Moderated by Emily, Digital Transformation Consultant at Hyperbots
Emily: Hi, everyone! This is Emily, and I’m a digital transformation consultant at Hyperbots. I’m very pleased to have Claudia Mejia on the call with me. Claudia is the managing director at Ikigai, and the topic that we’d be discussing today is that why the industry is struggling to achieve straight-through processing of invoices honestly, there could have been no better person to speak about this. So glad to have you on board, Claudia.
Claudia Mejia: Thank you, Emily. It’s a pleasure to talk with you this morning. Thank you for having me.
Emily: Amazing. So to quickly start things, Claudia. Let’s start with the 1st question, What is the straight-through processing of invoices?
Claudia Mejia: Well, basically, it’s the process where the invoices come in. And then we automatically can process them to the ERP system. It includes not only the data capture of the invoices but the validation of the data and also the integration with the accounting systems.
Claudia Mejia: So it’s usually in a conceptually, I see process, but it doesn’t work as simple as it sounds.
Emily: Understood. So, Claudia, have you seen anywhere wherein, straight-through processing of invoices is being achieved?
Claudia Mejia: Honestly, with my experience, I usually consult small and medium-sized companies, and I have not said a thorough end-to-end process that is seamless, it usually is fragmented. You have various stages to the process, and you have a lot of manual manipulation and validation of the data. So it has been one of the challenges that I have seen with the CFOs. The procure-to-pay process is challenging and very manual. It requires a lot of effort. However, with hyperbots, we have seen that this is the solution that has been able to bridge all these gaps and bring the invoices from beginning to end very seamlessly.
Emily: Great. So because you mentioned, Claudia, that the process is a little fragmented, especially in the small to mid-sized businesses. What are the primary challenges that you have seen in achieving straight-through processing of invoices?
Claudia Mejia: Well, there are several issues, right? When you receive invoices, you receive them from different sources. You have portals, you have email IDs, all kinds of sources of information, and different types of formats which means you have structured data, and in structured data, it is very hard to control the variability in one aspect, and taking all that data and integrating it into legacy systems that they are not flexible. And so those have been some of the challenges, and on top of that, you have the resistance to change. Corporations don’t want to change their workflow. They’re fearful of technology. And the new technology that we see with AI is new. But it’s very powerful. I have seen it and is amazing for this process but those are the challenges that most companies have regarding it.
Emily: Understood. So, Claudia, if any organization is, say, PR-PO driven. Is it a better fit for straight-through processing of invoices as in what should a company do, if most of its purchases are still without a purchase, requisition, or purchase order?
Claudia Mejia: Well, PO, some PR structures are very important for this process. The more you can standardize this process at the beginning of the process the better it’s gonna be on the data validation through the process. So we want for that process to be very standardized, the templates, the formats but it’s not as easy, right? But my recommendation for the companies that don’t have a PR structure is just to try to standardize those processes, because once you bring the technology, on top of that, then it will make the whole process a lot easier.
Emily: Understood. So, Claudia, since the chart of accounts and GL codes in each company is different, you know. GL coding of each invoice is difficult to automate. Why is it so? And what can be done to handle it?
Claudia Mejia: Well, deals have all kinds of structures for different companies right, so it’s very difficult to just standardize one charge of accounts but the solution that now we have with Hyperbots is that the system now can learn by itself, so something that invoice, we can code in particular GL accounts. The system will learn that over time, and it will be very accurate and will place that spending into that GL account in the past with other processes. This is not as simple you will require manual manipulation and somebody validating that particular GL code. So the technology is there, let’s use it.
Emily: Got it. Also, I’ve seen there are so many solutions in the industry specifically OCR. Why do you think the current OCRs are not, you know, sufficient to understand the content of an invoice and is it a big handicap?
Claudia Mejia: Well, it is. Let’s describe what OCR is, which is optical character recognition. This technology is very good for converting PDFs and images into text but it has a little bit of a struggle with consistency, on the other hand, natural language processing is not only able to recognize and transfer the images to text but also understand the context behind that images on text, so it can learn by itself, which you will never find in the OCR technology, because it’s not. It wasn’t meant to be like that. So that’s what, now with AI we will be able to kind of do the end-to-end process in a way that is not as fragmented as we talk about.
Emily: Got it. So you know, diverse formats, unstructured data and lack of standards is it the reality? So is that the primary challenge for straight-through processing, and what exactly can be done to address all of these?
Claudia Mejia: Well, you have. This is something that I say. I am a process person, usually. So when I look at coordinate states or processes. I always start with the process and then we bring the technology that can fix the process. Okay, help the process not to get fixed process. But in this particular case, the technology is the one leading the process because of the technology. Now we’re able to push through the data from end to end and so my recommendation is, to make sure you understand your processes in the beginning and standardize those formats, making sure your vendors understand those formats, and then use the technology to push through the data, validate data, capture the data, and make sure it goes directly to the ERP without much manual intervention, unless you want it to be right. There are pieces that you say no, I don’t want to go directly to my ERP until I approve certain expenses so those are the main points that you can put through the system to make sure that you have the controls that you want.
