The three-part pattern. Not a sector SKU.
Spectacle fits where demand is uncertain, capacity is fixed in the short run, and labour and material must be committed before demand is known. Pattern, not a sector. Agents on Spectacle — Freigabe.
I use one fit test before I talk about agents on Spectacle.
Demand is uncertain. Capacity cannot change in the short run. Labour and material must be committed before demand is known.
If those three do not hold together, I do not force the product. If they do, the conversation is about work inside that bind — not about a dashboard, not about a chatbot, not about a sector homepage.
Why the three matter
Most software pitches start with features. This piece starts with the bind.
If demand is known far enough ahead, capacity can flex overnight, and you can wait to commit, you already have room. Excel, WhatsApp, and Erfahrung may be enough. That is not the buyer we are writing for.
The buyer we care about still loses mornings in redeciding because the bind is structural. The Dienstplan goes out before the day proves itself. Wareneinsatz is locked before the order book closes. The van is loaded before the tickets settle. Someone still has to re-rank when the morning hypothesis ages.
That someone is usually the Inhaber or Betriebsleitung. A late BWA does not help at eleven o’clock. A wall of numbers that did not name a measure is still reporting.
The three parts are not a mood. They are the operating pattern that decides whether agents on Spectacle have a job to do.
What agents do inside the pattern
Spectacle is the layer over the Betrieb you already run: ontology as objects and relationships; forecast and simulation in code; agents that take hand-work; actions with Freigabe.
Agents assemble what ERP, Warenwirtschaft, and Zeitwirtschaft already hold into a named measure. A person releases it. Logged. Reversible where the source system allows.
Arithmetic stays in code. The model only writes language. We will not auto-assign shifts without human release. We will not write sealed systems of record. We will not invent accuracy bands so the pitch sounds finished.
The useful output on a live day is often small: one ranked move for the hours left. Call in. Send home. Hold. Hold still counts when eleven o’clock did not change the ranking. Time back is the want — fewer mornings spent re-ranking the same bind by hand.
Same-series cuts already live: the category (one core, every kind of business); the release gate (Freigabe holds the switch); the stack law (above the systems, not instead). Arithmetic stays in code. This piece locks the fit test under those cuts.
Six bands equal
Six public bands sit as equals. Venue operations last on purpose:
- Manufacturing and production
- Warehousing, logistics and distribution
- Technical service and maintenance
- Health and care
- Retail and multi-site
- Venue operations
A named account is not a category. Gastgewerbe is one band among six, never the product story. The fit is the operating pattern, not a sector SKU.
What we are not selling
Not a dashboard. Not a website. Not a chatbot. Not restaurant software. Not “AI for hospitality.” Not a gastro SKU. Not invented hours, euros, or ROI.
We do not write the Kasse. We do not silently auto-roster. We do not predict illness.
If the three hold in your Betrieb, twenty minutes is enough to see whether agents on Spectacle have a job. That is the Erstanalyse — start at Spectacle. If the data cannot answer the question, we say so.
FAQ
What is the three-part pattern? Demand is uncertain. Capacity cannot change in the short run. Labour and material must be committed before demand is known. Those three together decide whether agents on Spectacle have a job.
If the three do not hold, do you still sell Spectacle? No. If those three do not hold together, I do not force the product. Excel, WhatsApp, and Erfahrung may already be enough.
Is this restaurant software or “AI for hospitality”? No. Pattern, not a sector. Six equal bands. Venue operations last on purpose. Not a gastro SKU.
Do agents auto-assign shifts or write the Kasse? No. Agents assemble a named measure. A person releases it. We will not auto-assign shifts without human release. We will not write sealed systems of record. Illness prediction is not the job.
Do you invent hours, euros, or ROI? No. Time back is the want. We do not invent hours, euros, ROI, or accuracy bands.
Where should I start? Spectacle — Erstanalyse. The Insights series is how we write about the operation.