Selected work

Outcomes, not case studies

Anonymised engagements from Nordic enterprise AI deployments. Real constraints, real infrastructure, measurable results.

01

Manufacturing · Denmark

600K documents, zero cloud dependency

Problem

A 3,500-person Danish manufacturer had document-heavy compliance processes consuming 40% of engineer time on manual retrieval. Critical knowledge was scattered across legacy systems, SharePoint, and shared drives.

Approach

Deployed a private RAG pipeline over 600,000+ internal documents on on-premises infrastructure. Full-text semantic search with no external API calls. Integrated with existing authentication and access controls.

Outcome

Retrieval time went from hours to seconds. Engineers now query the knowledge base in plain language and get sourced answers with document references.

Key metric

40% reduction in time spent on document retrieval

02

Energy · Nordic

AI that works when the network doesn't

Problem

Offshore teams with intermittent satellite connectivity needed AI assistance for maintenance scheduling and incident reporting, but cloud-dependent APIs were unreliable and created compliance exposure.

Approach

Edge-deployed LLM on local hardware at the asset level. Works fully offline with complete audit trail for regulatory compliance. Synchronises when connectivity is restored.

Outcome

AI-assisted maintenance scheduling and incident reporting available around the clock, regardless of connectivity. Full regulatory audit trail from day one.

Key metric

24/7 AI availability — zero cloud dependency

03

Enterprise · Nordics

From 3% Copilot adoption to targeted AI workflows

Problem

An enterprise running M365 Copilot at scale had 3% active user adoption. Generic responses, no connection to internal knowledge, and staff who did not trust the output.

Approach

AI strategy audit identified three high-value use cases where AI could deliver measurable impact. Built targeted agents connected to internal data sources instead of relying on generic Copilot capabilities.

Outcome

Shifted from broad Copilot rollout to targeted AI deployment in specific workflows. Higher adoption in fewer, higher-value scenarios.

Key metric

Targeted workflow adoption vs. failed broad rollout

Facing a similar challenge?

Every engagement starts with a conversation — no obligation, no sales pitch.

Book a Sparring Session