Learn how to build AI agents for procurement Join us in Aarhus May 21

Growing capacity 50%. Not growing overhead to match.

Aker QRILL is adding a new vessel that will grow production capacity by 50%. Rather than scaling headcount alongside volume, they built software to absorb the complexity.

Customer Story

Results

11 applications built — five deployed, six in development
10% shipment error rate targeted for elimination
300–400K NOK in freight errors caught, fully retained

What this means

People do judgment work, not manual work. Teams review exceptions instead of checking every page.

Customers and suppliers feel it. Fewer errors, faster confirmations, same-day responses.

AI capabilities that compound. Each app is a building block, all without adding headcount.

The Company

Aker QRILL is the world's largest krill producer. PE-owned carve-out from Aker BioMarine, 650 employees, 150 customers in 50+ countries. Around 70 people handle all onshore admin and sales.

They're investing 3bn NOK to grow capacity by 50%. A new vessel is joining the fleet. The question wasn't catching more krill. It was running the business behind it without growing overhead to match.

They'd run early AI pilots but nothing had reached production. With Riff, they scoped a focused program: admin and support functions only, real pain points, one app per area. Small team, domain experts in from day one, builders-first mindset. Six apps shipped in three months. 23 more in the sandbox.

“We need to scale volume and complexity without scaling headcount and overhead at the same pace.”

The Solutions

Five example applications, one strategic outcome

Shipping Document Verifier - Order Management

Problem: Every shipment needs documents to match: container IDs, product names, invoice values. The team checked each one by hand. One in ten still had mismatches that triggered customs holds.
Solution: The app scans each document set, extracts key fields, and flags mismatches before anything ships. The team reviews only the flagged exceptions instead of reading every single page.
Results: Error rate on track from 10% toward zero, team freed for complex work, fewer customs delays.

Order Allocation Assistant - Order Management

Problem: Matching orders to inventory meant juggling PDFs, packaging specs, and import rules across scattered systems. A single allocation decision could generate over 150 emails spanning months.
Solution: The app extracts PO requirements, checks live inventory, applies customer preferences, and recommends allocations. Sales gets self-serve stock visibility without emailing the team.
Results: Manual work eliminated from allocation workflow, faster order confirmation, care team freed from questions.

Supplier Invoice Checker - Supply Chain + Procurement

Problem: Over 100 supplier invoices monthly, up to 20 line items each, in multiple currencies. Too complex to check by hand, so it was outsourced to auditors who kept 50% of every error found.
Solution: The app reads each invoice, extracts line items, validates currency conversions via live rates, and verifies the math. Clean invoices pass straight through; exceptions get flagged for review.
Results: 300–400K NOK in errors caught annually, hours per invoice reduced to ~2.5 min, external auditor eliminated, 100% of savings retained.

Supplier Statement Reconciler - Finance

Problem: Suppliers send statements covering 300+ invoices and want payment status. The volume was too high to handle manually, so most statements simply went unanswered — straining supplier relationships and risking late-payment charges.
Solution: The app accepts statements in any format by email or upload, reconciles each invoice against live ERP data, and generates a ready-to-send status report. Low-confidence items get flagged.
Results: Hours down to 2 min per statement, interest charges reduced, stronger supplier relationships.

Market Intelligence Dashboard - Leadership

Problem: Market position, competitive dynamics, and seasonal patterns all drive planning decisions. Leadership tracked this through manual PowerPoint decks where data was always stale by the time it reached a meeting.
Solution: The app pulls official industry reports, integrates internal data, and presents real-time dashboards with auto-generated summaries. Leadership uses a built-in chatbot for follow-up questions.
Results: Used daily for strategic decisions, manual data compilation gone, data accuracy and consistency improved.

By the Numbers

The economics of leverage

Metric	Before	After	Impact
Shipping document verification	~10% error rate	On track toward zero	Fewer delays, happier customers
Order allocation	Manual, multi-step process	Manual work removed, faster response	Improved customer care
Supplier invoice checking	Hours per invoice	2.5 minutes per invoice	95% time reduction
Supplier invoice error recovery	300–400K NOK found, 50% kept by auditor	200K NOK in savings, external auditor eliminated	Full savings retained in-house
Supplier statement reconciliation	Largely unaddressed	2 minutes per statement	Fewer delays, stronger supplier relationships
Market intelligence	Manual decks, always stale	Real-time dashboards	Strategic insights reviewed daily
Time to deploy new app	6–18 months (custom IT)	2–3 weeks	10–20x faster

“We go from business problem to production software in weeks, not months, at a fraction of the cost.”

How It Happened

Three months, 11 apps, one new operating model

Aker QRILL didn't hire more developers or implement generic enterprise software. They mobilized domain experts from Order Management, Finance, Supply Chain, and Operations, then worked intensively with Riff to translate business processes into working applications. Five are in production today, with six more in development.

Mobilize domain experts

Order Management, Finance, Supply Chain, and Operations teams brought the knowledge of how work actually happens.

Translate processes to software

Working with Riff, business rules and exceptions became production-ready applications in weeks, not quarters.

Train internal builders

Internal teams now own and evolve applications going forward, reducing dependency on external vendors.

Deploy with governance

Role-based access, audit trails, and security gates from day one. No shortcuts on compliance.

What's Next

From apps to agents

Deeper into workflow automation

Moving from single-task apps to multi-step agents that handle end-to-end processes with human oversight at decision points.

More advanced human augmentation

Giving domain experts tools that amplify their judgment rather than replace it, across more complex operational scenarios.

The Broader Lesson

Scaling through domain expertise

The old model: Business needs identified → Added to IT backlog → Custom development scoped → 18-month delivery cycle → Technology no longer matches reality
The new model: Business needs identified → Domain expert + AI builder → Working application → 2–3 week delivery cycle → Continuous iteration based on real use

For a company facing 50% capacity growth with limited headcount expansion, this difference is existential. It's the difference between scaling profitably and scaling into a cost crisis. This isn't just automation. It's strategic leverage.

Ready to turn your expertise into leverage?

About

Aker QRILL Company is an Antarctic krill-harvesting company dedicated to improving health across species. They develop krill-based ingredients for nutraceutical, aquaculture, and animal feed applications, backed by rigorous science and research. This case study represents their journey to scale operations through AI-assisted business process transformation.