Emily: Got it also a few obstacles for straight-through processing of invoices are inaccurate data in the invoice, or, you know, supplier side error, duplicate invoices. How to address these challenges?
Claudia Mejia: Well, the beauty about hyperbots specifically, because I have seen it is that the technology can not only read the data, understand the errors make sure it stops any invoices that have the errors, and then also provide recommendations. So there are a lot of good stages through the process that somebody can say, Oh, here I have the error, here I have inaccuracies. So that’s the beauty of having 1st a robust standardization, but also a good process through the book.
Emily: understood and from a regulatory standpoint, Claudia, you know what regulatory and compliance aspects should be evaluated for straight-through processing of invoices?
Claudia Mejia: Well, from an accounting point of view, there are a lot of controls that you need, right? So you need the all details, to make sure there is transparency in the transactions. You need the tax regulation that when you have invoices from different countries and different tax regulations, the system also has the flexibility to grab those types of regulations, and any inaccuracies through the process also can grab them or stop them. right? and data security. We are all concerned about data security, and make sure that all the data that goes through the system is secure and protects sensitive information.
Emily: Got it. And just to wind things up, the last question that I wanted to ask you, Claudia, is, what advancements in AI can help achieve straight-through processing to the highest possible degree?
Claudia Mejia: Well, AI has different levels of technology, right? So we have the machine learning algorithms which will help extract the information from the invoices, then we have the natural language processing which will be able to learn by itself and predict, and make recommendations and so does the magnificent view of this technology, right? This is something that we were not able to do before. And then we have the advanced analytics and so now we combine all these factors into a system like hyperbots, and we will be able to truly do it end to end. That’s what I said in this particular process technology leads it and I’m very happy to see that Hyperbots has been able to put it all together for us, and you’ll see the ROI come through.
Emily: Alright. Thank you so much, Claudia, for talking to us about the different challenges that the industry is facing with the straight-through processing of invoices, and also suggesting a couple of different measures. It was a fruitful discussion, an insightful one. So thank you so much for joining us today.
Claudia Mejia: No, thank you, Emily, thank you for having me.
Moderated by Niharika Sharma, Head of Marketing at Hyperbots.
Niharika: Hi Mike, it’s a pleasure to have you here today to discuss the role of AI in banking and finance. Jumping right into the question, how do you perceive the current landscape of AI and human interactions within the banking and finance industry?
Mike Vaishnav: AI implementation in banking and finance has been significant and is evolving very rapidly. It has played a crucial role in enabling the banking sector to keep pace with market changes. Financial services and banking handle vast amounts of data, and AI simplifies and improves the accuracy of processing this data. For example, in commercial banking, AI helps with credit card processing and credit analysis. It enables quick and accurate decisions regarding customer creditworthiness, enhancing credit lines and financial solutions. AI-driven chatbots and virtual assistants are becoming popular for customer service, providing 24/7 support and simplifying transactions. Natural Language Processing (NLP) can analyze and understand customer needs and sentiments, making decisions based on trends in customer behavior. This helps in customizing products and services to meet individual needs. In trading, AI plays a significant role in algorithmic trading and portfolio management, providing quick decisions based on market trends and optimizing portfolios. Credit scoring and underwriting are also enhanced by AI, making these processes more efficient and accurate. Additionally, AI aids in risk management and fraud detection by conducting quantitative analysis and identifying market trends to detect risks and fraud. It also assists in regulatory compliance by providing accurate data for audits, ensuring banks adhere to regulatory requirements. These are just some examples of how AI is playing a significant role and will continue to do so in the banking and financial industries.
Niharika: Thank you for those insights, Mike. What are some areas of successful AI applications in banking and finance that have caught your attention recently?
Mike Vaishnav: As mentioned earlier, handling credit card applications, credit analysis, and risk assessment are prominent areas where AI has made significant strides. AI-powered chatbots and virtual assistants offer 24/7 services and automate transactions, enhancing customer service. Financial services also offer robo-advisors for portfolio management, analyzing customer behavior, and providing personalized investment recommendations. Voice recognition is another important AI application, that enhances security for logging into banking portals. AI also provides customer insights and personalized recommendations based on trends and behavior, helping banks tailor their product offerings. Algorithmic trading is another key area where AI optimizes market information and enables quick decision-making for investments and portfolio management. These applications show how AI is deeply integrated into banking and finance, driving efficiency and personalized services.
Niharika: Absolutely. However, there must be some challenges around the implementation of AI as well. What potential challenges do financial institutions typically encounter with AI solutions, and what are the strategies to overcome them?
Mike Vaishnav: One of the main challenges is data quality and accessibility. Inconsistent or outdated data can lead to incorrect decisions. This can be mitigated through data cleansing, normalization, and ensuring robust data governance and regular audits. Regulatory compliance is another challenge, given the stringent requirements in the banking industry, such as KYC (Know Your Customer), AML (Anti-Money Laundering), GDPR, and CCPA. AI can assist, but compliance teams must continuously monitor and ensure adherence to these regulations. Explainable AI techniques can provide transparency and auditability. Security and privacy concerns are also critical, as banking data requires robust cybersecurity measures, including access control and regular monitoring. Additionally, AI can sometimes perpetuate biases in data analysis, leading to ethical issues. Banks must ensure proper data audits to avoid biased outcomes. Lastly, the talent gap in AI expertise is significant. Investing in training and developing AI talent is crucial to overcome this challenge. Despite these challenges, the benefits of AI far outweigh them, and with proper monitoring and compliance, banks can effectively harness AI’s potential.
