In September 1945, thousands of elevator operators walked off the job in New York City, leaving office workers crowded into lobbies beneath offices they could no longer easily reach. Much of commercial Manhattan ground to a halt because in 1945, no one rode an elevator without a human at the controls.
Here’s the strange part: automatic elevators had existed for nearly half a century, yet no one would step into a box that moved by itself. Building passenger confidence took safety features they could understand and use, including an emergency stop button, a telephone, and a recorded voice explaining what to do. The gap between invention and widespread adoption stretched across generations.
An automatic elevator has parallels to an autonomous agent: you give it a destination and trust it to get you there. Enterprises are now deciding how much responsibility to hand over to AI agents that can take action on their behalf.
At Delta-v, we call the distance between what AI can do and what it is actually doing inside enterprises the AI Adoption Gap. We believe closing it may be the largest value creation opportunity of the AI era, as companies build the trust needed to put AI to work.
That trust breaks down in three places. Enterprises hesitate to deploy AI they cannot secure, lose confidence when early results disappoint, or need help translating its capabilities into their own work.
Closing the gap takes three things: you need to Secure It, Equip It, and Walk It into the Building.

Secure It
As agents gain access to sensitive information and the ability to act, enterprises need to understand how they can be manipulated, what data they can expose, and how to stop an attack in progress. Security gives enterprises the confidence to expand what they delegate as agents take on more responsibility.
One of our most recent investments, HiddenLayer, is securing models and agents so enterprises can put them to work with confidence.
Equip It
AI adoption can follow a J-curve, much like onboarding a new employee. Early work needs correction, productivity dips, and people revert to doing things manually. Equipping an agent means giving it the right information and tools to do the job reliably. As the foundation improves, each reliable result earns trust, and teams gain the confidence to delegate more work.
You.com’s search APIs connect agents to current web information, grounding their work in what is true today.
Walk It into the Building
Many enterprises do not know how to apply AI to their own work yet. The gap closes when people redesign the work around AI, often with Forward Deployed Engineers working alongside customers. Enterprises build trust in the teams and applications that show them how to make AI work for them.
As an early Anthropic partner, Praxent is redesigning financial services workflows around AI, while Albert is bringing AI into the work of scientists developing chemical formulations.
Across software and services, the opportunity belongs to teams that can turn AI’s capabilities into results customers trust enough to change how they work.
Today, most of us step into an automatic elevator without thinking about it. We expect AI to earn a similar place in the enterprise as the companies securing it, equipping it, and walking it into the building make delegation feel routine. We are investing in the businesses that make that confidence possible.
Capability is inevitable. Trust is investable.
If you are building a company that helps close the AI Adoption Gap, we would love to hear from you. Get in touch with our team.
The companies named above are Delta-v portfolio companies, selected to illustrate this article’s themes and not for their returns. They are not a complete list of Delta-v’s investments or representative of them. Past results do not guarantee future results, and no representation is made that any investor or company will achieve similar results. This article is not an offer to sell or a solicitation of an offer to buy interests in any Delta-v fund.


