Artificial Intelligence (AI) is more than just a buzzword; it's a transformative force reshaping how businesses operate and innovate. In recent years, AI's prevalence in enterprise settings has soared, with Statista reporting that over 37% of businesses and organisations have adopted some form of AI as of 2020. This figure is expected to grow exponentially, with PwC predicting that AI could contribute up to £12.56 trillion to the global economy by 2030. Since then, the underlying models have matured considerably: outputs are more accurate, more consistent, and better performing across the board.
This burgeoning interest is not unfounded. AI offers unprecedented capabilities that streamline operations, enhance decision-making, and boost productivity. AI works best as an assistant: it makes work easier, and it doesn't replace human judgement.
AI-enhanced traceability: The future is now
An example of AI's potential can be seen in its application within traceability systems, such as iov42's solution, Interu, a leading technology in ensuring the integrity and transparency of supply chains to help towards compliance, improved efficiency and to save time (and money).
We know that at present, a large number of organisations are still using Excel spreadsheets to store and share their data. This just isn't scalable. With incoming deforestation regulations, like the EU Deforestation Regulation, the amount of data required to conduct the necessary due diligence to meet compliance will only increase.
Interu is already helping its users to securely store, manage and share data to help towards compliance, and we are continuing to explore and implement how AI can help manage the scale of data and conduct risk assessment. For example:
- Automated translation of documents AI-driven translators have matured significantly and continue to improve, with each generation of models handling more language pairs and more technical vocabulary reliably. For example, users of Interu can view a translated version of a foreign-language document they see in front of them with just the click of a button, and store it alongside the original.
- Document classification and management AI excels in classifying various document types, such as invoices or logs, which are crucial in trade and commerce. AI can also suggest optimal ways to split large PDFs into individual documents, enhancing file management. For example, users of Interu can upload a PDF containing a scan of multiple documents, and let AI suggest the start and end of each document contained, and identify its type.
- Optical Character Recognition (OCR) AI's ability to interpret printed and handwritten text transforms data entry, making it faster and more accurate. Rather than scanning line by line like traditional OCR, modern models read a whole page at once, so they can make sense of where information sits on it and attribute meaning accordingly. This is particularly useful in processing forms and extracting crucial data. For example, users of Interu could upload a scanned document and ask their AI assistant to automatically enter the data contained in it into Interu.
- Advanced data handling From extracting sender and recipient details to understanding line items on invoices, AI can parse complex document layouts and extract meaningful information, a task that is challenging even for seasoned professionals. For example, a user of Interu could engage the AI assistant to automatically collect issuer and recipient information, as well as the document's date, number, and all quantities and totals from an uploaded document.
- Fraud detection and security AI systems can perform plausibility checks and duplicate upload checks, adding a layer of security and integrity to transactions. For example, for a user of Interu, AI could check for each uploaded document, whether a very similar version of this was already uploaded earlier, and flag the uploaded version for review.
The challenges and considerations
While the benefits are significant, deploying AI in traceability systems still requires care. Today's leading models have been trained extensively by the major providers, all pursuing the goal of increasing general-purpose capability, so the old concern about training data quality has become negligible. The more pressing question now is how well a system is steered, controlled and governed in use: the guidelines, guardrails and oversight that determine whether AI behaves reliably within a business's specific workflows.
Getting this right matters more than ever. Integrating AI reliably into a business's specific workflows is one of the biggest challenges in taking AI from production. Mastering it, rather than simply experimenting with it, is what separates a demo from a system a business can depend on day to day, and only a handful of technology providers have shown they can do it well.
AI systems must still be used responsibly: they support human judgement, they don't replace it. The potential fines for EUDR non-compliance run up to 4% of an organisation's annual EU revenue (which for some large enterprises can be as high as €3.9 billion), so the responsibility for conducting due diligence stays with the user. AI cannot and should not replace that.
AI as a strategic enabler
As AI continues to evolve, its role in enhancing enterprise capabilities across industries will only grow. Businesses are increasingly looking beyond AI as a support to human tasks, towards AI's role in automating workflows and business processes directly, and this is where a great deal of executive attention is now turning. The two go hand in hand: automating a workflow well relies on the same steering, control and governance that keeps AI reliable and accountable everywhere else it's used.
The focus for businesses is shifting from questions of raw capability to questions of governance: how AI is steered, controlled and guided within a workflow. At iov42, we believe any AI implementation should be approached with careful testing, clear guidelines and close attention to its impact on existing workflows.
