Tellen launches AI platform in partnership with three US accounting firms

We are excited to launch Tellen's Accounting AI Studio.  It’s the first generative AI platform purpose-built for accounting firms to access and develop AI applications. Tellen’s Accounting AI Studio provides secure infrastructure allowing accounting firms to launch their own branded AI platforms for employees and clients. The solution empowers client-facing knowledge workers to access pre-built accounting AI applications from our app store, and build their own custom AI applications using natural language in minutes. Employees can easily use the applications across their portfolio of clients and share applications across their teams. The Studio enables accountants to develop productivity-boosting accounting AI applications as easily as they create Excel spreadsheets today. 

‍We are excited to announce our first three design partners, already rolling out Tellen’s Accounting AI Studio. Here’s a sneak-peak of what our platform looks like—and its key features

‍

‍

‍

‍

‍

Our founding story, and why we are building what we're building

How it started 

Just over six months ago, my co-founders and I started our journey of building Tellen. We started our research with a simple question: how can AI change the way audits are done? 

It was clear something needed to be done. Audit regulator PCAOB found a more-than 40 percent deficiency rate among public company audits last year. For context, the audit of an average S&P 500 company takes around 30,000 hours and costs over USD 13m. So even with all this time, money and resources spent on audits, the resulting ‘audit quality’ is often poor. Major accounting scandals and frauds are not uncommon amongst public companies and there are often restatements of financials in subsequent periods. 

We instantly fell in love with the problem—and knew we could solve it. Ultimately, the US economy is built on the foundation of good bookkeeping, and the only way to enforce that is good auditing. Having previously worked as an auditor at Ernst & Young, I found the concept of ‘audit quality’ cited ubiquitously, and the firm clearly spent hefty resources ensuring that audits were of the highest possible quality. Poor audit quality is a multi-faceted problem, and AI as a technology—from Generative AI to Machine learning—is the perfect tool to address the problem. 

‍Fall in love with the problem, not the solution

Over the last six months, we have completed over 90 product research calls with participants in the audit and accounting industry. We have spoken with Partners, Directors and Managers from national accounting firms, as well as representatives from regulatory agencies, standard setting boards and software solutions providers to understand the problems around audit quality that are unsolved in this space. 

Our objective was to learn about the problems and understand how AI can play a role in transforming the way auditors do their work. We used the feedback from the conversations to build prototypes. We have recently signed our first three design partners and we are rolling out our solution to their employees. 

‍What did we learn?

The industry is going through major structural changes. A record number of CPAs are leaving the profession causing an acute shortage of qualified accountants in the country. At the same time, the costs associated with outsourcing are going up significantly. Accounting firms are constrained for resources and are squeezed for margin. They are actively looking for areas that will improve efficiency to improve their margins and take on more work. Another thing that was very clear to us during our research and conversions was that other than the Big 4, there is a lack of innovation in the audit software industry. The industry is going through a massive shift from on premises to Cloud solution for the first time and AI provides a great opportunity for firms to leapfrog innovation. 

Here are our learnings and insights around how AI can transform this profession.‍

Not so surprising findings

1. Almost everyone recognized that AI could be a game changer for their profession. Most people came up with several ideas and repetitive tasks that they do which could be partially or fully automated using AI (if only they knew how)
2. Most firms have banned ChatGPT citing data security concerns
3. Some people were skeptical about the reliability of results from Large Language Models and whether technology is production ready for real client use cases

Surprising findings

1. Majority of the people had their own ideas on how they can solve their own problems and automate their productivity using AI but they don't know how to get started. The tech is seen as intuitive, but not necessarily easy or quick to use. There’s a lot of trial and error and context that is needed to get the best results and who has the time for that ?
2. Majority of the firms are thinking about building their own AI Chatbots with proprietary data. This goes way beyond using ‘Chat GPT for Enterprise’ or Microsoft Co-pilot. Firm owners are thinking about how AI can be a fundamental shift for the professional services business model and the way in which services are provided (i.e an employee and client facing chatbot) trained on the firm's knowledge and expertise. ‍

So how do we solve all this ?

Problem : I cannot trust AI models with confidential client data 

Solution: You need to be in control of the data and models you use. This means getting private instances of all AI models and hosting data within your own cloud infrastructure. We never store any of your firm or client data on our servers. Our platform is hosted on the firm’s infrastructure with private end points to all AI models. We will be using encryption for data in transit and advanced network security features for users including user access controls and MFA, Intrusion detection and prevention and DLP Monitoring.

Problem:  I cannot trust the output from LLMs

Solution: Large Language models like GPT-4 that power chat applications (ChatGPT) are very important but only one piece of the puzzle. Our solution goes way beyond LLMs and our platform brings together various state of the art AI models that are good at particular tasks to offer a complete AI System that allows you to automate complex workflows. Our platform also has the best validation tools in the industry. We see this technology as an enabler and a tool that allows you to use AI to augment your work. 

Problem: It takes too long, LLM doesn't have all the context

Solution: Results from AI models can be enhanced significantly through optimization. We are building the client data infrastructure that allows auditors to integrate various data sources, and provide the relevant information and context to the AI models right within their existing workflows. Our models are also optimized for Accounting use cases and specifically trained on documents and knowledge that are relevant to Accountants. 

Problem : I have ideas but I am not a coder and I don't know where to start

Solution: This is where our builder tools come in. We make it really easy for you to build any workflow automation that will help you with your client work. We have built the underlying technology that allows Accountants to use natural language to build powerful applications. For example, you can use natural language to extract structured data from documents, analyze the data, match data across various documents, fill out checklists, update templates and generate workpapers. 

Once we went through the exercise of building a framework to solve everything we heard, we had our ‘Aha’ moment and were decided on what our MVP needs to address. We are very excited today to launch our AI platform with three clients, and are looking forward to playing a key role in revolutionizing the Audit industry. 

If you are an accounting firm owner that is interested in learning more about services, please contact us here.