Niharika: Looking ahead, what emerging trends do you foresee shaping the future of AI in banking and finance?
Mike Vaishnav: Advanced analytics and predictive modeling will continue to evolve, enabling deeper insights into market trends and customer behavior. Conversational AI and chatbots will enhance customer service, providing 24/7 support and improving customer experience. Robotic Process Automation (RPA) will automate repetitive tasks like data entry and document management, increasing operational efficiency and cost savings. Cybersecurity and fraud detection will also advance, with AI playing a key role in safeguarding financial data. Explainable AI will provide transparency in AI-driven decisions, helping meet regulatory requirements. Overall, automation, advanced analysis, personalization, risk management, and operational efficiency will drive AI’s future in banking and finance.
Niharika: And how do you anticipate the role of AI evolving over the next decade?
Mike Vaishnav: AI will further enhance automation, advanced analytics, personalization, risk management, operational efficiency, innovation, and R&D investment. Scalability will be crucial, as AI can scale much faster than manual processes, supporting growth and efficiency in banking.
Niharika: Are there any regional or cultural differences that influence the adoption and implementation of AI in banking and finance? How does one navigate these differences in a globalized industry?
Mike Vaishnav: Yes, regional and cultural differences impact AI adoption. Regulatory environments vary by country, and technology infrastructure differs between advanced and developing nations. Cultural attitudes toward technology and AI adoption also vary, as do customer preferences. To navigate these differences, financial institutions must localize AI implementations based on specific regional requirements. This involves partnering with local stakeholders, ensuring collaboration, and customizing solutions to meet local needs. Education and training are also essential to address challenges and promote AI adoption.
Niharika: Thank you for answering that, Mike. The discussion on the role of AI in banking and finance has been quite insightful. Thank you for contributing to this conversation.
Mike Vaishnav: Thank you for having me. I’m glad to share these insights.
Conversational UX is gaining traction in tandem with rapid advancement in AI tech. It seems intuitive that humans would want to communicate with AI agents or bots as naturally as possible. Nothing about conversational UX is new, of course. We just happen to be at a tipping point where various AI technology trends are pushing it into prominence. Substantial research and successful application of that research for real-world scenarios over the past decade have made conversational UX ubiquitous and ready for primetime where the best is yet to come.
The new age of sophisticated applications using generative AI demands that designers dig deep into the art of conversation. This is an exciting time to explore possibilities to make a dialogue between humans and bots, natural, meaningful, fun, and engaging. On the B2B SaaS front, we are just scratching the surface.
With AI infused in all kinds of finance process automation, there are a ton of possibilities to make conversational UX a key part of such applications. It begs a radical question. What if there is no traditional UI layer in finance applications? Can business outcomes be achieved through good old easy-going conversations between humans and AI solely through a chat window with no regular app interface to speak of? How can designers design and orchestrate the creation of these environments? Designers must investigate, for instance, whats the equivalent of a casual business meeting in a cafe versus a mission-critical exchange in a conference room about budgeting between a CFO and their AI assistant.
At Hyperbots, designers are exploring ways to create a humane, relatable avatar for the powerful AI capabilities of our platform addressing the automation needs of finance processes like Accounts Payable and Expense Processing for the CFOs office processes that are still woefully manual.
The core work that the Hyperbots AI Assistants do are:
Accountant queries can be as practical as asking the AI assistant about invoices that can be safely bulk-approved or the ones that need their manual review.
The AI assistant reponds with real-time actionable data about pending invoices that need manual review.
Analytics critical to business decision making can be easily pulled up with a simple query.
These are sneak peeks into the early work that’s emerging as part of the conversational UX design charter for the design team at Hyperbots. Within broad chatbot categories that exist today, here’s where we might fit in.
Designers at Hyperbots know that if they want to create distinct AI Assistant identities, they need to focus beyond the visual elements of an avatar or the UI layer of a dialogue box. They must ask the question what makes a dialogue meaningful? Especially between a machine and a human. They must dive deep into the science of Human-Computer Interaction and the art of conversation. So far our secondary design research has pointed to some seminal work already in the public domain like the recent ethnographic study by NNGroup into usage patterns of ChatGPT, Bing, and Bard users suggesting there could be 6 different types of conversations with generative AI.
These provide a great basis for brushing up on fundamentals and taking the right first step. What should follow is arriving at a solid hypothesis of what specific approaches might work for ur CFOs and their teams and then testing these hypotheses with rigor.
We are early in our exploration of conversational UX at Hyperbots. We are more than convinced this space cannot remain untapped if we are to create a groundbreaking experience for our customers grappling with legacy applications to conduct their finance operations central to the customer experience we want to build for our CFOs and their teams. As they say, watch this space!
