In June, the Financial Times published the article, ‘AI is promoted from back-office duties to investment decisions’. Generative AI, it suggests, is evolving from automating marketing operations to making complex financial decisions (e.g. whether to sell high-value stocks).
The article implies that fund managers who use AI for back-office purposes, such as simple compliance checks or writing emails, are underutilising its true capabilities. You could interpret the article as a call to investigate more sophisticated functions for generative AI, thereby making its simpler functions obsolete.
Let’s examine how back-office AI is being deployed. Currently, many businesses are deploying AI to fulfil back-office roles – like administration and HR – plus other straightforward or repetitive operational procedures, including:
Successfully automating routine tasks, however, raises concerns about potential job displacement. So far, in 2024, we’ve seen a wave of redundancies across several industries, with companies like Spotify, Tesla and Goldsmiths (University of London) implementing mass layoffs. While a direct link to AI remains unclear (and unconfirmed), these events fuel the debate around AI's role in many of those job losses.
The question remains: is AI the culprit behind these redundancies, or can it (actually) help mitigate them?
In the past, the general presumption was that AI would automate repetitive jobs (like data entry). It would also assist less-skilled accountants and auditors, allowing AI to benefit but not replace more experienced employees. However, the Financial Times article shows us that it’s not just back-office workers whose cognitive functions might be replicable by AI algorithms. Skilled fund managers must – for now – work alongside AI, essentially training and experimenting with these new technologies.
Back-office AI is much more prolific while also less sophisticated in its output than predictive or analytical AI. In the future, companies could prioritise AI that generates less developed yet more ambitious outputs – such as investment strategy recommendations, insights from virtual analysts and AI-powered portfolio decisions. The obsoletion of back-office AI would result from multiple companies redirecting their resources from back-office AI solutions to those that can tackle more complex functions.
Back-office AI products becoming obsolete is a real concern for vendors and consumers alike. A more versatile or efficient solution will consistently outperform any outdated AI model.
Of course, you could argue that, with so many back-office AI models packaged as Software as a Service (SaaS), it’s now easier to switch vendor models if industrial developments outpace the capabilities of an AI model. However, for businesses whose workflows connect to a SaaS (e.g. via API), switching vendors isn’t as easy as re-establishing a new subscription.
Switching AI vendors can be tedious and lengthy for projects with custom requirements (i.e. automated workflows for data extraction, validation and file uploads). Vendor lock-in is a side effect of outdated back-office AI, meaning many firms won’t be able to access the newest AI innovations.
To avoid allying with outdated back-office AI technology, you’ll want to do the following:
1. Use an AI vendor dedicated to maintaining the performance of their technology in line with industrial developments.
2. Build an internal AI model and proactively maintain its functionality.
3. Use a lightweight integration that will allow you to switch vendors if necessary.
One way that back office AI could become redundant is if it gets subsumed into more encompassing technologies. Let’s take data extraction, for example.
A predictive analytics tool might automatically extract data from previous annual reports as part of its workflow. In this case, data extraction, as a singular automated back-office tool, could easily become obsolete in favour of ‘promoting’ a more complex analytics function.
Yet, even in this instance, back-office AI hasn’t become obsolete; rather, it's just been repackaged. AI-powered underwriting is another example of how a firm can achieve a specific outcome by harnessing back-office processes. The speed and accuracy of AI-powered underwriting heavily depend on the development of faster and more accurate automated data extraction, as well as other back-office processes like document classification and file storage.
The line between back-office AI and its contemporaries – advanced analytics AI trusted with decision-making – is therefore more blurred than previously thought. Take Financial Statements AI, our data extraction and analysis tool for financial statements, for instance. Not only does it leverage the extracted data to calculate financial ratios, it also makes contextual decisions, such as what can be considered a ‘current asset’, etc.
So, is Financial Statements AI back-office AI?
Not quite.
But neither is it the type of predictive analytics tool that companies like JP Morgan and Legalist are striving to create. Many of these tools cannot exist without robust technological back-office building blocks.
The rate of AI’s evolution raises serious concerns about its shelf life. The AI arms race – with its influx of AI startups and competing technologies – means that committing to an AI tool can feel like a far riskier endeavour than it (actually) is. Committing to a back-office AI tool isn’t risking your tech stack becoming obsolete. Instead, it’s less risky than allying with newer and less developed AI platforms.
Conceptually, back-office AI is unlikely to become obsolete (due to its functionality and role within other AI applications). On the other hand, specific back-office AI models are likely to become inferior to other more developed models.
Interested in obsoletion-proof AI data extraction and computation? Email hello@evolution.ai or book a demo with our financial data team.
