In August, I started building Nerve. The idea was fairly simple: a SaaS company could have perfectly healthy dashboards while one of its biggest customers was unable to use the product.
Think of a system handling a million requests a day. Almost all of them succeed, so nothing looks particularly wrong. But the failures could all belong to one customer whose integration has stopped working.
Existing monitoring tools can surface those errors. I wanted Nerve to connect them to the affected customer, recent deployments, support conversations, and the actual business impact.
I spent weeks building it.
The beta was working
By the time I had a working beta, I'd built secure data ingestion, background workers, account-level incident detection, and tests covering 15 provider contracts. I could replay a simulated incident where one customer's requests kept failing while the rest of the system appeared healthy.
Technically, I was making progress.
Commercially, I had almost nothing.
I sent 403 cold outreach emails trying to find companies willing to test Nerve. Some failed to deliver, and the campaign produced almost no useful design-partner interest.
The question I couldn't answer
I started participating in public discussions about production incidents and monitoring. Those conversations were more useful. Engineers were already dealing with customer-specific failures, and some were using their existing tools to track them.
That last part mattered.
The problem existed, but I still hadn't shown why someone needed another product to solve it.
During a conversation with an investor, I was asked to explain what Nerve actually added beyond existing logs and APM tools.
I thought I had explained that pretty well. Apparently, I hadn't.
I went back and rewrote the landing page around a single workflow, from detecting a failure to identifying the customer, investigating the impact, and verifying recovery.
The explanation got better. But I was still missing the one thing that mattered most: a company willing to connect Nerve to its actual production environment.
I had built a system that worked on simulated incidents. I hadn't proved that it would be useful enough for a real team to adopt, much less pay for.
Back to the codebase
And every time I ran into that uncertainty, I somehow ended up back in the codebase.
There was always another integration to build, another edge case to handle, another part of the infrastructure I could improve.
I knew exactly what to build next. I just didn't know who would use it.
Eventually, I decided to stop.
I still believe customer-level reliability is a real problem. I haven't concluded that Nerve was a bad idea. But I was spending too much time building around an assumption I hadn't properly tested.
What came after
After Nerve, I experimented with WishCodes, a consumer gifting product. That came with its own problems, particularly getting people to discover and use something I'd built.
Those experiences eventually led me to Distrarc, where I'm now working on helping startups find potential customers who are already discussing problems their products solve.
This time, I'm speaking to potential users much earlier. There's still a long way to go, but at least I'm getting feedback from people who might actually use the product.
I haven't ruled out returning to Nerve someday.
If I do, I'll start with one company, one real incident, and a clear answer to whether Nerve helped them catch something they would otherwise have missed.
I probably should have started there the first time.