Turn meetings into follow-through
A decision needs an owner, a date and a next action. Then somebody has to stay with it until it is actually done.
Snevagten A/S · Executive Operations Associate · selected fit
Keep people, partners and systems moving. Use AI where it actually helps.
My read of the job: help Marcus turn meetings, partner issues and projects into clear follow-up, while learning Snevagten's AI tools well enough to test them against the real operation.
My recent title is Technical Buyer. The work underneath it was broader: chasing commitments, resolving exceptions, checking ERP data, keeping different functions aligned and making sure the next action had an owner. That is the experience I would transfer here.
30+international suppliers coordinated
2,000+ERP / master-data records
100+ itemsaffected by a discrepancy caught before quotation
5 / 40+production lines / variants supported
What I understand the job to be
Turn decisions into movement.
Snevagten's public material is unusually clear about the gap: meetings and ideas are not enough. Somebody has to create structure, keep the right people involved and make sure the work is actually carried through.
Keep partner issues moving
Know what was promised, what changed, what is missing and who needs to act before a small exception becomes a bigger operating problem.
Make data useful for decisions
Check the source, reconcile conflicting information and turn the issue into something a manager can understand and act on quickly.
Test the tools against real work
Use normal cases, missing information and awkward exceptions to see whether an AI-assisted workflow actually helps the operation.
WHERE THE ROLE SITSBetween the manager, the partner network and the teams building the tools.
Marcus
Day-to-day partner operation and the work this role is there to support.
Christoffer + Heidi
Direction, onboarding and the wider operating rhythm around the role.
Project + IT
Build and improve the systems, automations and AI tools used in the business.
120+ partners
The external reality where the process, information and follow-up have to work.
WHY MY BACKGROUND FITS
Procurement was the setting. The useful part transfers much further.
I was often working in the layer between the transaction and the operation: supplier commitments, data, exceptions, documentation, internal priorities and the next action. This role puts that coordination and follow-through at the centre, then adds a modern workflow and AI layer that I have deliberately been building through my own projects.
How I would support Marcus
Give him fewer things to chase.
Marcus already carries a large partner operation. I would not add another reporting layer. I would keep the important facts and next actions visible so he can spend his time where his judgement and relationships matter most.
01TODAY
What needs action now?
Open decisions, partner exceptions, missing information, urgent promises and anything that needs escalation.
02THIS WEEK
What must not drift?
Projects, long-term partner work, data corrections, agreed follow-up and the next milestone that can disappear behind daily noise.
03IMPROVE
What keeps repeating?
A manual check, unclear handoff, repeated question or weak data point that is worth fixing instead of working around again.
WHEN I BRING AN ISSUE TO A MANAGER
Keep it short enough to use.
01What happened02Impact03Facts checked04Recommendation05Owner + next action
Real example: when I disagreed with a manager on steel timing, I brought the market and production evidence, gave my recommendation, respected his final decision and kept monitoring the risk. I am comfortable challenging with evidence without confusing support with authority.
How I think about AI at work
I do not start with AI. I start with friction.
First understand the task, the trusted source and who owns the decision. Then decide what can be automated, what still needs review and what should stop when the information is not good enough.
Understand the task
What is the user trying to get done? Which information is trusted? What should the tool never guess?
Test actual cases
Routine input, missing information, conflicting information and unusual but valid situations.
Check it against reality
Compare the output with the source and the operating context. High-impact actions stay with a person.
Record the failure and retest
Show the technical team what failed, what should have happened and whether the revised behaviour now works.
Where I fit with IT
I would not design the agent architecture. I would bring the business problem, approved sources, important exceptions and realistic test cases, then give the developers clear feedback on what worked and what did not.
REAL STOP-RULE EXAMPLE
Missing documentation meant stop, not "probably fine".
Material arrived without the required CoC/COO. I kept Production informed, followed the supplier and checked alternatives. The documentation later showed an origin the customer did not allow, so the material stayed blocked and we moved to an approved alternative. That same judgement matters when an automated workflow is missing a trusted source or reaches a high-impact exception.
I have already been practising this way of working
Three projects, three different contexts, the same discipline.
01Independent workflow project
Clementi AI Workflow
I built it around a simple rule: source of truth before automation. It uses workflow mapping, clear owners, review points and reusable handovers rather than treating AI as the starting point.
View project ↗02Human-reviewed process pilot
FørsteMatch
A process pilot built around task clarity, support needs, privacy, ownership and human review. No automated selection decision. The point was to make the process clearer before adding technology.
03Live implementation
Marzieh Nail Atelier
A real business implementation where scattered owner input became one clear system for services, prices, policies, booking, bilingual information, QA, release and handover.
View live site ↗Working toolkitDynamics AX/D365 · Excel + Power Query · Power BI/DAX · SQL fundamentals · Python basics · Jira · GitHub / Cloudflare
Evidence from real operations
The technology is the new layer. The operating judgement is not.
01PARTNER / SUPPLIER FOLLOW-UP
Recovered a capable supplier when volume became the constraint.
I controlled commitments, prioritised urgent demand, coordinated alternatives and kept Production and other stakeholders informed. I did not wait for the bottleneck to fix itself.
Supplier commitments · prioritisation · recovery actions · escalation02DATA QUALITY
Caught a commercial-data problem before a major quotation went out.
I found inconsistent cost and sales-price information affecting 100+ items, traced the source and coordinated the correction. A system value was something to verify, not something to trust automatically.
100+ affected items · source tracing · correction before release03ERP + CHANGE
Kept changing technical input connected to the people who had to act on it.
Daily AX/D365 work covered 2,000+ ERP and master-data records and 200+ complex structures. A change could affect material, suppliers, routings, quality records and production readiness, so impact and ownership had to stay visible.
AX/D365 · Planning · Engineering · Quality · Production · suppliers04SERVICE OPERATIONS
Owned international service files while the operation was still moving.
At AOT Inbound / ATS Pacific I coordinated French and European customer files, supplier confirmations, changes, complaints and urgent recovery. The habit was simple: keep people informed and keep the next action clear.
Customers · suppliers · changes · complaints · urgent recoveryFirst month
Learn the operation, take work off the manager, improve one thing that matters.
I would rather make one useful workflow reliable than arrive with ten AI ideas before I understand the job.
Learn the rhythm before changing it
Understand Marcus's recurring meetings, partner flow, current trackers, decision points, approved data sources, Claude Workspace and the internal AI tools already in use.
Take ownership of a real follow-up queue
Start carrying actions, partner exceptions or project follow-up so fewer open points depend on Marcus remembering or chasing them himself.
Improve one useful workflow with the team
Choose one repeated pain point, test the existing or proposed flow with realistic cases, document what breaks and help retest a better version.
The short version
Operations first. Follow-through every day. AI where it makes the work better.
I bring real supplier and service coordination, ERP and data discipline, cross-functional execution and a practical interest in using modern tools without giving away human accountability.
ExecutionPartner follow-upProjectsData qualityAI testingProcess improvement
Executive Operations + AI
Romaric ClementiContact
Operations-trained, systems-minded, comfortable with people and modern tools.
Greater Copenhagen · EU citizen · French native · English fluent · Danish operational (PD2/B1+).
Scope note: Company and role context comes from Snevagten's public vacancy, official pages and public leadership posts available in September 2026. My work examples are kept separate from that public context.