Release night is one of technology’s more peculiar rituals. A room of clever people spends months preparing a change, then stays awake to discover whether the change was clever too. Pizza arrives. A deployment stalls. Somebody who promised to be home at eight begins negotiating with midnight. Ted Elliott knows the scene because he spent nearly two decades building software in the Salesforce ecosystem. When he explains what Copado sells, however, he generally skips the catechism of pipelines, metadata and continuous integration. The product, he says, means people “get to go home for dinner.” He occasionally adds a second benefit, with the timing of a fellow who knows enterprise software could use a joke: fewer divorces.
This is not the usual way to describe DevOps. It is much better. A category that can vanish into diagrams becomes a story about an empty office chair and a full chair at the family table. The description also contains Elliott’s advantage. He did not arrive at Copado from a distant corner of management theory. He had already lived the customer’s problem for 18 years.
Act IThe long road to the short release
Elliott grew up in San Francisco while Silicon Valley was acquiring both its fortune and its mythology. He studied history at Washington and Lee University, graduating in the class of 1994, then attended the University of San Francisco School of Law. His subjects included securities, venture capital and intellectual property. He paid his way by working full time as a recruiter in biotechnology, a practical education in how companies find the people who become the company.
The early career was an improbable but useful braid: recruitment, law and life sciences. Elliott served as general counsel for Eximias Pharmaceuticals and became a partner at OPV Lifescience Partners, which combined executive search, seed capital and corporate advice. In 1999 he started Jobscience, recruitment software that would eventually find its defining platform in Salesforce.
“Eventually” matters. Jobscience did not pivot to Salesforce until 2010. The company became one of the ecosystem’s earliest independent software vendors, but its progress was the patient kind, measured across market turns rather than demo days. Elliott built sales, consulting and engineering teams. He learned how cloud software breaks, how customers complain, and how a release that looked harmless can occupy an entire evening. In March 2018, Bullhorn acquired the business.
Later that year, he joined Copado as chief executive and a board director. The fit had a satisfying circularity. Federico Larsen and Philipp Rackwitz had founded Copado in Madrid after enduring the same sort of overnight deployments Elliott knew from Jobscience. Their ambition was to make release days obsolete. Elliott arrived with the scar tissue of a customer and the habits of an operator.
Act IIA unicorn is a milestone, not a personality
Copado expanded after Elliott’s arrival. It opened a North American headquarters in Chicago, established a wider international presence and built a certification community around Salesforce DevOps. In 2021 the company raised a $140 million Series C at a reported $1.2 billion valuation. That year it also acquired Qentinel, adding robotic testing to its reach. Copado’s current company history lists more than 1,200 customers, more than 350 employees, operations across at least six countries and a community above 100,000 practitioners.
The company’s geography tells part of the story. Two engineers began it in Madrid. The headquarters moved to Chicago. Teams spread through Europe and India, while Elliott led from New Orleans. Copado was selling coordination to enterprises while practicing a version of it across its own time zones. That made the dinner test more than a slogan. A failed handoff in distributed software can chase the sun: one team logs off, another inherits the puzzle, and a third wakes to the consequences. Release management is partly the art of preventing a small surprise from becoming an international itinerary.
The numbers explain scale; they do not quite explain Elliott. For that, consider the chess clocks. He once gave them to his managers, a sly piece of office furniture that made meeting time visible and reciprocal. If one person monopolized the conversation, their clock revealed the expense. It is a comic device with an operator’s point beneath it: time is a budget, even when nobody has put it in a spreadsheet.
He is based in New Orleans, a city connected to his mother’s family and sufficiently distant from the standard Silicon Valley mise-en-scène. The company he leads is distributed across countries and continents. He also serves on the advisory board of Washington and Lee’s J. Lawrence Connolly Center for Entrepreneurship and has been associated with support for public education and youth sailing. The pattern is relational. Companies are systems, but Elliott tends to explain them through the people trapped inside the system when it fails.
Act IIIThe useful embarrassment of being wrong
Artificial intelligence tested that instinct. Two weeks after ChatGPT appeared in late 2022, Copado went hard at the technology. The company applied AI to customer support and saw encouraging early results. Scores rose. Answers looked better. Case-closing time fell. Management concluded that automation could handle roughly half the work and reduced the support organization.
Then the bill arrived. Complaints grew louder and accounts began to leave. The AI could produce fluent answers, but the smaller human team had become less practiced at recognizing when an answer was subtly wrong. Elliott later wrote about the failure plainly. The noteworthy part was not that an ambitious software company got carried away with a new tool. This has happened roughly every Thursday since the invention of software. The noteworthy part was that its CEO made the correction part of his argument.
His position now is neither retreat nor rapture. AI can write, diagnose and review at startling speed. It can also be confidently mistaken. The answer is to give it context, narrow its permissions, preserve an audit trail and build quality gates around what reaches production. Humans should remain responsible for approval. Elliott’s shorthand is the junior developer: productive, energetic, worth encouraging, and absolutely not allowed to merge unreviewed code because it sounds certain.
Context is the unglamorous hinge. A general coding assistant may know how software usually works; an enterprise system contains years of local decisions, exceptions and dependencies. Elliott has pointed to the difference between quickly producing code and producing code that fits an old, complicated Salesforce organization. The latter requires knowledge of existing metadata and logic, plus the humility to reuse what is already there. In a greenfield demo, novelty looks intelligent. In production, restraint often is.
a change
and policy check it
what ships
This emphasis brings him back to DevOps’ oldest promise. Automation is not valuable because machines are fashionable. It is valuable when it removes the fragile, repetitive work that keeps people awake. Security follows the same rule. Elliott has argued that controls are working when developers stop finding ways around them. “Complexity that nobody follows isn’t security,” he wrote. “It’s theater.” The good guardrail performs its duty without asking everyone to admire it.
The next releaseSpeed, with someone answerable
Elliott can sound blunt about what comes next. He has said that AI agents will affect entry-level Salesforce jobs and that traditional DevOps roles will evolve. Yet his more durable idea is about responsibility, not replacement. If agents create more changes, companies need more ways to verify those changes. The premium moves toward people who can define requirements, review architecture, spot an edge case and decide whether the machine’s confident answer belongs in the real world.
That makes his career look less like a collection of pivots and more like a continuous argument. The recruiter learned that talent matters. The lawyer learned that accountability has an address. The Jobscience founder learned that platforms create leverage and dependencies in equal measure. The Copado CEO learned that faster releases are only better when they are also safer. AI has not erased those lessons. It has made them arrive at greater volume.
The technology industry enjoys grand forecasts because tomorrow is wonderfully exempt from quarterly review. Elliott’s best line works in the opposite direction. Dinner happens tonight. Either the deployment lets a person attend, or it does not. Either the tests catch the mistake, or somebody returns to a glowing laptop while the plates go cold. For all the language now surrounding intelligent agents, the measure remains disarmingly ordinary: did the software work, could the team prove it, and did everyone get home?