The Release System

In the spring of 2026, Spare's release was approaching and the plan was the traditional one: an email campaign, plus Partner Service Managers walking clients through enablement one conversation at a time. For anything to get turned on, a client had to notice an email, connect it to their own operation, and reach out to their PSM.

Stef Lash, the product manager I worked with, and I had been circling a different idea: let customers discover features inside the platform itself, turn on the straightforward ones directly, and request enablement for anything that needed configuration from our team. About a week before launch, we decided to stop talking about it and build it.

We had one thing going for us: a mapping of every feature flag attached to the spring release. I designed and built the front end: an eye-catching, temporary advert in the platform that took users to a new features page, where each feature had a short walkthrough video, a plain explanation of what it did, and a request-to-enable button. A platform engineer built the backend and the Slack connection: a request fired a notification, and a bot tagged the client's assigned PSM to take the conversation from there.

While building, I went back and forth with the PSMs: what information did a request need to carry, and what could they reasonably expect a client to know, before they'd feel confident enabling a feature. Their answers shaped the page. The whole thing went from decision to production, PRs, QA, review, and deploy, inside the week.

It outperformed every other channel. About 80% of the interest we saw in the spring features came through the in-product flow and its Slack pipeline; the email campaign and social channels together accounted for the rest.

When the summer release came around, we upgraded the system instead of rebuilding it. The advert got a summer refresh, and instead of landing on a static page, it opened a conversation with the Spare AI Assistant, which walked users through the summer features, helped them figure out which ones fit their organization, and helped them turn features on.

That did two jobs at once: self-serve enablement, and getting more customers comfortable using the assistant itself, which we wanted on its own merits. We also piloted new ways of announcing the release inside the product, through the notification bell and a toast.

Assistant usage roughly doubled among users who went through the summer release experience, and the release drove a clear spike of interest in the summer features.

The bigger result is the system itself. Releases at Spare now have an owned, in-product channel: customers discover features where they already work, act on them immediately, and the right person gets pulled in only when configuration actually needs a human.

The spring sprint was scrappy on purpose: a temporary advert, a simple page, a Slack bot. It settled a debate in one release cycle that could have gone on for quarters, and it made the summer investment, wiring the release into the AI assistant, an easy call. The fastest way to find out whether a channel works is to ship the smallest real version of it.