Josh Malate is CoFounder & President of Ultimarii. Photo by Paulina Ochoa,
Josh Malate is no stranger to building things, whether it’s businesses, software, or infrastructure. The cofounder and president of Ultimarii, an AIpowered regulatory technology company, took the stage at YYC DataCon to break down what it really takes to grow an AIdriven business today.
As AI reshapes industries at an unprecedented pace, automation is no longer just a theoretical discussion. Professionals in law, engineering, and other highvalue fields are actively looking for ways to integrate AI into their workflows.
The challenge, however, is that while technology itself is advancing rapidly, the playbook for building enterprise software companies in this new landscape remains uncertain.
Malate’s session tackled this uncertainty headon.
Drawing from his experience scaling AIpowered software, he shared his strategies for navigating rapid change, achieving adoption in complex industries, and growing a business while staying capitalefficient.
His company, Ultimarii, is a case study in AIdriven expansion, moving from an idea to more than $1 million in contracted ARR in just nine months.
The success, Malate says, is due to five core strategies.
1. Verticalize and stop worrying about total addressable market
Most startups are pressured by investors to define their total addressable market, but Malate has abandoned that approach entirely.
“If a VC asked me what the total addressable market is, I say, ‘That’s your problem, you do the calculation.’”
Instead of starting broad, Ultimarii takes a hyperfocused approach, going deep into specific verticals where AI can make an immediate impact. The company targets professionals who navigate complex regulatory processes, believing that these specialized users provide a clearer path to product adoption than chasing massmarket applications.
“There are tens of thousands of verticals we can build against,” he explained. “Winning even a narrow market provides multiple expansion strategies — moving into adjacent verticals, integrating up and down the value chain, or merging with other verticals.”
By locking in on a niche and solving a painful, expensive problem, the company avoids competing with horizontal AI solutions that aim to serve everyone but risk being too general to be useful.
2. Focus on adoption, not just technology
For AI companies, having strong technology is not enough. The real challenge is getting people to use it. Malate has seen this firsthand with enterprise AI implementations that fail to gain traction.
“I go into these large enterprises, and my pitch is down pat,” he said. “I ask, ‘How’s your Copilot implementation going? Has it failed yet?’ And they always laugh.”
The issue, he argues, is that companies are implementing AI in the wrong places. Many early AI deployments focus on lowvalue workflows like HR chatbots or autogenerated IT support tickets. While these projects are easy to execute, they don’t provide meaningful improvements to professionals doing complex, highstakes work.
Ultimarii takes the opposite approach by embedding AI into highvalue workflows, where automation can have a tangible impact. Instead of replacing lowlevel tasks, the company builds AI tools for regulatory professionals who deal with thousands of pages of documents and shifting legal requirements.
Malate sees parallels to the 1990s, when companies struggled to integrate tools like Excel and Lotus 123 into their workflows. AI adoption, he believes, will follow a similar trajectory, requiring companies that specialize in configuring and deploying AI where it creates real value.
3. Embrace technical debt and rebuild constantly
Technical debt refers to the tradeoffs made when software is built quickly, often leading to outdated code or systems that need to be reworked later as technology evolves.
And it’s something Malate and his team think about a lot.
Ultimarii’s approach to software development rejects the traditional idea that technical debt is a problem to be avoided. Instead, Malate believes that companies in AI must be comfortable accumulating technical debt because foundational AI models are evolving so quickly.
“We’re going to accumulate a whole bunch of debt, but declare bankruptcy every six months,” he said. “Bankruptcy is that the foundational models are actually innovating behind us, and we can actually just leverage the innovation that they have done to get rid of a bunch of the debt that we’ve created.”
The bigger issue Malate is talking about is that the AI landscape changes so fast that anything built today will likely be obsolete within months. Rather than resisting that, Ultimarii embraces it. The company assumes that much of its code will be thrown away and rewritten in response to new advances in foundational AI models.
Malate sees this as an advantage for startups because companies that began building a year ago may already be at a disadvantage because they started with a tech stack that is now outdated. New entrants can build with today’s best tools instead of being locked into older choices.
To make this work, Ultimarii has structured its engineering process around rapid iteration and rebuilding. Instead of clinging to existing infrastructure, the company treats each sixmonth cycle as an opportunity to reassess and replace technology that no longer serves its purpose.
4. Iterate product ideas at a breakneck speed
Malate’s team operates on a level of speed that many product managers would find extreme.
While most enterprise software companies iterate on a quarterly basis, Ultimarii’s product roadmap is reviewed and updated every week.
“Can this go from concept into production in 45 days or less? If it can’t, we are looking extremely closely at that decision,” he says.
By forcing strict timelines, the company avoids investing in features that may no longer be relevant by the time they are completed. If a project can’t be built and deployed quickly, it’s likely that the problem itself will change before the solution is ready.
This mindset recently saved the company a significant investment.
Ultimarii had considered investing up to $500,000 in data enrichment and tagging but repeatedly postponed the decision. A week ago, a new AI embedding model was released that completely solved the problem, eliminating the need for that investment.
5. Be cautious with venture capital
Malate is both candid and selfaware about the role of venture capital in AI startups. While he has successfully raised funding himself, he encourages founders to question whether it’s the right path for their business rather than assuming it’s the only way to scale.
“I think VCs are crack dealers, and they get you addicted to crack, which is capital, and that is, their business is to sell you that,” he joked.
The real challenge, he explains, isn’t venture capital itself but the uncertainty around longterm pricing models for enterprise AI.
Unlike traditional SaaS, where companies can lock in longterm contracts and predictable pricing, AI software costs are fluctuating rapidly. As foundational models improve and become cheaper, startups may struggle to maintain pricing power, making aggressive growth strategies riskier.
Because of this, Ultimarii is careful about how much capital it raises and deploys.
Malate is wary of taking on excessive funding only to find that valuations later collapse due to shifting market dynamics. His advice isn’t to reject venture funding outright, but to be intentional about whether it aligns with the company’s longterm strategy.
The race to build AI companies that last
Malate’s approach to building AI companies is rooted in speed, adaptability, and a willingness to rethink everything. In an industry where foundational models evolve in months, not years, staying ahead means building for change, not stability.
Success in AI isn’t just about having the best technology — it’s about making it useful. Companies that focus on adoption, rather than just innovation, will outlast those that chase hype. Finding highvalue use cases and integrating AI into real workflows is what separates lasting businesses from shortlived experiments.
Raising capital is no different. Venture funding can accelerate growth, but it also comes with expectations that may not align with AI’s unpredictable business models.
Malate’s view is that capital should serve the company’s strategy, not dictate it.
The real challenge isn’t just moving fast, it’s knowing when to shift direction. AI companies that stay rigid risk being left behind, while those that embrace uncertainty and keep iterating will be the ones that last.
is the official media partner of YYC DataCon 2025.
This article was created with the assistance of AI. Learn more about our AI ethics policy here.
