Field notes · August 2, 2026 · 2 min read
Why Voco stopped starting with software
We originally started Voco with a classic software pattern: have an idea, build a SaaS platform, and find users who want to buy it. We built prototypes aimed squarely at recruiting teams, attempting to solve the painful candidate-rediscovery problem.
The software worked, and the recruiting problem was real. But as we put it in front of operators, we noticed a persistent kind of customer pull. Teams didn't just want another platform to log into, and they didn't want to bend their existing operations to fit our data models.
Instead, they kept pointing to the messy, manual, review-heavy workflows they were already doing — the ones that required human judgment and were expensive when delayed or wrong — and asked: "Can you just build something that fixes this?"
That pull taught us a few things that shifted our entire model:
Recurring operational workflows are the right starting point. The most expensive problems in a business aren't usually solved by adopting a new vertical SaaS. They are solved by looking at a specific, repeatable workflow where people lose hours every week to manual data entry, formatting, and drafting.
Service lets us learn before productizing. Instead of building a platform in a vacuum, we now use a service-led motion to learn how a workflow actually operates inside a real team. We diagnose the friction, build a Minimum Evolvable Product (MEP) around that specific workflow, and prove it works before expanding. We don't assume we know what the software should be until the workflow teaches us.
Human judgment is a feature, not a bug. A lot of AI development right now is obsessed with autonomous agents that run entirely on their own. But in regulated or high-stakes operational environments, taking humans out of the loop is a non-starter. Our systems are built to do the repetitive drafting and sorting, but they purposefully keep necessary human approval required at the end of the line.
One strong signal is not proven ROI. We are looking at strong signals from this new approach, but we are being careful not to confuse a good direction with a proven finish line. We expand only from real, measured evidence. We are looking for more operational teams with painful, manual workflows to partner with, so we can keep proving this model out.
We aren't a recruiter SaaS company anymore. We build AI systems around the workflows your team cannot afford to keep doing manually.