Including the uncomfortable ones about failure modes, accountability and whether any of this is proven yet. Short answers here; the detail is in the docs.
Live example: Grade change PM4 · 135 gsm kraftliner → 110 gsm testliner, no break, ≤14 min off-spec
The questions mill managers and process engineers actually ask in the first meeting.
It is a supervisory autonomy layer for paper machines. Nine agents and an orchestrator perceive the web, plan a run, rehearse it on an as-run digital twin, act inside limits the mill sets, and record every decision.
No. Pulpum reads from and writes through the control systems you already have. Base-layer regulatory control, scanner hardware and safety systems remain exactly where they are.
The machine keeps running on existing control. Any loss of inputs, models, network or policy means the agents stop writing and hand back to the DCS. Degrading to today's way of running is always an acceptable outcome.
Only at the autonomy level a site has enabled, only on tags in the allow-list, and only within declared magnitude and rate limits. Speed ramps and grade releases always require human approval.
Three to four weeks from rack to shadow mode, then roughly 90 days to a signed baseline reconciliation. Advisory usually starts around week seven.
Machine tier is $16,000 per machine per month. Mill tier is $95,000 per month. Enterprise agreements run from $700,000 to $9,000,000 in annual contract value depending on fleet size and outcome terms.
Not yet. Pulpum is pre-launch. Every performance figure on this site is a design target and is marked [ASPIRATIONAL]. We would rather lose a deal than misrepresent that.
Process data, video and the audit log stay on mill-owned infrastructure by default. Only what a site explicitly enables leaves, and model training on your data requires written agreement.
A Jetson-class mill-edge node per machine, synchronised line-scan cameras where existing inspection cannot be reused, and network access to the QCS, DCS, historian and MES.
The questions mill managers and process engineers actually ask in the first meeting.
Models are pinned per machine. A candidate must pass historical replay, evaluation gates and shadow mode before promotion, and rollback is a single command.
The mill retains operational accountability, which is why every write is bounded, logged and reversible. The audit log records the agent, the model version, the inputs, the limits checked and any human approval.
Kraftliner, testliner, fluting, folding boxboard, tissue and specialty grades today. Coated board and specialty release grades are on the roadmap.
Yes, and most mills should. Advisory mode is a complete product on its own — recommendations, reasoning traces and genealogy without any write access.
The break-precursor models aim for around 90 seconds of warning. On an actual break the agents stop writing, record the full signal window and assist with the restart sequence rather than driving it.
Acceptance rate on advisory recommendations is the metric we watch hardest, and every rejection is kept as training signal. Crews tend to trust the audit log before they trust the model.
Reel genealogy maps onto ISO, TAPPI and FEFCO measurement methods so conformance can be evidenced rather than asserted.
The mill edge is on-premises by definition. Fully air-gapped deployments, with training and simulation on mill-owned hardware, are supported for enterprise agreements.
Send twelve weeks of QCS and DCS history from one machine. We will tell you within a week what we can see and whether a pilot is worth running.
When an answer refers to "a run", this is the one it means.
Mill Orchestrator — Pulled the 110 gsm testliner spec, customer tolerances and the standing energy budget from mill MES; locked the target envelope for the run.
Mill Orchestrator — Simulated 48 candidate transition recipes on the as-run paper-machine twin — forming, press, dryer and calender — and ranked them on off-spec tonnes, break risk and steam.
Pulp-and-Stock — Stepped refiner specific edge load 1.9 → 1.4 Ws/m and pushed freeness toward 412 CSF while consistency held at 3.4%.
Wetend-and-Chemistry — Retention aid trimmed to 214 g/t and sizing to 1.1 kg/t against live charge and turbidity; first-pass retention recovered to 78% inside 90 seconds.
Form-and-Press — Re-cut the slice profile across 78 actuators and set jet-to-wire to 0.994 to hold formation index through the basis-weight ramp.
Form-and-Press — Nip load reduced 620 → 540 kN/m to protect the lighter web; post-press dryness landed at 47.1%.
Dry-and-Coat — Re-phased the steam schedule across 6 dryer groups and rebalanced the hood; reel moisture converged to 7.4% ±0.19 2σ at 6.1% less steam than the standing recipe.
Defect-and-Inspect — 18 line-scan cameras streaming; two edge-crack precursors detected at the drive side and cleared by a 40 kN/m nip trim before either propagated.
Mill Orchestrator — Speed ramp 1,180 → 1,245 m/min exceeded the site autonomy threshold. Held for the machine tender; approved by J. Okonkwo at 04:57:12.
Quality-and-Conformance — Reel R-24188 released: full genealogy written (furnish, chemistry, CD profiles, defect map, operator actions) and checked against grade spec.
Every tool invocation, argument and result is written to an immutable, human-readable log — and every reasoning step is expandable. Nothing about a run is hidden from the mill.
Plan → thought → action → observation, exactly as the orchestrator recorded it.
Move PM4 from 135 gsm kraftliner to 110 gsm testliner with zero breaks and under 14 minutes of off-spec production.
Simulate the transition on the twin first. Lead with stock and chemistry so the furnish is right before the basis weight ramps, then trim forming and press, then re-phase drying. Watch the web continuously; hold the speed ramp for a human.
