Speaker Notes — Gräper Invoice Approval Demo

Slide 1: Title

Set the tone: this is a tailored solution, not a generic product demo
"We've studied your current process and your requirements document — what follows is built specifically for Gräper"

Walk through the flow left-to-right quickly — this is the map, not the detail
Emphasize: "The blue box in the middle — the Approval Engine — is the core innovation. Everything before it feeds data in, everything after it acts on the decision."
"We'll go through each stage, but the key insight is: most of this is automated. Humans only get involved when they need to."

"You currently process ~30,000 invoices a year, mostly digital"
"Right now email invoices get handled manually from a central mailbox. We replace that with automatic pickup."
"We also connect to d.velop Cloud to pull documents you already have in your DMS"
"All three channels converge into the same processing pipeline — no manual sorting needed"

"Our AI reads the invoice and extracts everything: supplier, amounts, line items, PO references, payment terms, Skonto deadlines"
"This includes a cost center suggestion with a confidence score — the AI learns from your booking history"
"15+ compliance checks run automatically at this stage — before any human sees the invoice"
Gräper-specific: "Your current system requires FIBU to manually enter all data into LN — supplier, invoice number, amounts, tax codes, GL accounts, up to 6 sub-accounts. We automate all of that."

"Three categories of automated checks happen before any routing"
Formal compliance: "§14 UStG validation — your current process flags this as error-prone and time-consuming. We automate it."
Fraud detection: "IBAN changes compared against master data, duplicate detection, invoice splitting patterns — things that are almost impossible to catch manually at scale"
Supplier verification: "New suppliers get flagged automatically. Known suppliers are validated against your master data."

"This is where the biggest efficiency gain is. 80% of your invoices have POs."
"We match invoices to purchase orders and goods receipts automatically, with configurable tolerances for quantity and price"
"A common scenario you deal with: invoices arriving before the goods receipt. We park those automatically and match them as soon as the GR comes in — no manual tracking needed."
"Your current d.velop system doesn't do automated matching — this replaces that entirely."
Handle the question: "What about tolerances?" → "Fully configurable per supplier or category. Small freight surcharges under a threshold can be auto-approved, just like your FIBU handles the under-10-EUR cases today."

"The Approval Engine is an AI agent that gets the full picture before making a decision"
"On one side: your financial data from Infor LN — POs, goods receipts, budgets, cost centers, supplier master data. We build connectors to pull this in."
"On the other side: everything Orcha has already processed — the extracted invoice, matching results, validation flags, confidence scores"
"The agent combines both to make routing decisions — or to auto-approve when everything checks out"
Key selling point: "We connect to wherever your data lives. If it's in Infor LN, we connect to Infor LN. If some data lives in Excel files or another system, we build a connector for that too."

"The left side is what the agent knows. The right side is what you tell it to do."
"Context data is assembled automatically — the approver doesn't need to go searching through files or systems to make a decision"
"Rules are your business logic: who approves what, how deviations are handled, when to escalate"
Gräper-specific: "Based on your current process: per-supplier responsible approver, quantity deviations go to warehouse, price deviations go to purchasing, and cost center assignment gets an AI suggestion that the approver confirms"
"Skonto deadline escalation is critical for you — the system tracks these automatically and escalates when time is running out"

"This is where you configure everything — no code changes, no IT tickets"
"Your supplier-to-approver mapping lives here. Currently you have this in your heads or in informal processes. This makes it explicit and auditable."
"Deviation rules: if a quantity is off by more than 5%, it goes to the warehouse manager automatically. Price off by 2%? Goes to purchasing."
"Escalation timers: reminders after 2 days, escalation to backup after 5 days. Configurable per priority level."
"New suppliers automatically require Finance approval — no one slips through"
Handle the question: "Can we change these ourselves?" → "Yes, this is a self-service UI. Changes take effect immediately."

"Three possible outcomes from the engine"
Auto-approve: "Full PO match within tolerance, known supplier, everything checks out — goes straight to booking. No human needed. This is your target for 80% of PO invoices."
Route to approver: "When human judgment is needed, we send a notification with everything they need to decide. One click to approve or reject."
Escalate: "Multi-step approval chains, automatic reminders, deadline enforcement. No invoice sits in a queue forgotten."
Key point: "The goal is to eliminate approval fatigue. Humans only see what actually needs their judgment."

"This is what the approver actually sees — a Teams message with everything they need"
Walk through the mockup: "Invoice from Hilti, €12,450. Matched to PO, within budget, known supplier. The AI summarizes why it recommends approval."
"They tap Approve and they're done. No context switching, no opening another system, no searching for the PO."
"If they don't respond, reminders go out automatically. After the configured deadline, it escalates to the backup approver."
"Skonto deadlines are tracked — if a discount is about to expire, that's highlighted"
Gräper-specific: "You mentioned Teams and email as preferred channels. Both work identically."

"Once approved, the invoice goes directly to Infor LN"
"Full account assignment: GL account, cost center, all 6 sub-accounts — pre-populated by our AI and confirmed during approval"
"Line items mapped to POs and cost centers. Status feedback comes back to Orcha — we know if the booking succeeded or failed."
"Payment block is released automatically on approval"
"Complete audit trail stored — every step, every decision, every timestamp"
On the dashed section: "Payment preparation and SEPA export — this is something we can discuss as part of the detailed scoping. The core flow is intake through booking."

"Three concrete next steps"
Lastenheft alignment: "We've studied your Konzept V0.9 document. The next step is to formally align this concept with your existing requirements — make sure nothing is missed."
Infor LN integration: "We need a joint session with your ERP team — or your ERP provider — to define the interface. Our initial analysis suggests the BOD/API integration should work without blockers."
Go-Live planning: "Once we've aligned on scope and integration, we define the timeline and rollout strategy."
Close with confidence: "We build this as a solution tailored to Gräper. Not a one-size-fits-all product — a system that fits your process."