'Partners need to pivot to make the most of AI': CloudSmiths CTO
Thomas Folwer discusses the future of AI in the channel
The channel can benefit greatly from the rise of GenAI, but it needs to deeply understand, assess and then pivot its offerings and business practices to align with the technology, claims CloudSmiths CTO, Thomas Fowler, in an interview with CRN.
2023 was a time of discovery and exploration for generative AI in the channel, but to fully take advantage of the opportunity, businesses will need to rethink processes from the ground up.
"Generative AI provides channel businesses with an entirely new set of tools to help them do more with less resources," Fowler says.
"At its core, it can greatly increase engineering productivity, reduce time to market for delivering solutions, and automate many repetitive tasks to scale up solutions.
"This fundamentally impacts how tech channel companies need to think about approaching problems and estimating work efforts when AI capabilities like auto-coding, requirements analysis, content generation etc. can accomplish in hours what used to take days or weeks.
"Rather than seeing it as a threat, there is a major opportunity to leverage AI to operate more efficiently and competitively.
"But it requires channel leaders to have a vision for how to integrate it operationally across their organisation."
AI spending and budgets
AI may also play a role and impact on this year's IT spending, but Fowler explains it will depend heavily on the specific organisation.
"Large enterprises with extensive resources may choose to invest very heavily in their own compute infrastructure and AI talent to build custom models tailored to their needs.
"However, most companies will be more focused on consuming AI services from the major cloud platform providers.
"While spending on underlying infrastructure may not change radically for them, we are seeing budgets increasingly shifted to procuring skills, services, and platforms to enable adopting AI solutions effectively."
Fowler predicts this is where the majority of new spend will occur - acquiring specialised talent with AI/ML expertise and partnering with AI-focused consultancies and services firms.
On the other hand, Gartner distinguished VP analyst John-David Lovelock told CRN last November that while it "doesn't get more hyped than AI", GenAI more specifically hadn't had the chance to get off the ground yet.
"The new generative AI hasn't really had the opportunity yet to earn money from enterprise," he said.
"2023 really was the year of the story. Organisations had the time to come up with the story of how generative AI is going to affect their product, industry, go-to-market, valuation, interaction with customers, wherever they felt they were getting the most bang for their storyline.
"2024 is going to be the year of the plan, where enterprises are actually working through the plan.
"2025 is when we're going to really start using generative AI."
Fowler adds that spending also follows maturity, early on it was more speculative experimentation but as clear AI use cases emerge, companies are now ready to allocate real budget.
"While hype may have peaked, we're seeing meaningful investment start to turn on as businesses realise the competitive necessity of integrating AI operationally.
"The costs and challenges of adoption are coming into focus after the initial excitement phase."
CloudSmiths' journey in AI
Fowler explains that CloudSmiths has been focused on data engineering and machine learning for the past six to seven years.
In the last 24 months, the company fully pivoted to specialise in generative AI, which Fowler sees as a real advancement in AI capabilities beyond just classification tasks.
"As a consultancy, our core focus has been on helping organisations adopt AI solutions in a sustainable way to solve real-world commercial problems and drive ROI in their businesses.
"We work on building in-house expertise in AI as well as developing accelerators and centres of excellence to enable companies to effectively scale their AI deployments in a measurable, repeatable way.
"Rather than just hype and proof-of-concepts, we aim to help businesses integrate AI operationally and understand how to grow this capability strategically."
Automating sustainability?
GenAI could potentially also help companies process massive volumes of unstructured data related to sustainability much more efficiently, Fowler claims.
"For example, automating the analysis and standardisation of fragmented formats like PDF sustainability reports. This kind of data processing at scale could enable better visibility and decision making.
"However, we have to balance that benefit against AI systems' own growing energy and compute requirements."
In fact, some reports suggest AI could become one of the largest global electricity consumers.
"So while AI offers clear potential to optimise certain tasks, it's a complex equation and the overall net impact is very hard to quantify accurately," Fowler adds.
"We're still lacking comprehensive data and models to provide definitive conclusions either way.
"There are positive use cases but also real risks and unknowns. We can't fully understand the trade-offs and downstream effects yet.
"Any sustainability gains would need to outweigh efficiency gains in usage. It's an area that requires creative thinking and being aware of second order effects."
The future of AI
Looking ahead, the CTO explains that advancing GenAI capabilities will be the priority for 2024.
"Within that, two key areas stand out.
"One is expanding the context window for generative models.
"Currently, models can only leverage a limited context when generating content or answering questions before accuracy deteriorates.
"Work is needed so models can interpret and synthesise larger bodies of information and handle more complex, long-form tasks."
The second priority area, Fowler claims is developing more specialised AI models tuned for specific tasks or industry verticals.
"Rather than general utility, companies need tailored AI solutions that excel in legal services, medicine, engineering, etc.
"Pre-trained foundations will be fine-tuned to create niche, skilled models rather than being average at everything.
"Expert systems that augment human capabilities in defined domains will drive the most impact and value. We'll see more bespoke AI models engineered for purpose rather than crudely pre-trained ones."