AUTHOR/EN
Götz Kümmerle
WTS and PSP ( Peters Schönberger & Partner) have founded an AI Joint Venture that has the potential to shake up the industry. This is because they invented an AI approach which allows tax functions to build their own generative AI applications. This new approach puts pressure on consulting firms and tool providers alike as it turns the customers into kings.
It is no coincidence that these two consulting firms joined forces: WTS CEO Fritz Esterer and PSP partner Stefan Groß are not just two well-known figures in the German tax scene. They both have written industry history by revolutionizing the consulting market. Each of them in his own way.
Fritz Esterer and his WTS made possible what the former monopolists in the consulting market, the Big Four, declared impossible, namely offering tax consulting for billion-dollar companies without having an audit arm. Wirecard and the EU Audit Act proved Esterer right. Groß and PSP, on the other hand, showed the conservative Bavarian SME boutiques that process optimization, digitalization, compliance management systems and SMEs are not mutually exclusive - and that a boutique law firm can become a technology leader.
Now, Fritz Esterer and Stefan Groß are opening a new chapter which again could write industry history and once again this causes a lot of unrest in the industry. The two are joining forces in the field of generative AI by providing a “do-it-yourself tool kit” for tax functions.
The rise of ChatGPT produced by OpenAI and with the support of Microsoft has completely changed the digitalization landscape in the tax industry. There is no tax tool that can do without an AI chatbot. The Big Four companies are currently investing billions in software and application concepts based on generative AI. The industry's new tax AI products are diverse and cover a wide range of companies' pain points: from radically streamlining processes and workflows to completely new and modular content management systems, interactive simulations and automation.
Great hopes are pinned on these products - not at least because of the considerable investments involved. However, according to a market observer who wishes to remain anonymous, the new tax AI products often still contain a much older core. A chatbot based on generative AI merely supplements existing, traditional technology. Both are then repackaged together and advertised as an innovation.
Toolbox vs. black box
The problem for tax functions when evaluating such software applications is to find out what is old and what is really new and innovative. Tax applications are often a black box whose exact functionality sometimes does not even seem to be known to those who sell them.
The black box syndrome has a long tradition in the tax industry. According to market observers in the industry, KPMG's tax application TaxOne was long regarded as a prime example of a holistic large-scale application that appeared non-transparent from the outside. In the mid-2010s, almost every Big Four company had a large-scale application like this or similar with a more or less opaque structure, according to an industry insider. However, intransparency is not an obstacle to sales for software providers, but actually a recipe for success, as SAP and Microsoft have proven with their products for decades.
In recent years, however, tax technology has taken a different path. Under KPMG Head of Tax Innovation & Technology Christian Stender, TaxOne has been gradually transformed into a modern and modular platform solution, according to one industry expert. Industry leader EY has even dissolved its Tax Technology service line and anchored the topic back into the individual tax departments. The era of holistic software concepts seemed to be over.
With the advent of generative AI, there was suddenly a risk that the transparency that had just been gained in terms of software would disappear again. After all, a chatbot can perfectly disguise the content behind it. It is no coincidence that the main criticism of ChatGPT is that it is not clear which sources it uses and how the answers are generated.
Fritz Esterer and Stefan Groß are now moving in the opposite direction with their Joint Venture: they provide control functions with a kind of guided toolbox with which they can test and try out for themselves how they can use AI. "In order to understand what AI can do in a tax context, you have to experience it, and this is precisely the experience we create with our AI Playground approach," says WTS CEO Esterer, describing the maxim behind the joint venture idea. Every industrial company currently has AI on its agenda in one form or another. "You have to get people on board and get them excited, that's how reservations are broken down and a wonderful symbiosis between humans and AI is created," says Esterer.
This sounds more harmless than it is for the tool and consulting industry, because Esterer and Groß are now making consulting and software clients fit for AI themselves. Companies that cannot or do not want to afford the costly creation of generative tax AI themselves have so far had to rely on the advice of the consulting industry and tool providers. This is now changing thanks to the WTS PSP cooperation.
Power to the user
This is because the WTS PSP toolkit does not place users in the control function in front of completed solutions or front ends, but takes them directly into the engine room of the AI applications. The toolkit does not require any code or programming knowledge. The technology layman in the control function can thus build and create their own tools. If in doubt, they can try out the software solutions offered by consultants and tool providers for themselves. In this way, the toolkit reduces the dependency of tax functions on consultants and software providers when it comes to AI. The user is given the power to design. The customer becomes the AI king. This significantly raises the bar for consultants and tool providers.
Groß and Esterer describe their joint venture's offering as an "AI Payground", although it is much more than that. It offers tax functions a test and control environment for advisory content and technology. And it also allows in-house tax experts to develop their own applications without any external input. "In many cases, we are also dealing with highly specialized issues, such as Pillar II or transfer pricing," says Groß, explaining the starting point for the use of AI in tax functions. "The companies have already accumulated enough knowledge themselves, which they can now enrich with AI with the help of the Playground and make available to all employees." The AI solution serves to make this in-house knowledge quickly accessible and implement it.
As a modular system, the solution can be used by any company. However, the modular system will remain something for large companies and corporations. This is because, like any generative AI application, it also requires an appropriate amount of data sources. Companies will only gain real added value from the toolbox if they have enough of their own documents to feed their toolbox with. For small and medium-sized companies, the database is likely to be too small.
In order to provide the WTS PSP AI with a certain amount of basic knowledge, the solution will initially be filled with data and input from WTS. To this end, talks are currently underway to prepare selected tax content from NWB Verlag* in Herne, which will then be made available as a customized self-service. This makes the application different from the "Otto Schmidt Answers" product recently presented by Taxy.io and the Otto Schmidt publishing house. It provides an insight into the publisher's entire tax portfolio via generative AI.