The standing recipe drops basis weight and speed together. On the twin that produces a 3-minute moisture excursion because the dryer section unloads faster than the steam schedule can follow. Sequencing the steam re-phase ahead of the basis-weight ramp removes it.
twin.simulate(candidates=48) → recipe #31 re-phases steam 40 s before the basis-weight ramp and holds nip load until post-press dryness confirms.
Recipe #31 estimates 6.4 t off-spec against 9.8 t for the standing recipe, with break risk 0.07. Two candidates scored lower on steam but raised break risk above the 0.15 site limit and were discarded.
Execute recipe #31 under autonomy level L3: nine setpoint writes permitted, speed ramp routed to the machine tender.
Edge-crack precursor at 1,412 mm drive side at 04:56:41. Nip trim of −40 kN/m in press-2 cleared it in 16 s without a break.
Run complete in 11 m 42 s. 6.2 t off-spec, zero breaks, 6.1% less dryer steam than the standing recipe. Reel R-24188 released with full genealogy.
A mill does not go from manual to unattended in one step. Pulpum makes the level explicit, auditable and reversible at any time.
| Level | What the agent does | What the human does | Typical time to reach |
|---|---|---|---|
| L1 · Advisory | Recommends setpoints and explains why | Enters every change manually | Week 1 |
| L2 · Supervised | Proposes a write; it executes on approval | Approves each write in the HMI | Week 3–6 |
| L3 · Bounded | Writes inside tag, rate and magnitude limits | Approves ramps and grade releases | Month 2–4 |
| L4 · Unattended | Runs the envelope without prompting | Sets the envelope; reviews the shift record | Month 6+ [ASPIRATIONAL] |
Pulpum writes to production equipment. Every capability is scoped, every write is policy-checked, and every action is written to an append-only audit log the mill owns.
| Standard | Scope | Status |
|---|---|---|
| SOC 2 Type II | Cloud control plane | RUNNING In progress [ASPIRATIONAL] |
| ISO 27001 | Company-wide ISMS | QUEUED Planned [ASPIRATIONAL] |
| IEC 62443 | Mill-edge OT security | RUNNING Design-aligned |
| GDPR | Operator data | SUCCEEDED Compliant |
| ISO 9001 / FSC | Quality + chain of custody records | SUCCEEDED Supported |
The table people ask for after the first ten minutes.
| Failure | Detection | Agent behaviour | Machine impact |
|---|---|---|---|
| Sensor loss | Signal staleness check | Stop writing affected loop | Runs on existing control |
| Camera desync | Frame timestamp drift | Degrade to reduced camera set | Reduced defect coverage |
| Model confidence collapse | Confidence below threshold | Fall back to advisory | No autonomous writes |
| Edge node failure | Heartbeat loss | DCS retains control | None |
| Network loss | Control-plane unreachable | Local buffer, continue at current level | None |
| Policy violation | Pre-write limit check | Refuse and escalate | None |
| Twin divergence | Prediction error vs reel data | Widen risk margins, request review | More conservative moves |
| Human override | Operator action | Yield immediately, log the reason | Operator in control |
Land on one workflow with a measurable baseline. Expand by machine, module and autonomy level. Outcome components available on broke, breaks and energy.
Wet-end, forming and drying control or web inspection for a single machine. For mills validating ROI on the wedge.
The whole loop: all nine agents, the orchestrator and the as-run paper-machine twin across every machine on site.
Multi-site groups standardising on Pulpum, with custom grade, furnish and tolerance models.
Pulpum reads and writes through the systems already on the floor. No rip-and-replace, no parallel historian, no new HMI to learn.
Valmet IQ, ABB 800xA QCS, Honeywell Experion MX
Profiles, scans, lab results
ABB 800xA, Valmet DNA, Honeywell Experion, Siemens PCS 7
Setpoint reads and guarded writes
WIS/WMS line-scan, IR and transmission cameras
Frames, defect maps, break replays
SAP PP/QM, ABB cpmPlus, custom historians
Orders, grades, reel genealogy
OSIsoft PI, Aspen IP.21, InfluxDB
Time-series backfill and replay
NVIDIA Isaac, winder and wrapper PLCs
Reel, roll and clamp-truck motion
Azure AD, Okta, on-prem LDAP
SSO, RBAC, named approvers
NVIDIA Jetson Orin, IGX, on-prem GPU
Sub-100 ms inference at the machine
One policy model, one audit trail, one benchmark across every machine in every mill — with the grade and furnish models kept private to each site.
Autonomy earns trust one shift at a time. These are design-partner quotes from pilot deployments [ASPIRATIONAL].
"The first thing that convinced the crew wasn't the control — it was the log. You can scroll back and see exactly why it dropped the nip. Nobody argues with a timestamp."
Machine tender · PM4 · Nordkraft Mills
"We had two people who could do a clean 135-to-110 transition. One retired in March. The twin now does the sequencing and the second one supervises it."
Production manager · Aurora Board
"Break prediction was the wedge. Ninety seconds of warning is the difference between a nip trim and four hours of threading."
Process engineer · Ternvik Paper
Ask it directly. If the answer is that Pulpum is not right for your machine, we will say so.
↑↓ navigate↵ openesc close