Trained on an enormous pool of data, AI can give us information and ideas previously thought out of reach. But when that pool of information is too vast, and the answers it generates are too vague or wrong, who is there to push AI in the right direction? Meet the experts wrangling AI, and all the solutions it can offer, into a shape and form that makes a real impact for businesses.
Hired guns: "As a systems integrator, we come in as a hired gun to implement a mature process, create outcomes, solutions, and applications, build a repeatable set of processes, and hand over the keys—or manage them for clients," explains Nicholas Clarke, Head of Cognitive AI at Intelagan.
Intelagan works with startups and companies to modernize their technology in Agentic AI, Web3/Blockchain, Automation, and Cloud technologies. They are one of the few solutions providers stepping in to draw connections between AI’s latest developments and a company’s operations.
"That’s where you might see companies like Accenture booking billions in generative AI," Clarke adds. "They’re coming in to execute on low-hanging fruit in the portfolio so executives can actualize their pet projects."
Narrowing in: Generative pre-trained transformers (GPTs) and other foundational models are built on hypergraphs of relationships, a curated web of knowledge designed to apply across domains. While powerful, this collection of knowledge doesn’t reflect the specific needs, strategies, or goals of a business.
"Inside GPTs and pre-trained foundation language models, they have their own type of hypergraph of relationships between things," Clarke says. "But that’s not your business’s knowledge and strategy. If last year’s strategy is different from this year’s, there needs to be a way to differentiate the vast universe of foundational model knowledge from what makes your company unique and differentiated."
Framework: Clarke emphasizes that AI strategy is a roadmap, requiring careful prioritization that reflects where a company is from an operations standpoint. "Strategy is about getting to a place, how to get there, and what intermediate steps are needed," he says. "I take a broad look at a company’s portfolio, work with leadership to prioritize what impacts their top and bottom lines, and synthesize their mission and goals into a top-down prioritization. The effort has to align with the maturity of their ops processes. Many organizations aren’t there yet—they’re not mature in AIOps or deploying multiple generative AI models with low error and hallucination rates."
Situated AI: Together with AI strategy researcher Allende Kemp, Clarke is developing ‘situated AI’, a framework that involves codifying an organization’s unique attributes.
"To situate your organization means to codify what is differentiated about you," Clarke shares. "If everybody has access to AGI-like language models, how do you stand out? It comes down to self-awareness. The knowledge you have, how you systemize and codify it, interacts with foundation models to filter what makes sense for your company, strategy, business unit, domain, and specific processes."
