[Editor’s Note: Agents of Transformation is an independent GeekWire series, underwritten by Accenture, exploring the people, companies, and ideas behind AI agents. Join us Tuesday, March 24, for our Agents of Transformation event in Seattle.]
Just a year ago, AI-powered sales platform Nooks wasn’t using many AI agents, instead relying on prompts and pre-trained models to help customers bolster their sales strategies. But that’s changed in a big way.
“We’ve injected them into almost every part of the stack since then,” said Nooks co-founder and CTO Nikhil Cheerla, speaking at an event the company hosted in Seattle in February.
Nooks’ rapid adoption reflects the growing attention on vertical AI agents — tools built to do one job exceptionally well by combining models with domain-specific data, workflows, and context.
General purpose AI models can quickly generate text, write code, and summarize reports. But they aren’t as proficient at solving industry-specific tasks. And that’s where vertical agents come in.
For this installment in our Agents of Transformation series, GeekWire examined the rising trend of vertical AI agents, and the big opportunity for startups.
“Large AI platforms may become broad distribution engines for intelligence,” Madrona investors Sabrina Wu and Vivek Ramaswami wrote in a recent analysis of the AI landscape. “But specialized companies will continue to emerge by getting the hard parts right in specific domains.”
Jerry Zhou, CEO at Seattle legal tech startup Supio, described vertical AI as “a shift from tools to agents.” Supio’s software helps lawyers quickly sort, search, and organize case-related data.
“It’s not enough for AI to generate insights — it needs to operate within real workflows and take action,” Zhou said. “In legal, that means transforming complex data like medical records into verified, structured outputs attorneys can rely on without second-guessing.”
The new technology is helping startups uncover value for customers. Prophetic, a Portland, Ore.-based land acquisition intelligence platform, trained its AI on more than 20,000 municipal zoning codes in the U.S. “We removed a critical bottleneck and unlocked an entirely new way of operating in one of the world’s largest industries,” said Prophetic CEO Oliver Alexander. “That’s the true power of vertical AI.”
The shift is attracting attention from investors like Mia Lewin, a Seattle-based tech vet who just raised an inaugural $5 million fund for TheFounderVC, her new firm focused on vertical AI startups.
“We expect this space to mint over 300 unicorns in the next decade, with the first Vertical AI IPOs hitting the market within three years,” Lewin said.
Speaking last month at Nooks’ event, Pulumi CEO Joe Duffy described how Pulumi’s AI agent, Neo, helps companies automate cloud infrastructure tasks like optimizing costs and ensuring compliance. The goal behind Neo, which launched last year, was to make an AI agent capable of doing everything a human infrastructure engineer would — not just answering questions, but taking action across complex systems.
“One of the special parts of a vertical agent is that you can really go deep into one domain,” Duffy said. “And that domain is not just LLM tokens. It’s much more complex than that.”
Building these systems requires more than a model. It needs what some call an “agent harness” — the surrounding infrastructure that helps orchestrate tasks, find context, and verify outputs, Wu and Ramaswami noted in their post.
Vertical AI agents are already automating various types of manual work — going far beyond traditional software-as-a-service tools.
“Turning workflow context into execution is the opportunity for vertical AI agents, and what will distinguish winners from those who only generate content or recommendations,” said Doug Tallmadge, CEO at Seattle marketing AI startup Gradial.
Startups that pair vertical AI agents with strong contextual data could pose a threat to incumbents. Cheerla, the CTO at Nooks, said a company like Salesforce has billions of data points — “but they don’t know what’s good and what’s bad in that data.”
“The way we’re trying to design Nooks is to collect very high quality data, so we completely get the context that led to a decision being made,” Cheerla said.
Nooks’ agents handle end-to-end sales workflows, including identifying accounts, finding contacts, drafting emails, and assisting reps on live calls. They can be invoked manually, run in bulk, or operate in the background, and are designed to work in collaboration with human users.
The next phase of vertical AI agents could go beyond simple task execution. One emerging trend is agent-to-agent collaboration, where multiple systems work to solve complex problems.
“You can think of a swarm of agents actually collaborating together to get something done,” Duffy said, drawing parallels to his earlier work designing distributed systems.
Another shift is toward proactive agents — systems that don’t just respond to instructions, but initiate actions on their own. That transition, however, may take time. Even as agents become more capable, companies are moving cautiously when it comes to handing over control.
Duffy referenced an “autonomy slider,” a term coined by AI researcher Andrej Karpathy, that ranges from fully human-controlled systems to fully autonomous agents.
For low-risk tasks — like cleaning up unused cloud resources — companies may allow agents to operate independently. But for high-stakes actions — like deploying production infrastructure — human oversight remains essential.
“You need to first build trust and build quality into the systems you’re building,” Duffy said.
Capable vertical agents are already beginning to reshape how companies structure their teams. Cheerla described evolving the traditional model of engineering organizations, where product managers facilitate information sharing between engineers and customers. He said that process can be automated with agents, and engineers instead should be directly connected to customers and get ownership over outcomes.
“You need to get rid of these pipelines and bottlenecks,” he said.
At Pulumi, Duffy described a shift where every engineer is effectively the lead of their own team of agents. “The engineers who can think like a product manager and a staff-level engineer are able to literally be 100x developers,” he said.
Investors with Bessemer Venture Partners say vertical AI “represents a fundamentally larger opportunity than vertical SaaS ever did,” in part because of how it impacts workforces.
“Unlike vertical SaaS, which typically captures a fraction of Fortune 500 IT spend, Vertical AI taps directly into the labor line of a P&L,” they wrote in a blog post.
Sharbani Roy, a vice president at chip design firm Arm who previously helped build Alexa at Amazon, offered a unique frame for how human employees interact with agents: the apprentice model.
Rather than thinking of agents as automation tools, she encourages her team to ask a different question. “How are you using an agent to help act like an apprentice to make you better?” she said at the panel discussion. “What did you do this week that you were able to achieve — but better — because you had an agent helping you? How are you making higher and higher judgment calls?”
