Unleashing the Power of Gen AI with GaaS at Mollie

Here at Mollie, we're always exploring new ways to do things smarter – not just in payment solutions, but across our entire organization. It's all about making improvements, and we're excited to share a significant step forward in our AI journey: the continued growth and success of Gen AI as a Service (GaaS).

Beyond simply processing or analyzing existing information, GaaS focuses on generating new AI responses, tailored to your specific use-case. For instance, our Go-to-Market (GTM) team is leveraging GaaS for Conversation Intelligence, instantly processing sales calls to deliver concise summaries, key insights, and even draft follow-up emails, freeing up valuable time. Similarly, our Support specialists and Merchant Experience (MEX) teams are using GaaS for Support Call Follow-up, automatically transcribing and summarizing customer interactions, generating essential notes, and crafting follow-up emails.

GaaS is also breaking down critical language barriers. It powers Customer Feedback Translation for our MEX team, translating insights from any language into English to give us a clearer, global understanding. And for our Support specialists, GaaS enables truly expertise-driven communication through outbound language translation, allowing them to respond to merchants in their preferred language, seamlessly connecting with our diverse customer base. Finally, our Onboarding domain also benefits, as GaaS efficiently performs Prohibited Keyword Analysis on merchant websites, streamlining compliance checks.

In essence, GaaS efficiently transforms unprocessed data into polished, immediately usable content – be it narratives, structured insights, or multilingual communications – directly supporting our internal teams and integrated systems. These are just a few examples of how GaaS is already empowering our teams to work smarter and deliver even better experiences for our customers.

The 'as a Service' part is crucial. It signifies that this generative capability isn't a one-off script or a bespoke tool for a single use case. Instead, our GaaS platform delivers dedicated API endpoint routes for specific internal LLM-related use cases. This means that teams across Mollie can easily leverage powerful AI models for content creation by simply calling these endpoints. For example, in addition to generating reports, our GaaS platform can automatically extract key statistics and summaries from sales calls through a single API endpoint. Think of it as a central utility: you provide the input data, define the desired output via the API, and GaaS handles the complex generation process, integrating seamlessly with our existing systems to deliver automation directly where it's needed.

The Audit Reporting Challenge: A Manual Marathon (No More!)

Consider this: an auditor uncovers an important finding – maybe a process tweak, or a compliance gap. The evidence is all there, often in various structured data points: the process(es) reviewed, the work performed, the outcome, the risk, the root cause, the relevant regulatory impact, and so on.

The traditional workflow for this was a manual marathon, heavily relying on recreating data already documented, carefully crafting paragraphs to describe issues, and ensuring consistent report formatting and language across a team of diverse cultural and professional backgrounds — a significant challenge for standardization. This wasn't just time-consuming and prone to errors; it also pulled our talented auditors away from their core work of deep analysis and strategic engagement.

GaaS in Action: From Google Sheet to Gemini to Google Slides

We asked ourselves: "What if AI, as part of a scalable GaaS framework, could process our structured audit evidence to automatically generate the structured finding text needed for our audit reports, preparing it for human review and direct delivery from our internal systems into our presentation tools?"

Automating audit reporting starts with structured evidence and a precise, multi-stage AI process.

1. Our audit evidence is mostly stored in Google Sheets. Our GaaS solution automatically finds and gathers all the relevant information for each audit finding, making sure our AI models get a complete and full picture. This way of organizing wasn't just set up once; it grew and improved through many rounds of working closely with our Audit team. Their great willingness to change how they organize their data was key. This careful process gave us a clear and consistent data structure. This consistency is essential, as it creates a predictable path for our automated systems, letting them reliably find and use all the necessary information for AI analysis.

2. Once all the relevant audit information is gathered and ready, our LLM integration, powered by Gemini 2.0 Flash, takes over. While our goal is a Google Slide summary, this isn't done with a single AI command. We follow a 'one-task-one-prompt' approach: each section of the audit finding – like the Title, 'What We Audited,' 'Root Cause,' 'Risk,' or 'Action Plan' – is generated by a specific, tailored prompt. This detailed method allows us to apply precise rules, meet our internal audit standards (including tone, word limits, and restricted phrases), and maintain consistent tone and format throughout the report. It breaks down a complex task into manageable, high-quality results.

3. The final stage involves seamlessly integrating the generated content into our presentation tool. This LLM-generated content is intelligently structured to fit predefined placeholders within our standardized Google Slide template. Our GaaS framework ensures that the LLM's output for each specific section is precisely mapped and inserted into the correct fields on the slide, transforming structured text into a ready-to-present summary.

The result? Instead of auditors spending hours manually drafting and formatting, the LLM provides a high-quality draft of each finding's narrative sections in mere seconds, ready for review and finalization. This is GaaS in action: content flowing effortlessly from one system (Google Sheet) to another (Google Slides)!

The Tangible Benefits: Why GaaS Matters

This isn't just about automating audit tasks; it's a powerful validation of our broader GaaS and AI strategy at Mollie. Here’s why it’s a game-changer:

• Massive Time Savings: What used to take hours of manual work now takes seconds.

• Enhanced Consistency & Quality: The LLM ensures a consistent tone, structure, and level of detail across all reports, elevating their professionalism and readability.

• Reduced Human Error: Less manual data transfer means fewer chances for transcription errors or missed details.

• Faster Insights & Decisions: Quicker report generation means stakeholders get crucial information faster, leading to quicker remediation and risk mitigation.

• Scalability & Reusability: This modular GaaS approach for audit reporting proves the concept. The framework can be adapted to generate all sorts of structured content across Mollie, paving the way for impactful company-wide automation!

Looking Ahead: The Future of GaaS at Mollie 

Of course, human oversight remains absolutely essential. The LLM is an incredibly efficient first-draft generator, but the final review, nuanced wording, and strategic framing still rest with our experienced professionals. This ensures accuracy, context, and that unique Mollie touch.

We're actively exploring its potential across various departments, demonstrating its versatility far beyond internal reporting. The framework we've established allows us to identify and address countless opportunities to streamline processes, improve efficiency, and empower our teams by generating the right content at the right time, exactly where it's needed. We envision GaaS playing a pivotal role in transforming how we create, communicate, and operate across the entire organization.

Looking to the future, we're also exploring how Large Language Models (LLMs) will empower our teams in a more direct, user-centric way. We're actively developing specialized conversational agents – intuitive AI assistants, set up with clear instructions, for instance, to ensure content perfectly captures the 'Mollie' brand voice. These will offer powerful assistance that's easy to get started with.

By thoughtfully integrating AI into our workflows through strategic initiatives like GaaS, we're not just streamlining processes; we’re unlocking new levels of insight and effectiveness across every function. This is how we push boundaries, making Mollie more robust, secure, and efficient, with pragmatic, impactful applications of the latest AI models.