What you can measure our work by

Starting point

Call-offs change faster than the monthly planning cycle

• Late reaction: Call-off cuts, requests and lead-time jumps assessed only in the monthly cycle
• Fragmented data: Data spread across separate systems, analysis takes heavy manual effort

Solution

Orchestrated AI agents detect, assess and recommend, people decide

• Detect: A signal agent checks new call-offs and customer emails on its own
• Assess: Agents quantify the impact in euros and units, every figure calculated from the planning data
• Decide: The best mix of actions as a proposal, approved by the planner with a log

• AI agents
• Process automation
• Forecasts and scenarios
• Data integration

Result

Counter weeks earlier, load production evenly, avoid emergency measures

• Utilization: Peaks smoothed, free capacity reallocated instead of extra shifts
• Costs: Fewer special freights, excess stock reduced early instead of written off
• Delivery performance: On-time delivery secured, customer requests confirmed only with feasible volumes

Starting point

Ramp-up under pressure, causes of loss unknown

• Delivery at risk: Output below customer demand, backlogs growing
• No reliable picture: Downtime and scrap spread across paper, spreadsheets and separate systems

Solution

Losses made visible per station and eliminated systematically with lean routines

• Data foundation: Existing systems connected, a daily cockpit of overall equipment effectiveness (OEE) and losses per station
• Analytics: Biggest OEE loss drivers identified in each work cycle, actions derived from the data
• Steering: Implemented with the line team, the impact of every action measured

• Data integration
• KPI systems and dashboards
• Impact measurement
• Change management

Result

Ramp-up stabilized, more output from the same line

• Technical availability: +18 percentage points (from 70% to 88%), the biggest OEE lever
• Scrap: Cut from 12% to 7%
• Output: +85% a little over a year after start of production

Starting point

Ramp-up in crisis, start of production at risk

• Risks spotted too late: Schedule and start of production at risk at several sites
• Line overloaded: Problems solved alongside daily business, resources spread without focus

Solution

Taskforce started at short notice, bottlenecks worked off in fixed cycles

• Start: Core team free of line duties, in tandem with the line managers
• Steering: Short work cycles, a weekly steering committee, actions measured against KPIs
• Handover: De-escalation criteria defined in advance, open topics handed to the line

• Taskforce lead
• Situation diagnosis and de-escalation
• Risk and decision management
• Training and enablement

Result

Start of production secured, the line carries on by itself

• Good-part rate: Raised from about 30% to above 80%
• Bottleneck cycle time: Cut by more than 40%
• Embedded in the line: Routines and KPIs taken over by the line team

Starting point

Many programs, many sites, no shared picture

• Scattered data: Sites report their status in separate spreadsheets, each in its own format
• Manual consolidation: Status picture merged by hand every week, deviations spotted late

Solution

One reporting logic, one data model, one executive cockpit

• Capture: A standard template with upload and automatic checks per site
• Data model: Programs, sites and reporting dates in one model, history through stored snapshots
• Cockpit: KPIs per program and site, deviations at a glance

• Data integration
• KPI systems and dashboards
• Data quality and governance
• Program planning and steering

Result

One picture for all programs, deviations visible early

• Transparency: 12 KPIs in one cockpit for the entire site network
• Rollout: Started with about 20 programs, more sites to follow
• Effort: 20 hours of manual consolidation per week eliminated

Starting point

More than 80 companies order individually instead of as a group

• Buying power unused: Each company negotiates on its own, the same items at different prices
• Manual ordering: Phone, fax and email, most documents checked by hand

Solution

Platform selected, integrated into the IT landscape and embedded in the group companies

• Selection: Criteria catalog and an interface test with real data before the contract was signed
• Integration: ERP interfaces and master data aligned with group IT
• Enablement: Rollout in waves, key users as multipliers, training and supplier onboarding

• Project management office (PMO)
• Program planning and steering
• Data integration
• Change management
• Training and enablement

Result

Pilot started on plan, supplier base for regular demand integrated in the system

• System decision: Suitable platform chosen together with all stakeholders, contract negotiated with experts
• Pilot on plan: First companies live on schedule, first orders placed successfully through the system
• Broad involvement: Group functions, business units and operating units involved from the start, with high acceptance

Starting point

Same items, different prices, no transparency

• Scattered data: Invoices only as PDF, stored in separate systems at each company
• Prices rose unnoticed: Item-level increases went unchallenged for months

Solution

AI reads invoices down to item level and compares across all companies

• Data foundation: PDF invoices of all companies made usable, without touching existing systems
• AI extraction: About 150,000 invoices from 160 suppliers captured in structured form and checked
• Analysis: Prices and terms compared per supplier, negotiation levers prioritized

• AI document processing
• Data integration
• Data quality and governance
• KPI systems and dashboards

Result

Price transparency across all companies, without extra work for the business

• Savings per supplier: Up to €50k per year
• Consistent terms: 1 to 2% savings from consistent discounts
• Lever per supplier: Prioritized by volume and deviation, ready to negotiate

Starting point

A flood of documents, manual coding, correction loops with the tax advisors

• Manual: Every document coded one by one, payments matched by hand
• Error-prone: Special cases such as reverse charge, entertainment expenses or foreign currency

Solution

The AI agent books daily, fixed rules safeguard the tax logic

• Automation: Read documents, code them, match them with bank transactions and book them
• Booking rules: Tax booking rules fixed in the system, uncertain cases go to review
• Learning loop: Every correction by the tax advisors becomes a rule and a test case for quality assurance

• AI agents
• AI document processing
• Process automation
• Data quality and governance

Result

Accounting runs automatically every day, the tax advisors review instead of booking

• Match: About 91% of bookings identical to the tax advisors’ final version
• Volume: About 4,500 logged AI decisions on more than 850 documents
• Multi-entity: Each company with its own rules, data strictly separated