MSPs are missing out on a huge opportunity to use generative AI (gen AI) to their advantage.
That’s according to Nicole Reineke, distinguished product manager and director of AI strategy at N-able, who said that many MSPs have failed to fully embrace the technology due to a lack of expertise, as well as being worried about making a mistake.
It’s reflected in the fact that while gen AI is estimated to be worth more than £120 billion to the Channel by 2028, 25 per cent of MSPs aren’t currently using the technology at all, N-able’s Horizon report has found.
Reineke has been in technology for about 30 years, including 20 in AI. She started working in machine learning (ML), data and image processing algorithms in 2004 with Accusoft.
Reineke has worked in the MSP space three times, starting as a founding member of Connected Corporation, which was acquired by Autonomy, which was itself bought by HP, before moving onto Stratus Technologies and then Hanock Software. She also worked for Unidesk on virtual desktop software, as a senior distinguished engineer at Dell Technologies leading advanced research on information and intelligent data management in AI and ML-based technology, and was most recently senior vice president of innovation for Iron Mountain.
The biggest gamechanger Reineke has seen during her time in the industry has been the advent of transformers. She added that, over the last three or four years, the technology has accelerated the development of deep learning and gen AI.
As far as present challenges are concerned, Reineke said that the most significant one is the growing complexity of AI. That requires having the right expertise in place to make sure it’s being correctly used, she said.
“You can do a lot of things with AI, but there’s a little technicality called drift, which means that even if you programme rules and create models, it’s possible for those to shift over time,” said Reineke. “That’s why you must only use the technology with humans who can augment it and determine whether it’s rational or not.”
A key risk that emanates from drift, said Reineke, is AI or ML bias. She said that this occurs when the information input by humans to train the model is faulty, poor or incomplete, resulting in inaccurate and biased results.
The problem, said Reineke, has been exacerbated in recent times by the shortage of in-house experts who can operate AI. Those two factors alone, she said, have made it extremely hard to implement the technology within the telecommunications industry and to ensure its reliability.
Adoption challenges
Another reason why MSPs have struggled to adopt AI, said Reineke, is fear of failure. She said that they were also anxious about making a mistake when implementing it, which could ultimately materially impact their customers.
“The other problem is a lack of understanding of how to create a return on investment,” said Reineke. “Even when they are doing something as simple as using the technology to automate people’s day jobs, they have to put in an initial upfront investment of both time and finance, and to do that without being able to calculate when they’re going to see a return on it makes it very difficult for them to implement.”
Reineke said that AI’s most common use among MSPs was for content generation and marketing. That is closely followed by enhancing customer support through the use of tools such as chatbots, and data analysis and automation of routine tasks, she said.
“As far as enhancing customer support, it’s about moving to a more natural language sounding answer as opposed to a pre-canned answer,” said Reineke. “It’s also moving beyond the standard Q&As to suggestion engines that prompt the customer to make further choices, thus enabling businesses to increase their average sales by three- or four-fold.
“Where data analysis is concerned, the technology is being used to improve sales and the operational aspects of the supply chain, and work on business strategy. But it also requires people to walk alongside those models who can question how the technology can be used to project what may happen next and augment the decision-making.”
By industry, Reineke said that AI has been most widely adopted in the B2C market, particularly in decision-making and upselling. It’s also enabling businesses to better run their back-end operations, such as microservices, she said.
Key opportunities
In terms of new AI opportunities, Reineke said that the big one was security enhancement. She said that the technology can be used to detect anomalies in user behaviour and data, and suggest it is reviewed.
“That’s a really great opportunity for MSPs to start pushing to their customers,” said Reineke. “By using AI in this way, they can also counteract the AI being used by bad actors.”
Another key opportunity area within AI, Reineke said, is in improving the customer experience. In that respect, she said that repeatable tasks can be automated to take simple actions, which can be quickly and easily reviewed by humans.
“Real-time support is key,” said Reineke. “That requires proactively identifying an action or activity that impacts the customer and then reaching out to inform them about it rather than waiting for them to identify it.”
AI can also enable MSPs to reduce their ticket response times and costs, said Reineke. Additionally, it can be used to identify security vulnerabilities that need to be patched, she said.
Reineke said that there are four levels of AI adoption: buying another company’s product, tuning (taking your data and adding it to another company’s AI model), combining different types of models, and building your own models and rules from scratch. The more complex the level, she said, the larger the investment required.
“If you don’t have much quality and well curated data to work with and you aren’t an expert in dealing with it, the first option is the best,” said Reineke. “But if you have the data and know what you want to do with it then tuning is ideal, and that’s what many larger MSPs are already starting to do.”