You cannot buy a product from people who don't live in it.

In the summer of 2024 a job application came through the form on our website. Name: Chris Vlessis. Position: contractor. Why are you a great fit for Repli: “I have built multiple solutions that are used extensively at Repli.”

Which was true. Chris had owned Shape, the paid-media budget software we ran our ads through for years, and he had sold it to NinjaCat in 2022. He was applying to us the way you text an old friend to see if they want to get a beer.

Some context on why that form mattered as much as it did.

Four years, two contractors

MultiHub started in 2020 as a way to solve our own problems. We were running websites, search, ads and reporting for operators all over the country, by hand, with a dozen logins and a spreadsheet for everything, and I said out loud that I wished there was a Shopify for multifamily. So we gave it a name and did what a company with no engineers does. We hired some.

I want to be fair here, because the second contractor did real work. The data pipelines, the BigQuery tables, the integrations our performance reporting still runs on today. That was them, and it was good. What they could not give us was a product that solved anything for the people inside Repli beyond showing a customer their numbers. Part of that was on us. We never had a project manager who fully understood the problems we needed solved, so we handed a team of very capable strangers a list of features and hoped the product would show up.

It doesn’t work that way. A contractor can understand your tickets. They cannot understand your problems, because they don’t live in them. That is not a knock on anyone. It is just what the arrangement is.

Four years in, we had something almost working, and a team that had watched two rounds of engineers come and go.

Ten days

Chris came on as a contractor on August 1 to look at what we had and tell us the truth. Ten days later he sent me a document.

Some of what it said: “What tangible progress has MultiHub made towards solving our own problems?” And: “I have more concerns with what is being built than how it is being built.” And the line I have thought about most since: “It is incredibly tempting to try to build a comprehensive platform that solves all problems. I have never seen this work.”

He also wrote, almost as an aside, that every engineer on the project being a contractor was probably making all of it worse.

He was right about all of it, and I knew it while I was reading it. That is a specific feeling. Someone tells you the thing you already know, in plain words, with no stake in your feelings about it, and you either get defensive or you get moving.

Day two

I flew him to Atlanta. Day one was the plan. He walked through everything we should do, and the elephant in the room was that he was explaining it exactly like the person who was going to lead it. At some point I stopped him. Man, all signs are pointing to you being the one leading this team. This is exciting to you. You’re behind the vision.

That night the two of us had dinner, just him and me, and it was a good conversation. The next morning I had his press release written. Chris Vlessis joins Repli as Chief Technology Officer. He had not said yes.

I wrote it because I thought it would get him over the hump. Crush whatever doubt was left. Get him excited about getting back to the work he loves, get him out of San Francisco and back to Bend, and give him the kind of freedom he’d had running his own company. Show him the ending and let him decide whether he wanted it.

He had three conditions. An office in Bend, Oregon. A salary. Equity. I didn’t have to think about any of them for long. Chris was the unicorn we needed, and it was now or never.

He started as CTO on October 1, 2024. Bend became our second office.

Untangling the slinky

Chris went balls to the wall. He built a GraphQL API and moved the whole front end onto it. He got the data model solid. He ripped out more bad code than I want to admit we had. In his words, he spent the first three months untangling the slinky.

The part I did not expect was the trust. Early on he built quick tools for the team running our client accounts, small things that took real work off their plates. That team had been told twice before that the software was coming. Within a few weeks they knew this time was different, because they could feel it in their day.

We built more in the first three months than we did in four years without him.

What I’d tell you

If you are paying a contractor to build your product right now, here is the whole lesson in two parts.

The foundation and the architecture are the entire game at the start. If you don’t get that right, you are going to rebuild from the ground up, and you will do it later, when it costs more and the team has stopped believing.

And a contractor is great for a proof of concept. Past that, you are doing more harm than good, unless you find a unicorn. We got lucky. Ours filled out the form on our website.

You cannot buy a product from people who don’t live in it. So we stopped trying.

Chris would tell this story differently. He’d probably use a lot more curse words.