The World’s Best Companies Are Still Running on “A Guy Named Steve”

I still remember the moment at DoorDash when I realized we had a scaling problem.

It was 2014, before sophisticated logistics systems, before algorithms optimized every delivery route. Back then, we had Steve. Literally, a guy named Steve manually assigned deliveries. Any new driver who joined would share their location with Steve via Find My Friends. He looked at incoming orders, checked out who was where, and texted the closest person to pick up the order.

Steve was excellent at his job. The system worked…until it didn’t.

You can’t scale a company on heroics forever. As DoorDash grew, we couldn’t hire enough Steves to keep up. So we did what tech companies do: we turned that manual workflow into software. Code that could run billions of deliveries with the same reliability Steve brought to dozens. Once the repetitive coordination became software, Steve moved into building and leading the support functions around it. The work didn’t disappear. It evolved.

Since then, I’ve seen the same pattern across enterprises, whether in accounting, operations, account management, or the back office more broadly. Most are still running on armies of Steves: smart people manually holding critical workflows together because the systems around them haven’t kept up.

Most of these teams don’t have engineers waiting to turn those workflows into software, so they’re stuck. AI has started to bridge the gap, but most of the rollouts still miss the point. AI mostly shows up as an assistant inside an existing workflow — it may help someone work faster, but it doesn’t take the work off their plate. It optimizes the current system; it doesn’t change how the business operates. Software engineering has already seen what happens if you shift to a fundamentally new paradigm: AI has created the 100x engineer by fully taking over coding. We believe the people running the business deserve that same leverage.

It is time for the 100x operator.

That requires digital teammates. A digital teammate isn’t a tool you use. It’s a peer that owns the execution of a process. You teach it the goal, spot-check the work, and manage the outcomes. It takes the manual work off your plate entirely.

Over the past year, a small team and I have been building Convey quietly alongside early customers and partners, with support from Khosla Ventures, Pear VC, SV Angel, and a group of operators and founders we deeply respect. Today, we’re bringing Convey out of stealth and sharing publicly what we’ve been building.

How Convey Works: Train Digital Teammates by Just Doing Your Job

The idea behind Convey is straightforward. The people doing the work should be the ones training the digital workforce. Operators simply describe a process or share their screen and the AI Teammate observes, learns the process, and takes over execution. When the Teammate hits a new edge case, it pauses, flags the issue, asks for input, and learns for next time.

Convey is built on three principles:

1. 1. Operators Become Conductors
   The person closest to the work should own the relationship with the digital teammate. They should not have to translate their process for external consultants or forward-deployed engineers every time it changes. That handoff creates friction when you are trying to get work off your plate, and dependencies when the process evolves. Convey lets operators teach digital teammates directly by doing the work themselves, shifting their role from manual executor to conductor of a digital workforce.

2. 2. Built for Unsupervised Execution
   There is a big difference between an AI that helps you draft an email while you watch and a digital teammate you trust to process invoices all night. Most AI tools are built for one-off, supervised tasks. Convey is built for unsupervised reliability. Training an AI to execute complex workflows without hand-holding requires a high-fidelity process, and we have proven that process across hundreds of thousands of hours of unmonitored work with our customers. If a system is going to operate in the background, it has to earn your trust.

3. 3. A True Agent Identity
   Many AI products are designed for individuals, not enterprises. You can’t just drop a raw AI agent into your corporate systems and processes. Just like human employees, a digital teammate needs a holistic identity to govern their access, centralize their updates and output, and establish escalation paths for oversight and administration. Convey provides the end-to-end infrastructure to coordinate with your digital teammates so you can deploy them where they matter most — at the heart of your operations, not the periphery.

Real Results from Real Customers

Teams are using Convey to delegate workflows they previously assumed were too complex for software.

Today, we work with forward-looking businesses like NBC Universal, Samsara, Unity, Faire, and TelevisaUnivision. Each organization has its own complexity. Each has its own edge cases. That is exactly what Convey is built for.

“Convey is a cheat code that is giving us a real edge over our competitors. We have used it across every department within our organization and saw an ROI within weeks. We were able to automate inbound reservation requests, something our engineering team has not been able to solve in over a year, at a fraction of the cost. In certain job functions, we will be able to delay hiring for over 2 years, and our employees love that they get to focus on the more impactful parts of their job.”

Dom is not alone.

At ChargePoint, the sales team saves hundreds of hours per week by delegating the updates of high-volume deals across Salesforce and NetSuite.

At a large streaming service, the operations team delegated custom reporting and critical ad ops workflows to AI Teammates, saving 500+ hours per week.

At Samsara, a staff accountant trained an AI Teammate to allocate IoT spend across a global portfolio. A process that once took hours now completes in minutes.

Across our customer base, non-technical users train their first end-to-end digital teammate in about three hours.

Looking Forward

AI is moving fast, and the future is uncertain. But that’s something our team has seen before. At DoorDash, 2014 was about people believing food delivery was a thing. 2017 was getting them to believe it could be profitable. 2020 was keeping up with incredible demand. No matter the year, though, our approach was always the same: go through 100 deliveries one by one every single day and make each one better. That’s where progress came from, where strategy came from, where success came from. That’s what we’ll do with our customers at Convey.

If your team is buried in repetitive workflows, if your IT backlog is six months deep, or if people are spending hours moving information between systems, we built Convey for you.

The era of relying on “a guy named Steve” to manually run critical business processes is over. The people who understand the work best should stay at the center of it. The repetitive execution should not.