{"id":24419,"date":"2023-06-30T10:55:00","date_gmt":"2023-06-30T10:55:00","guid":{"rendered":"https:\/\/silverfin.com\/resources\/lessons-learned-bringing-ai-to-accountancy\/"},"modified":"2026-02-16T13:21:39","modified_gmt":"2026-02-16T13:21:39","slug":"lessons-learned-bringing-ai-to-accountancy","status":"publish","type":"post","link":"https:\/\/silverfin.com\/en-gb\/resources\/lessons-learned-bringing-ai-to-accountancy\/","title":{"rendered":"How AI is Transforming Accountancy: Lessons Learned"},"content":{"rendered":"\n

This update looks way beyond the last month and even into future years, so buckle in!<\/p>\n\n\n\n

On the 15th of June in the heart of Battersea Park, a real show for accountants<\/a> was organised. With laser lights and silent disco headsets inside, food trucks outside, it was truly an impressive event, but the most memorable were of course the customer interactions with the speakers and exhibitors on the latest technologies and trends in accounting<\/p>\n\n\n\n

In a recent keynote with our GM, Cameron Ford<\/a>, we discussed the transformative impact of AI on accountancy, sharing valuable lessons from Silverfin’s journey<\/span>, as well as to give our modest view on what the future holds.<\/p>\n\n\n\n

We were also excited to announce our upcoming launch of the Silverfin Assistant in the UK.
<\/p>\n\n\n\n\n\n\n\n

For those new to the scene, we\u2019ve been working on bringing AI to accountancy since 2017, started development in 2018 and launched in 2019. Back then it was called machine learning, but, as the famous quip goes, the difference between machine learning and AI is that machine learning is written in python and AI is written in powerpoint.<\/p>\n\n\n\n

So what have we learned over those years of building, running and selling our AI solutions? Let\u2019s focus on four main things:<\/p>\n\n\n\n

1. Start with time savings<\/span><\/strong><\/h2>\n\n\n\n

For deploying new technology to accountants, or in general any change in people\u2019s way of working, one may expect an initial hesitation, call it healthy scepticism. By initially focusing on the time-saving benefits of AI, users quickly recognise its value, streamlining daily accounting tasks and improving efficiency.<\/span> It also makes it an easier sale, taking into account that we don\u2019t sell technology like AI or cloud, but specific solutions that make the day to day of accountants easier. Finally, also internally at Silverfin, being able to save time for our users shows success. Success breeds success, so it\u2019s really motivating for our AI team and our company as a whole.<\/p>\n\n\n\n

2. Trust is key<\/span><\/h2>\n\n\n\n

Trust is key. AI is probabilistic and often a black box. That means you need to build up trust in the results. If 19 out of 20 analysis are good, and one is way off, you\u2019re not going to trust it. Especially in so-called number professions, experts are used to excel like right-or-wrong calculations with clear inputs, not probabilistic suggestions or indications, even though they are super valuable and insightful. While we recognize this trust needs some time to build up, we did find that typically after 5 to 10 sessions users become familiar with it and trust its outcomes. Accountants are sometimes less conservative than they\u2019re given credit for.<\/p>\n\n\n\n

3. The User Experience (UX) is just as important as the AI<\/span><\/h2>\n\n\n\n

Again, with these AI models given probabilistic answers, it becomes crucial how you present and show the outcomes on the screen. It\u2019s important to err on the safe side. Your AI model may have highest accuracy but if the outcomes are confusing or not presented well, the user may not trust it and stop using it. We learned that a poka yoke<\/a> approach of iteratively taking away possible misunderstandings, brings about the best experience for a broad range of users.<\/span><\/p>\n\n\n\n

4. Assist, don\u2019t automate<\/strong><\/span><\/h2>\n\n\n\n

Finally, we strongly believe in assisting accountants with AI rather than automating their work. By pointing out action points or suggesting inputs or opportunities, we make their work process more comfortable and efficient and keep them in charge. Silverfin offers many points of automation, as any platform should, but for these advanced methods, our users prefer to confirm suggestions rather than having to trust that the automation is perfect. It\u2019s often described by our customers as having an extra pair of eyes for more peace of mind. So you are effectively hiring an extra member of staff, who\u2019s never off sick, who\u2019s always available and who is willing to do the boring stuff very quickly. That\u2019s why we call it Silverfin Assistant!<\/p>\n\n\n\n

For all of this, the safety aspect can not be overlooked, and we consider this just as crucial as our cloud security, personal data protection, confidentiality etc. I would argue that it is nothing new, but you need to take it just as seriously and bring in the experts.<\/p>\n\n\n\n\"Download<\/a><\/span>