There’s nothing wrong with Bob the Builder, but should companies invest their money and scarce IT talent in do-it-yourself solutions? During the roundtable on AI, the present specialists don’t think so, although this might change in the future.
‘Bob the Buildering’ becomes a verb at the AI roundtable, organized by ITdaily. Bob is synonymous with tinkering and do-it-yourself. The little man is capable, but the children’s show doesn’t address one pressing question: if you can build something, does that mean you must build it yourself?
Self-Build
The experts at the table certainly don’t think so. “Companies are starting to build the entire infrastructure around agentive AI themselves,” observes Gianni Cooreman, Presales Director at Salesforce Benelux. “They combine building blocks from Azure, OpenAI, set up private instances, integrate an LLM, and think it’s solved. Then I think: that’s possible, but are you really going to keep up with that technology, and do you have the experience and people in-house to continue doing so, while everything is evolving so quickly? I think we’re back to having many Bob the Builders these days.”
I think we’re back to having many Bob the Builders these days.
Gianni Cooreman, Presales Director Salesforce Benelux
Cooreman is not alone in this observation. He’s at the table with Christophe Robyns, managing partner of Agylitic, Joachim Ganseman, research consultant at Smals, Lander T’Kindt, co-founder of Donna, and Maarten Callaert, co-founder and COO of Paperbox.
AI vs. Classic Project
Bob has supporters in the group. “You can be a Bob the Builder,” Callaert thinks. “But you need to work on where the revenue comes in. AI is strategic, but also a hype. We see companies wanting to keep AI internal, but that’s not always the best choice. It’s not easy to keep up with the latest developments.”
read also
From Agents to Your Thermostat: AI Cannot Be Put Into One Box
Ganseman also sees that difference with more traditional IT solutions. “AI is different from a classic software project. Here, you don’t just need to update a version number occasionally. You need to continuously actively monitor your AI project. The development is never finished, and that brings higher costs. It’s possible that just a few months after a successful implementation, you need to make architectural changes again.”
The group nods in agreement. Around the table are representatives from different corners of the AI field: Cooreman represents a larger player with Salesforce, while Callaert and T’Kindt bring the vision of two young startups. Ganseman approaches AI from the risk-averse perspective of the government, and Robyns focuses on project implementations tailored for SMEs. In fact, Agilytic is a bit like the real (professional) Bob the Builder in the story. They all agree on the complexity of AI and the high pace of evolution.
Done Is Far From Done
T’Kindt concretizes: “It may be that everything works well, but after some time model drift occurs. New data in business systems can cause a successful AI implementation to deteriorate over time and perform less well.”
New data in business systems can cause a successful AI implementation to deteriorate over time.
Lander T’Kindt, Co-founder Donna
“Maybe after a year you really need to completely overhaul everything,” Ganseman also emphasizes. “Something like that is not comparable to traditional bug fixes. Such problems can really occur, and you need to take that into account.”
Think before you act, is the message. Those who consider themselves do-it-yourselfers and screw together an AI solution internally are only at the beginning of a long and intensive maintenance journey.
The Same Mistake Over and Over
Cooreman compares the situation to the beginning of the internet. “We were first in the phase of HTML and CSS, WordPress only followed later. It’s the same now. I do see more and more no-code and low-code solutions appearing in the market, but today there still remains a certain complexity. Think of multi-agent architectures across different platforms, multilingual speech technology, or unlocking data and APIs towards agents. I am convinced that this will be solved by the sector in the coming months and years.”
IT seems to consist of an endless repetition of similar mistakes. Ten years ago, people like Cooreman had to explain to companies why it might not be a good idea to put everything on-premises. The added value of the cloud, where IT is outsourced so that companies can focus on their core tasks, was far from obvious back then. With AI, this lesson needs to be learned again.
Maturity is on the horizon. ‘Bob the Builders will be much simpler in three years,’ predicts T’Kindt. ‘I still advise against starting completely from scratch. In a few years, every company will have set up its own combination of AI systems, but just as we saw with IT systems in the past, not every business can develop everything themselves.’
However, today the pace of innovation is still too high for customers to keep up with everything themselves. ‘I’ve been working at Salesforce for nine years, and the pace of innovation has never been this high,’ illustrates Cooreman. ‘Half of all our R&D people are working day in and day out on AI development.’
What is Your Core Task?
The message from the experts to companies is clear. To begin with, AI is a tool to solve a problem, not the goal of an IT project itself. With problem-solving in mind, AI can offer added value, but is costly AI development at this early stage really part of your company’s core task?
Specialized companies, both large and small, are currently ready to offer solutions and take over the complexity. This seems to be a more realistic approach for now. Those who put on their own overalls, plaid shirt, and yellow helmet to get started themselves should realize that they will have to continue intensively playing the role of Bob the Builder for a long period. A self-made AI project is never completely finished.
This is the second editorial article in a series of three on the theme of AI in practice. Click on our theme page to see all the articles from the round table, the video, and our partners.