“Let’s go buy that business together. We’ll be a minority shareholder, we’ll put skin in the game…We all win when we do that.”
While Artificial Intelligence (AI) undoubtedly has the long-term potential to revolutionize industries, its short-term impact will be most felt in automating existing workflows. Fractional AI co-founder and CEO Chris Taylor said his company, which builds custom AI solutions, is well-positioned to deliver both those early gains and, eventually, the transformational ones.
Most of those early benefits will be realized by existing companies. Taylor said incumbents have strong market positions and deep domain knowledge. Because of the volume flowing through their systems, AI can help automate onerous workflows, enhance call center processes and provide superior customer service. Take small changes and multiply them by a company’s scale, and AI can attack previously unsolvable, billion-dollar problems.
However, to truly leverage AI’s potential, more people need to master it. Taylor explained that Large-Language Models (LLMs) are so new that few have mastered them. Traditional machine-learning knowledge won’t cut it.
Fractional AI is forcing the issue by taking top talent and creating an environment where they will succeed.
“We know how to build great engineering teams; great engineers are what is required to build these systems,” Taylor said. “We started building out a team with some of the best engineers we’ve ever worked with, a lot of former founding CTO-type profiles. We have everyone in one room here in San Francisco focused entirely on AI projects, and we foster an environment where everyone can learn from every project we do.”
Twice per week, the Fractional AI team gathers for lunch to share what they are learning on current projects — what worked and what didn’t. How did they successfully adapt APIs? How did a new AI voice system perform? Lessons are added to the company playbook, so future employees quickly get up to speed.
Taylor believes this system will scale.
“I think it will serve us well, especially given how fast this landscape is changing,” he observed. “New tools are coming out all the time, so both for our use of these tools to be more efficient in our day-to-day, and keeping up with the latest and understanding where we can apply these things to projects like the deep research API just came out.
“How have we used that on projects thus far? And what learnings can we take from that and apply to other projects? All that kind of stuff was this continuous learning curve and being at the forefront of applying these technologies to solve real-world problems for enterprises.”
Fractional AI found a niche in serving PE firms seeking to add value to portfolio companies. Taylor said the ideal involvement is when a company has some sense of what they want to accomplish with AI, and Fractional AI takes them the rest of the way.
Often, it’s a solution that was previously not cost-effective to scale to smaller clients. However, AI can replicate some of those activities to make scaling feasible.
As companies like Fractional AI mature, they can turn one-off fixes into repeatable tools, providing off-the-shelf solutions that address common problems. Taylor said that every solution his team builds is customized, but they do leverage existing models when it makes sense. Most customizing occurs in the last mile.
“That is where the majority of the effort goes, and so that’s partially why a services approach makes sense and a bespoke approach makes sense,” Taylor said. “That said, we do see a lot of patterns in these projects that enable us to move faster. So we’ve built voice projects that we have techniques for putting in place. And every time we do a voice project, we get a little faster at it; we get more familiar with all the different tools that are out there. If somebody has an idea for a voice agent they want to build, we’re going to be faster at building that.”
Taylor said improving document processing and configuration tasks can take six months to implement. Clients ask Fractional AI if AI can reduce that time frame. Often, it can be reduced to days.
The client needs to have a point person with a deep understanding of its internal workflows and at least a working knowledge of the technologies supporting the new solutions. Fractional AI helps that person navigate the implementation and maintenance.
Will AI develop to the point that companies look to it to serve as a CTO or CFO? Taylor said finance is one area where companies can consider buying instead of building.
“Every company has a finance department; every company has books. That is a trade that lends itself well to doing something off the shelf versus building custom workflows for finance. It’s not a universal truth because if you’re a very large organization, you have some very bespoke workflows in your finance department, and you might want to build something custom. But to the extent you just have a standard type of finance department that looks like everybody else, then you probably want to reach off the shelf.”
Could these AI-based systems impact organizational structures? Not much, Taylor said; it’s generally like maintaining most software.
“Newer, more accurate models are always coming out, and you might want to upgrade to the latest model. If that’s not true and cost is what you’re after, you might naturally benefit from the fact that the cost of running these models goes down dramatically over time. You might just leave the system with automatic improvement built in. As it gets cheaper to use, your variable cost of running it is going to come down about 90% over the next 12 months.”
Taylor returned to the importance of AI’s ability to scale the previously unscalable. One client performs AI opportunity audits, where it interviews a client’s employees about workflows and AI use. They show the company where they are on the AI spectrum while recommending areas most amenable to AI.
This company cannot physically conduct all of the interviews itself. AI systems do it. That allows the company to accept more work while incurring lower expenses.
Taylor acknowledges that some companies are wary of using AI due to perceived risks. Perhaps they don’t want to pay a flat cost and accept that risk.
When Fractional AI is confident that it can deliver value, it offers solutions where it gets a percentage of the cost savings. Fractional AI also deploys a co-investment model.
“Perhaps there is a PE firm looking to acquire a business with high AI potential, but it has a long client onboarding process; their whole sales process is bottlenecked, and their revenue is bottlenecked by this,” Taylor said. “Let’s go buy that business together. We’ll be a minority shareholder, we’ll put skin in the game, and then we’ll condense that onboarding process to unlock the revenue.
“We all win when we do that.”
