Should my team use this?

A short read for managers deciding whether to invest team time and licenses.

The short version

AI tools are already inside Qvantum's M365 environment. Every employee can use Copilot today; Claude is available on a smaller, paid-license basis for harder work. The cost is real but bounded. The upside depends on whether your team uses it for the right tasks — not all work benefits.

What this is — and isn't

Useful for:

Drafting and editing internal documents.
Summarising long content (meeting notes, specs, customer threads).
Asking questions about pasted content.
First-draft code, first-draft analysis, first-draft anything where the human edits the output.
Reasoning over structured inputs (XML, JSON, tables).

Not useful for:

Anything where being wrong has high cost and no human checks the output.
Anything requiring real-time internal data the tool hasn't been given (it won't read Qnet, PM Portal, or your inbox unless explicitly connected).
Replacing domain expertise. The output quality depends entirely on what you give it.

What it costs

Copilot: [FILL: Confirm whether Copilot is included in current M365 licenses for Qvantum, or if it's a separate per-seat add-on. Per-seat cost from Petter.]

Claude: [FILL: Per-seat license cost. Confirm with Alenka or Johan. Note current license count and who has them.]

No-cost option: Every Qvantum employee can use Copilot today at no incremental cost (assuming bundling above).

What's already working here

Real examples from colleagues:

R&D — XML tool built with Claude. [FILL: one-sentence outcome, e.g. "Saved ~3 hours per week on supplier-data conversion."] See the page.
Operations — NPI step-by-step. Coming soon (post-Federico meeting, week of 2026-05-22).
IT — Internal search across Qnet / PM Portal. Under investigation.

See all use cases →

If you want your team to try this

A small, structured approach beats a free-for-all:

Pick one task. Something your team does often, that has clear inputs and outputs, and where being slightly wrong is recoverable.
One person prototypes. They try Copilot or request a Claude license; spend a few hours figuring out a prompt that works.
Share it. They submit it as a use case. The whole team can then use the same approach.
Measure. After 2–4 weeks, ask: did this save time, change quality, or neither? Adjust.

For larger ambitions — say, "I want my whole team using AI for X by end of quarter" — we recommend a workshop / "Claude coaching" session. Contact Jacob to set one up.

Risks

Confidentiality. [FILL: short statement about Qvantum's M365 data boundary for Copilot and the equivalent for Claude. Confirm wording with Petter/IT.]
Wrong answers, confidently delivered. AI tools sometimes invent facts. Anything that goes to a customer, regulator, or production system needs human review.
Over-reliance. New employees who use AI from day one may not develop the judgment to spot when it's wrong. Pair AI use with mentoring on what the output should look like.
External sharing. Treat output as internal Qvantum work. Don't paste it into public AI tools or share externally without checking.

Want to talk this through?

The fastest way to figure out if this fits your team is a 30-minute conversation. I can show you what's already working, what isn't, and what a sensible first project looks like for your area.

Book a chat with Jacob Request Claude licenses for my team

Maintained by Jacob Skogström and Alenka [FILL: last name] · Last updated 2026-05-15