Data protection should be taken care of: Although the application uses ChatGPT 3.5 technology, it stores data in a separate cloud hosted on European soil. Company data only remains available to the respective company and does not leave EU borders. This is an essential prerequisite for tax functions to be able to use the toolbox at all.
Questions to ask questions
In addition to special bots, users can access various text- and data-based basic tools or use special assistants that replace or perfect prompting. "For the development of additional AI applications, we have set up an AI community with numerous representatives from well-known tax departments. Here, we are particularly interested in listening to what the tax function wants and not what we think it might want," emphasizes Stefan Groß.
Special attention is paid to prompt engineering: the question of what questions to ask the AI. As with the Oracle of Delphi, your own fate depends entirely on asking the right questions. Here, WTS and PSP rely on so-called intelligent prompts, i.e. prompts that help with the creation of prompts. It sounds a bit like the Matrix or Terminator films, and in essence it is: artificial intelligence helps you to improve yourself. These intelligent prompts ask very specific questions to determine the facts of the case so that a valid tax result emerges from the query at the end. This is why the process is also known as reverse prompting, because the prompt asks questions instead of answering straight away.
These prompts can be improved as required in the WTS PSP toolbox and stored in your own prompt libraries so that they can be used to create more and more AI applications. In this way, the tax functions toolbox allows you to delve deeper and deeper into the rabbit hole of your own tax-relevant business issues.
Thinking a little into the future, tax functions could connect their tools to PowerBI, SAP S4 / Hana and business intelligence applications. It is already a milestone that SAP S4/Hana not only connects the finance module to the tax function as before, but also enables access to warehouse logistics and sales. There are completely different possibilities, especially for VAT and customs applications. Regulations such as CBAM and CSRD are also forcing tax functions to delve ever earlier and ever deeper into company facts and processes in order to check their tax and customs relevance and make appropriate statements on business decisions.
"One fits all" is passé
The toolbox therefore offers tax functions the opportunity to delve into the depths of their own corporate structures and create different applications from company to company. Holistic applications from consultants based on the "one fits all" principle are now a thing of the past.
But why is WTS investing in a concept with the toolbox that gives the customer independence when, as a consultant, you actually have more of a dependent customer? The fact that WTS, of all companies, is pursuing a concept that is basically destructive and counter-intuitive from a consulting perspective can be explained by the history of the former Siemens tax department.
While consulting firms such as EY, PwC, Deloitte or KPMG work for tax functions, at WTS it has been part of the core of the brand from the very beginning to take on tax functions in whole or in part. This was just as true for Siemens as it was for E.ON or Allianz. For WTS, Business Partnering therefore not only means being an extended workbench, but also always taking over the tools used. With the control functions, WTS also took their software solutions on board. In contrast, the other large consulting firms only struggled with these business applications on the mandate and could always argue that the applications from PwC, KPMG, Deloitte or EY were actually much more suitable.
According to an industry insider, WTS already had to manage an almost unmanageable number of individual solutions, VBA macros and Excel programs in the mid-2010s, the number of which increased with each additional takeover of a tax function. From WTS's point of view, it is therefore logical not to rely on a single generative AI application in the style of "one fits all", but on a modular system that can also be used simply and modularly in your own business partnering model.
Projects as money-makers, not knowledge transfer
But Esterer and Groß would not be who they are if they were only shaking up the market in this respect - their toolbox contains yet another tip. By making the content of NWB and WTS tax functions available for self-service, the two are initiating a further paradigm shift: pure knowledge is no longer how consultancies earn money. This is because the toolbox provides consulting knowledge without automatically generating a consulting order.
This is knowledge that might have been billed as a memo a year or two ago," says Stefan Groß. However, the pure provision of knowledge is no longer something for which the client is prepared to pay for hours in the long term. "Instead of replicating knowledge, consulting in the future must be about marrying technology and knowledge in order to break up established processes and consistently rethink them," says Groß. Esterer agrees with this: "Theoretical knowledge is no longer something that clients will ask for in the future."
Just a few years ago, however, law firms in particular had a reputation for producing excellent expert opinions on specialist tax issues - and not caring at all about how a tax function could then implement these in practice. The classic lawyer's self-image was very similar to that of the Oracle of Delphi. Anyone who ventured to seek advice would receive a more or less cryptic answer from the other side of the desk from the spheres of tax law. And were then allowed to see for themselves what they did with it after paying the hourly rates.
By simply making this knowledge available using generative AI, Groß and Esterer are burying this type of business model in tax consulting. 'Not the provision of knowledge, but its implementation in the specific business process' is the new service credo for consulting that Groß and Esterer are pursuing.
AI turbo for the entire industry
The two companies are thus burying the hopes of the Big Four of being able to distil additional benefits via generative AI from the tens of thousands of documents and expert opinions that these companies have created for hundreds of clients over the years. Simple and fast secondary marketing of reports that have already been billed will no longer work in this way - not even if all client-specific content is removed. Simply pouring existing knowledge into an AI and thus opening up new business areas will not be enough if the tax function already has the toolbox. By providing knowledge by the way, Gross and Esterer are forcing other consultancies to go beyond the mere provision of knowledge in their offerings.
With their Joint Venture, Esterer and Gross are therefore becoming an AI turbo for the entire industry - from tool providers and consultancies to knowledge brokers such as specialist publishers. As if the development of AI wasn't already happening fast enough in terms of content and time.
* JUVE Verlag is a subsidiary of the NWB Group.
*This article has been translated by DeepL Übersetzer: Der präziseste Übersetzer der Welt
