CUSP — Operator Demo
Decision engine · cusp.so
Metal / fabrication · 10 Apr 2026 · CUSP early build

What you should do this week

Primary action
Protect margin now — increase pricing immediately, cut variable cost, lock energy before the window closes
High confidence — PMI + new orders both below contraction thresholds simultaneously
PMI mfg
44.9
New orders
41.0
Copper MoM
+8%
Gas WoW
−12%
Assuming ~30–50% material cost exposure — typical for metal fabrication. If your exposure is different, some numbers shift.
Three signals stacking. Costs rising, demand falling, energy window open. Not independent — together they point one direction. Timing is the decision:
Pricing — this week Energy lock — 2 weeks Labour — ongoing
Act now · 0–7 days
Raise prices — lowest-margin products first
2–4% increase starting with your thinnest-margin lines. Flag existing contracts for next renewal. Don't wait for permission.
Protects ~2% EBITDA (assuming materials ~40–50% of cost base) If ignored → margin erosion in 4–6 weeks
Plan · 1–2 weeks
Cut overtime first, then agency shifts
Stop overtime immediately. Review WIP and push non-urgent production.
If orders don't recover in 2–3 weeks, begin reducing agency headcount. Don't cut permanent staff yet — the signal isn't that severe.
Avoids £30–80K working capital lockup If ignored → overstock builds as orders thin
Plan · this week
Lock energy forward contracts now
Call your broker today. Lock 3–6 month forward at current rates. This window may close within days if gas rebounds.
Save 5–10% on energy bill vs spot If ignored → window closes, no second chance
Trade-off: locks in savings but removes upside if prices fall further. At −12% WoW, the lock is worth it.
Before Jan deadline
Made Smarter automation grant — worth doing
Welding robots, press automation, CNC upgrades all qualify. Application is ~2 hours.
Up to £20K non-repayable If ignored → next window is 12 months away
Do not do this
Do not build inventory — even though energy is cheaper right now
This is where most operators get caught out. Cheaper inputs is not the same as real demand.
New orders at 41. You'll hold stock you can't shift while cash gets tight.
Do not expand capacity or launch new product lines
Chasing volume you won't get is how margin disappears twice.
PMI 44.9, fifth consecutive month below 50. This is not a blip.
What copper % would actually make you raise prices?
Drag to your real threshold — not the safe answer
+2% +20%
+8%
At +8%: direct BOM hit. We call this act-now. Is that where you'd actually move?
"What's the one recommendation here you disagree with most?"
Ask it. Then stop talking.
If they seem convinced too easily: "This might be wrong for your business — that's what I'm trying to find out."
1 / 9
Before we finish
Based on what you've told me, here's how I'd adjust the engine for your business.
Read this back to them. Let them correct it. This is the moment it becomes theirs — not your tool, their system.
Does that feel right?
Before we finish
Based on what you've told me, here's how I'd adjust the engine for your business.
Read this back to them. Let them correct it. This is the moment it becomes theirs — not your tool, their system.
Does that feel right? Correct anything.
Analysing responses...
Interview summary
Calibration output
Operator profile
Confidence mismatch
Calibrated thresholds
Where the engine needs fixing
Language upgrades — their words
Kill question signal
Loop close reaction
What to change before next interview