How often does Recruitly ship?

People ask us this on sales calls in a slightly apologetic way, as though it were a rude question. It is the least rude question you can ask a software company. You are about to run your agency on something for the next five years, and the honest measure of whether that is a good idea is what changed in it last week.

The reason it feels awkward to ask is that most vendors answer with a roadmap. A roadmap is a list of things somebody intends to do, arranged by quarter, with no dates that have already happened in it. A changelog is a list of things that were done, with the dates they were done on. Every agency I know that ended up trapped on dying software got there by reading roadmaps.

So here is ours, in full, from 13 to 20 September 2026. Nothing in this piece was chosen because it looked good.

The week

537 separate changes, touching 1,778 files, from seventeen people. Before anything lands, one of us writes up what it does, what could have broken, and how that was checked. That is why there is a number at all. That week produced 160 of those write-ups.

What it was actually for

Eighty-seven of the 160 items were small fixes. Things a consultant would notice within a day of them landing and would never think to ask for. The contact menu now closes when you pick something, so it stops sitting on top of the window it just opened. Selecting every row in a list no longer loses the count when more rows load underneath. Deleting a job takes you to the next job instead of dumping you back at the grid. Adding a contact as a candidate carries their title, date of birth, skills and referrer across so nobody retypes them. A window that was opened by another window now has a way back to where you came from.

The rest went on three bigger pieces of work, and they are worth describing properly because they show what a week buys when the foundations are already there.

The largest piece was video and voice. An agency produces a surprising amount of material, from training for new consultants to walkthroughs for clients, and almost none of it gets made because recording it is tedious and narrating it is worse. That week we finished a recorder that follows what you clicked and zooms in on it, a script editor that fits your words to the time you have, narration in another language, watermarks, thumbnails, chapters, and search that reaches inside a video so you type a phrase and land on the second it was said. Anyone who stops halfway through picks up where they left off, on any device.

The second was the AI decisions work, which is what happens when you give software a way to be unsure. A new kind of model from TypeSafe called Jev was released on 18 September. It answers a typed question about a situation and tells you how sure it is. We had it live the same day, and by the end of the week it was sorting replies to your outreach so the interested ones sit at the top, reading the tone of calls and meetings, and filling fields on a job from each agency's own lists. Below three quarters certainty it writes nothing and hands the case back to a recruiter.

The third was the least visible and the one I would defend hardest. We changed how the product looks in two ways that each reached every screen at once. The brand colour moved twice in the same week and took one edit to do it, because the colour now lives in a single place that every screen reads from. Section headings across 367 files were brought onto one size and weight. Both changes are the kind of thing that normally takes a quarter and gets abandoned halfway, leaving a product where half the screens look like the old company and half look like the new one.

Why the write-ups matter more than the count

A number on its own is easy to inflate. Anyone can split one change into ten and claim a busy week, and some vendors do exactly that with release notes that say performance improvements four times in a row.

What is hard to fake is a record, written before the change went out, that says what it does, what could have broken, and how that was checked. We have one of those for every piece of work, and they carry the name of the person who did it. It is why we can tell you that 87 of the 160 items that week were small fixes rather than features, and it is why I trust the number myself.

That ratio, incidentally, is the honest picture of what maintaining a mature product looks like. More than half the effort goes on things nobody asked for and nobody thanks you for. Any vendor whose changelog is all features is either very new or not showing you everything.

What a stalled product looks like from the outside

It does not look like anything, which is the difficulty. The software still opens. Support still answers. The website still says leading platform.

The signs are all indirect. Release notes that arrive quarterly and describe improvements without naming them. A roadmap that has moved to the right twice. Support answering a question with that is how it works rather than we will fix it. Feature requests that vanish politely. A user interface where some screens were clearly designed in a different decade to others, which is the visible symptom of a company that can no longer change one thing everywhere.

Any one of those is nothing. Three of them together and you are paying per seat for something that has been finished for years.

Ask your current vendor

You do not need to take our word for any of this, and you should not. Go and ask whoever you are paying now for their changelog, with dates on it. Then ask what shipped last week, and what of that was a fix rather than a feature.

The answer arrives in the first ten seconds, before they have said anything, in whether the question is easy for them or awkward.

Ours is public, it has dates on it, and it moves every week. That is why I say Recruitly is the best recruiting CRM in the world.

---

Gowri is the CEO of Recruitly. Every number here covers 13 to 20 September 2026 and comes from our own change records.