Automate supplier invoice coding in IFS using self-learning AI
Overview
Snowfox is an official IFS compatible AI add-on that automates purchase invoice coding and workflow routing with over 90% accuracy.
IFS and Snowfox will show how self-learning AI automates invoice coding and workflow routing, helping existing IFS users reduce manual effort without changing their current processes.
It runs silently in the background – no new interface, just a smoother, more automated process within your IFS environment.
What you'll learn
- High-Impact Automation: How AI predicts invoice coding, reviewers, and routing within IFS to boost throughput.
- Eliminate Manual Touchpoints: Reduce manual effort across non-PO invoicing and other exception-heavy workflows.
- Fewer Errors, Better Data: Improve consistency, reduce human error, and clean up financial reporting.
- Faster Approvals: Get invoices through approval and posting in hours instead of days.
- Real-World Outcomes: Ways teams have cut costs, freed capacity, and scaled without adding headcount.
Join IFS and Snowfox co-hosted 45-minute webinar to see how Snowfox AI helps finance teams eliminate manual invoice coding, without having to replace IFS invoice workflow module.
Want to explore this for your own IFS environment?
If you’d like to discuss how this could work for your organisation, you can leave your details below. Our team can follow up after the webinar with relevant examples or answers to your specific questions.
- Industry expertise from IFS and Snowfox
Your hosts
This session is hosted by AP automation specialists from IFS and Snowfox. You will hear directly from the teams shaping IFS invoice workflows and delivering real-world AI automation at scale for finance teams.
Jonas Silvero
Product Manager, Finance, IFS
Miikka Savolainen
Chief Operative Officer, Snowfox
Tuomas Haapsaari
Chief Growth Officer, Snowfox
- Built specifically for IFS finance teams
Who is it mainly for
This session is designed for organisations running IFS that want to remove manual invoice coding without changing how their teams work. Snowfox operates directly alongside IFS, using AI to automate coding and routing while keeping your existing workflows, controls, and ownership fully intact.
AP Managers & Finance Leaders
Reduce invoice processing effort without disrupting established finance operations. Snowfox automates coding in the background, improving speed and accuracy while giving teams more time for oversight and exceptions.
IFS Admins & ERP Owners
Reduce system maintenance effort without altering your IFS configuration. Snowfox automates coding and routing using historical data, lowering admin overhead while keeping workflows and controls unchanged.
Digital Transformation & Automation Teams
Reduce manual dependency without introducing new platforms or tools. Snowfox applies AI directly within IFS, learning from invoice history to drive continuous automation and measurable efficiency gains.
IFS Webinar FAQs
How the AI Works
Is the AI function learning on customers own invoices or is the model static and updated by IFS?
Snowfox automates and enhances your invoice coding and approval routing with AI-powered predictions, reducing manual work, increasing coding quality, and speeding up AP processing times.
Key value points:
- Up to 90% automation rate in invoice coding
- Direct cost savings of 2 – 4 € per invoice
- Faster month-end closings and fewer bottlenecks
- Significant reduction in human error and manual touchpoints
- Seamless integration with IFS – no new UI or system change required
- Improved data consistency and reporting quality
With Snowfox, IFS customers gain:
- Advanced AI expertise purpose-built for invoice coding and workflow routing within IFS
- Continuous innovation and development
- Seamless integration – AI intelligence added on top of the IFS workflow
How long history data is enough?
Snowfox can automatically predict all invoice coding dimensions used in your IFS setup, such as:
- GL account
- Cost center
- Project / Work order
- Department / Business unit
Tax code - Booking date, periodicals
- Supplier-specific dimensions
- Description/open text
- Any custom dimensions defined in IFS (e.g. asset, region)
How does the AI model understand if there has been organizational changes where invoices during the historical period it trains based on understand what is the correct cost center to post on?
Yes. Snowfox can automatically predict multiple coding lines per invoice, not just one.
