isn't enough history to say. → - 06
~7 min
How AutoML Turns Your Best Run Into a Live Reference The AI mines every completed cycle for the fastest sustained stretch a machine has run, then watches recent cycles against it continuously, with the chart and the dollar figure behind the finding. → - 07
~8 min
Alarm Sequences That Precede a Stop: How a Precursor Is Found and Turned Into a Trigger How Spall mines a machine's alarm history for codes that reliably fire before a downtime reason or a scrap event, and turns a real pattern into a one-click trigger. → - 08
~8 min
AutoML on Any Controller: Learning Normal From Telemetry Shape, Not a Brand's Tag Names How Spall decides which signals a model can fairly learn from, using what a value actually does over time instead of a tag name, a protocol, or a role a person forgot to set. → - 09
~7 min
AI Without a Data Scientist: What Needs Configuring, and What Needs Deciding Spall's model selection, training, and scoring run with no setup screen at all. What a plant still has to decide: cost rates, reason codes, and which findings to act on. → - 10
~7 min
What the AI Needs From Your Plant: Data Readiness in a Week Which models come online in a day, which need a full week, and which take longer, mapped against what a floor actually has to do to get there. → - 11
~7 min
How to Read an Insight Card: Evidence, Confidence, and the Blank That Means Something Anatomy of an ML insight card: what the title claims, what the evidence backs it with, what the dollar figure means, and what a blank or a learning state is actually saying. → - 12
~8 min
From AI Finding to Verified Savings: The Loop That Proves It Linking an insight to the action taken opens a measurement window. What comes out the other side: verified, no change, or worse, priced only against a real rate, confounders named either way. → - 13
~7 min
Connect Your AI Assistant to Spall With MCP Point Claude Desktop, Cursor, or Copilot at your Spall tenant with one package and a read-only key. What each tool reads, what it never does, and the ask scope and budget. → - 14
~6 min
Your Plant in Claude, Copilot, or Cursor: What a Read-Only MCP Server Does, and What It Never Does Why a plant would let its own AI assistant read Spall, what fourteen read-only tools return, the basis every answer carries, and the three things the package refuses to do. →
Looking for the full set? See every guide →
Bring your machine list. We'll tell you exactly what plugs in.
30 days, hardware included, line pilot $4,500 or plant pilot $8,500, fully credited when you expand.