ContactYour experts already answered it. Continuum finds the answer and hands it over — without interrupting them.
Marcus Chen
Shared Service Centre · SSC Kraków
DN50 regulator flange — torque spec after gasket change?
Hi Anna — quick one before I sign this off. What's the recommended torque for the DN50 regulator flange after the 2023 gasket change? The SSC runbook still shows the old figure and I don't want to guess.
Thanks,
Marcus
Answer — sent 14 March 2024
After the gasket change, torque to 85 Nm in two stages — 40 Nm cross-pattern first, then 85 Nm. Do not use the pre-2023 figure of 110 Nm; it over-compresses the new gasket material.
Anna Reber
Process Owner — Regulators & Valves · 22 years
Without Continuum — the question waits ~3 days for the expert to be free to reply.
With Continuum — answered in seconds, from an answer Anna already wrote — she's never interrupted.
18 months
before anyone notices how often the questions come back
2–3 people
how many experts a plant of 2,000 really depends on
0
records of what those interruptions cost
The problem
~6 hours
of senior expert time a month, spent re-answering questions they've already answered — on a conservative 40 repeat questions.
The work moved. The questions didn't. They still route back to two or three people — every one a senior person's afternoon, on no cost line anywhere.
What Continuum does
Continuum finds it, shows who wrote it and when, and hands it to whoever asked.
Watch the whole loop — match, refuse, capture, match again:
Marcus Chen
Shared Service Centre · SSC Kraków
DN50 flange torque after gasket change?
Hi Anna — quick one before I sign this off. What's the recommended torque for the DN50 regulator flange after the 2023 gasket change? The SSC runbook still shows the old figure and I don't want to guess.
Thanks,
Marcus
Answer — sent 14 March 2024
After the gasket change, torque to 85 Nm in two stages — 40 Nm cross-pattern first, then 85 Nm. Do not use the pre-2023 figure of 110 Nm; it over-compresses the new gasket material.
Anna Reber
Process Owner — Regulators & Valves · 22 years
A question comes in. Continuum finds the answer Anna already wrote.
Why this clears review
41% — below threshold
Never generates. Below the confidence threshold it refuses and routes to the expert.
Confirmed with [NAME] at [SITE].
Confirm current lot with [SUPPLIER].
Strips identities. People, sites and suppliers are removed before anything is stored.
Sent Items
Zero extra steps. The expert just replies to email. There is nothing to maintain.
Every answer carries its author, role and the date it was written.
The pattern underneath
Seventeen questions about one valve isn't a knowledge gap. It's a missing document, a training gap, or a supplier issue. Continuum shows you the pattern.
Questions we get asked
No. Continuum only retrieves text a named expert actually wrote. It never generates, paraphrases, or summarises. If it has no close match, it says so and routes the question to the expert.
Names, sites, suppliers and customers are stripped at the point of capture, before anything is stored. The stored answer keeps the technical content and the author attribution, nothing else.
In the EU and Switzerland. Nothing leaves that boundary.
That's a conversation to have early, not late. In Germany a tool that observes an expert's mailbox falls under co-determination, so the Betriebsrat has to be involved. We start engagements in Switzerland for that reason, and we'd rather raise it with you than have you discover it.
No. The diagnostic is a consulting engagement, not software. We tag 60 days of one expert's inbox and hand you a written finding. Nothing is installed and nothing connects to your systems.
Every surfaced answer carries its author, role and the date it was written. The person asking can see it's from 2023 and judge accordingly — which is more than they get from a chatbot.
Copilot generates. That's the point of it, and it's also why it can't be used where a wrong answer is dangerous. Continuum's constraint is the opposite: it can only return something a human expert already wrote.
The diagnostic is priced per engagement. The product is priced per protected expert per year, not per seat — the people asking are many, the experts are few.
Where to start
Before you buy anything, we measure it. We tag 60 days of one expert's inbox — what was asked, how often, and whether last time's answer would have worked. Anonymised, nothing installed. If the questions are mostly one-offs, we'll tell you, and you shouldn't buy the product.