On July 4, 2026, I reached a strange milestone: the system I was building had started helping build itself.
There was still babysitting. Plenty of it. But Ticinuum was reading tickets and doing work that improved Ticinuum. The loop I had been thinking about for months was starting to close.
That was a pretty fun moment.
It started with a side project, some frustration with Jira, and a question I couldn’t leave alone: how much of the software delivery lifecycle could agents handle if we gave them a clear contract for the work?
2025 · Ava
Back in 2025, I started experimenting with agentic systems and orchestration. I built a side project called Ava. It was fun, entertaining, and educational. It also suffered from a fairly basic problem: it was too generic.
I could make agents do interesting things. I hadn’t found a compelling enough reason for someone else to pay for those things.
Late 2025 · Tickets
Toward the end of that year, I started thinking about tickets.
Partly because I was fed up with Jira. I won’t pretend that wasn’t a factor.
But mostly because tickets sit at an interesting seam in how work gets done. An idea becomes a request. A request gets clarified, planned, implemented, tested, reviewed, and eventually delivered. Along the way, someone has to decide what matters, what’s allowed, what “done” means, and when a human needs to get involved.
A ticket could carry that contract.
That became the central idea:
Tickets as contracts for agentic work.
The contract could describe the intended outcome, the boundaries around the work, and the evidence needed to consider it complete. Agents could operate within those boundaries, with controls governing how far they could go and when they needed help.
And that could apply across the entire software development lifecycle: design, grooming, planning, estimation, development, testing, review, deployment, and support.
All of it.
That was the ambition, anyway. Getting there would require much more than connecting a model to a backlog.
One of the clearest lessons from experimenting with these systems was that you have to start with limited autonomy. You watch closely. You constrain what they can do. You review outcomes and intervene when something goes sideways.
Then you improve the system.
You strengthen the processes, tighten the controls, and make the outcomes more predictable. As particular kinds of work become dependable, you give the system more room to handle them.
Autonomy has to be earned.
That is what I mean when I talk about diminishing human oversight. The amount of supervision should change as the system demonstrates what it can reliably handle. Some work will continue to need close human judgment. Other work should eventually need much less attention.
The goal is to spend less time babysitting execution and more time deciding what is worth doing.
Late 2025 · The name
Toward the end of 2025, I bought the domain Ticinuum: Ticket + Continuum.
The name came from the loop I wanted to build. Agents would read tickets and carry out work. They would also identify problems, surface opportunities, and introduce new tickets. Humans would set intent, priorities, and boundaries, and make the decisions that required them.
Work would produce results, and those results would inform the next round of work.
June 2026 · The prototype
By June 2026, I was ready to put hand to keyboard and see whether the idea held up. I wrote a small prototype and started experimenting.
The initial results surprised me enough to keep going.
I also documented the approach in a defensive publication, Cost, Safety, and Durability Guardrails for a Ticket-Driven Autonomous Software-Delivery Control Plane, published on June 29, 2026. I wanted a public record of the mechanisms I was developing.
July 2026 → now
A few days later came that July 4 milestone: Ticinuum was helping build Ticinuum.
That didn’t mean I could walk away. It meant I had something concrete to work with. I could use the system, see where it struggled, improve it, and put those improvements back to work.
Since then, it has been months of nights, weekends, and obsessing over details that probably did not deserve quite as much of my attention as they received.
Some of them did.
You can follow the progression in the interactive roadmap.
There is still a lot I want to solve. The ambition remains an agentic system that can carry work through the entire delivery lifecycle, with governance and accountability built into how that work happens. There are plenty of hard problems left between here and everything I want it to become.
But I’ve reached the point where I’m willing to put it in front of other people and let them critique it.
Today
So, here it is: Ticinuum.
Right now, signups are for the early-access waitlist. I plan to open broader signups in the coming months. There will probably be a free tier so people can try it, test it, and decide whether it is useful.
But I want to build something worth paying for.
I’m interested in working with people and companies that have real problems to solve, are willing to give honest feedback, and will spend a few bucks on something that makes their work better. I like that relationship: you pay for the product, and I have to keep earning that payment by delivering value.
Ava gave me a place to experiment. Ticinuum gave those experiments a purpose.
Now I want to find out how useful it can be to someone besides me.
Take a look. Tell me what makes sense, what doesn’t, and where you’d want to put it to work. If you’d like to help shape what comes next, join the early-access waitlist.
— Troy Sampson