Published on the 18/03/2025 | Written by Hayden McCall

Are buyers looking for AI features? Or is the reverse more true?

As someone who sits at the junction between IT news, technology vendor hype and the interpretation of it all into advice for businesses buying software, AI has been an unavoidable cacophony in recent times.

Artificial intelligence (AI) has become the dominant theme in business technology, with every tech vendor on the planet hyping new features as AI-driven this or AI-empowered that. And until recently, the noise has been just that: Best ignored when it comes time to get out the cheque book.

Often, AI capabilities have simply been embellishments of typical IT capabilities such as process automation, data analytics, or integration tools.

“AI is the new IT” I’ve said before, while also thinking, like most, that I’d better keep across what’s going on here.

While buyers remain sceptical, AI is rapidly maturing as the large language models that support it grow in sophistication and reduce in cost.

As AI gets embedded into business software platforms, this sophistication is arriving in the hands of software users and piquing the interest of both execs and IT procurement teams.

The AI translated a list of ERP product parameters into full sales-ready descriptions ready to upload to the web store, saving days and days of work.

The explosion of generative AI tools has excited users with the potential for this new breed of technology and driven rapid uptake. It has also heightened expectations that, at least in the immediate time frame, are likely to be dashed.

But there’s no doubt the cycle is trending up the enlightenment slope as AI use cases become clearer, LLMs improve, and vendors move from experimentation to embedding AI tooling into their products.

There are strong use cases emerging for ‘agentic’ AI that add genuine value to call centre staff helping customers resolve complex issues (which is why we called them in the first place, right?). And to be clear, I’m not talking the dreaded chatbots here, although they too are being improved (to avoid the call in the first place).

In CRM and online retailing, AI is delivering real value in surfacing customer insights and augmenting data collection.

I recently heard a great example where AI transformed a painful process of loading thousands of new products to an e-commerce site. The AI translated a list of ERP product parameters into full sales-ready descriptions ready to upload to the web store, saving days and days of work. Copilot can deliver some real value it seems.

We’re also seeing product recommendations and x-sell/up-sell suggestions that are actually useful, using AI’s ability to understand the semantics of the customer journey.

While the benefits of AI in enterprise resource planning (ERP) systems is less obvious, AI’s ability to generate insights from data, particularly in forecasting likely or possible outcomes, is looking promising for predictive analytics.

The use of natural language in search queries or in problem statements is making it easier for users to engage naturally with systems, without technical skills or detailed knowledge of the data they are searching.

GenAI is valuable for accelerating tasks, especially in contextualising search results, and is delivering improved personal productivity, but not necessarily something that sits comfortably inside an ERP.

It has also become an asset for developers by enabling quick retrieval of relevant code blocks or open-source solutions to resolve integration mapping or complex logic, significantly enhancing dev team productivity.

It [AI] is starting to not only identify data gaps or outliers, it’s then prompting users with suggestions to resolve the issue, in some cases even asking if it should go ahead and fix it.

Despite the advancements in AI, many buyers remain pragmatic and even sceptical about its capabilities.

Believing AI magic will happen can be a distraction away from the hard work of automating or integrating processes, or from improving transactional data quality. Those things are the true enablers of digital success.

But again, AI is having an impact as vendors introduce anomaly detection, alerting users when transactions are outside of norms so they can be fixed at source thereby improving data quality.

It is starting to not only identify data gaps or outliers, it’s then prompting users with suggestions to resolve the issue, in some cases even asking if it should go ahead and fix it. The net result is better quality, where AI-driven analytics can then start to play their part.

But are these trends translating into requirements that buyers are taking to market?

From my experience, not yet. Buyers are very engaged with the possibilities of AI, but they may not know yet the specific AI features that they require. What they want to do is align with vendors that have a strong story around how AI will be incorporated into the product they are evaluating. Which explains the hype. But the story must be believable and in pragmatic terms that people can get their heads around.

In our recent work interviewing local ERP leaders as part of iStart’s 2025-26 ERP Buyer’s Guide, the use cases listed in the table below were the common threads, as an attempt to meet that measure.

I’d expect features such as these to appear more regularly in software buyer requirements lists as the market matures.

One thing is for certain, AI technology will continue to evolve and improve. As it does, its role in business technology will create tangible applications that drive genuine efficiency and innovation.

The AI news is all good for innovation and productivity, but I worry about only one thing: Who will end up paying for the massive investment that the sector is attracting? Watch this space.

EMERGING USE CASES FOR AI IN BUSINESS TECHNOLOGY

 

Function	Use Cases
Process automation	Agentic AI is embedding capability within ERP software to allow a user’s transaction patterns to be added to knowledge bases resources, or to be automated for rote processes (aka embedded RPA). AI applied to AP & Sales processes to augment existing OCR scanning or to create quotes off sales enquiries, assisted by semantics, anomaly sensing and fraud detection. Improved bank reconciliations with AI logic & reasoning to recommend matches and reduce manual input.
Analytics	Enhancing demand planning and predictive analytics tools to create forecasts or predict outcomes off historic trends, tailored with plain English queries. Generating contextual insights off large data sets e.g. summaries for board packs, periodic financial statements or customer profitability. Cash management tools to evaluate scenarios and determine (and enact) the best e.g. chase debtors sooner, offer (or take up) early payments, set up dynamic discounting etc. Anomaly detection alerting where transaction exceptions occur such as in fraud protection (bank account matching) or keying errors. Supply chain insights assessing actual delivery performance or factoring in vessel location tracking, shipping disruptions, strikes or weather impacts.
Content	GenAI is providing tools to jump start content creation tasks (including in data entry auto-population).
Software engineering	GenAI applied in code development is accelerating problem resolution for developers.
Support	Plain English queries by users to get instructions on how to complete tasks or comply with company policies. ChatBots to contextualise queries and provide self-service solutions.
Search	Surfacing summarised search results from natural language queries, either via AI-enhanced web search or chatbots configured to search internally across policy documents and unstructured data.

Source: iStart ERP Buyers Guide 2025-26, vendor interviews and related literature reviews

 

Hayden McCall is managing director of digital publisher iStart technology in business and owns Software Shortlist, a consulting business helping companies to buy better software. He has been involved with IT and the software industry for over 25 years.