The AI analyses your organisation’s historical multi-line codings to learn how costs are typically distributed. When invoices that need multiple coding lines arrive, Snowfox predicts:
- Several coding lines with the right GL accounts, cost centers, projects, and other dimensions
This makes Snowfox ideal for handling multi-line and complex invoices – with the same accuracy and confidence as simple, single-line invoices.
If an invoice has a note that it has been manually rebooked, can the AI notice this?
Snowfox predictions are based on your own historical invoice data and are continuously validated:
- Each prediction comes with a confidence score, showing how certain the model is about that particular prediction.
- Customers can set a confidence threshold, for example, only predictions above 90% confidence are auto-applied, giving you full control over automation
- AI also learns from every accepted and corrected prediction – improving accuracy over time
Customers typically reach 85–95% accuracy after a short learning phase - Finance teams always have the final control – no automatic postings without human approval (unless configured otherwise)
Customers can set a confidence threshold, for example, only predictions above 90% confidence are auto-applied, giving you full control over automation. This helps avoid errors and builds trust in AI-driven processes.
Is the model trained from scratch for each customer since that much histroical data is needed? How is training made, can snowfox extract needed historical data from ifs via ifs API/projections)?
Snowfox predictions are based on your own historical invoice data and are continuously validated:
- Each prediction comes with a confidence score, showing how certain the model is about that particular prediction.
- Customers can set a confidence threshold, for example, only predictions above 90% confidence are auto-applied, giving you full control over automation
- AI also learns from every accepted and corrected prediction – improving accuracy over time
Customers typically reach 85–95% accuracy after a short learning phase - Finance teams always have the final control – no automatic postings without human approval (unless configured otherwise)
Customers can set a confidence threshold, for example, only predictions above 90% confidence are auto-applied, giving you full control over automation. This helps avoid errors and builds trust in AI-driven processes.
Am I understanding correctly that manual validation is required before final posting, and that these corrections are used to train the AI over time? Do all invoices initially require manual review before final posting? If so, how long does that review phase typically last before the system can start auto-posting invoices with sufficient confidence? What automation rate do customers typically achieve after the first 3 to 6 months?
Snowfox predictions are based on your own historical invoice data and are continuously validated:
- Each prediction comes with a confidence score, showing how certain the model is about that particular prediction.
- Customers can set a confidence threshold, for example, only predictions above 90% confidence are auto-applied, giving you full control over automation
- AI also learns from every accepted and corrected prediction – improving accuracy over time
Customers typically reach 85–95% accuracy after a short learning phase - Finance teams always have the final control – no automatic postings without human approval (unless configured otherwise)
Customers can set a confidence threshold, for example, only predictions above 90% confidence are auto-applied, giving you full control over automation. This helps avoid errors and builds trust in AI-driven processes.
How much can you customize the AI - can you "feed it" with certain instructions for some special case suppliers etc?
Snowfox predictions are based on your own historical invoice data and are continuously validated:
- Each prediction comes with a confidence score, showing how certain the model is about that particular prediction.
- Customers can set a confidence threshold, for example, only predictions above 90% confidence are auto-applied, giving you full control over automation
- AI also learns from every accepted and corrected prediction – improving accuracy over time
Customers typically reach 85–95% accuracy after a short learning phase - Finance teams always have the final control – no automatic postings without human approval (unless configured otherwise)
Customers can set a confidence threshold, for example, only predictions above 90% confidence are auto-applied, giving you full control over automation. This helps avoid errors and builds trust in AI-driven processes.
Is the model trained per customer or globally across all tenants?
Snowfox predictions are based on your own historical invoice data and are continuously validated:
- Each prediction comes with a confidence score, showing how certain the model is about that particular prediction.
- Customers can set a confidence threshold, for example, only predictions above 90% confidence are auto-applied, giving you full control over automation
- AI also learns from every accepted and corrected prediction – improving accuracy over time
Customers typically reach 85–95% accuracy after a short learning phase - Finance teams always have the final control – no automatic postings without human approval (unless configured otherwise)
Customers can set a confidence threshold, for example, only predictions above 90% confidence are auto-applied, giving you full control over automation. This helps avoid errors and builds trust in AI-driven processes.