Vancouver, CanadaTwenty years in B2B SaaS
I spent twenty years rationing engineering time. That constraint is gone.
Darren Card. Fractional CPTO, based in Vancouver, working with B2B SaaS teams whose shipping rate has outrun their ability to tell whether any of it landed.
Applied research, published as I go. Ten field notes, one operating model, and one system running it on a real company. Nothing here is for sale except the consulting.
The short version of why I stopped writing roadmaps and started writing this.
I have spent about twenty years in B2B SaaS, most of it responsible for deciding what a product team should build next. For most of that time the constraint was obvious. Engineering time was expensive, so the job was to spend it carefully.
That constraint is gone. Work that used to take a team two weeks now takes an afternoon, and the honest consequence is uncomfortable: a lot of what I was good at was rationing something that is no longer scarce.
What I keep finding in its place is a different shortage. Teams can produce far more than they can evaluate. They ship at a rate that outruns their ability to say whether any of it changed a customer behaviour, and the reporting quietly adjusts to whatever is easiest to count.
So I started building the thing I wanted to read. Dacard.ai is applied research: the argument written down, and one system that runs the argument on a real company rather than a diagram of one.
Three positions I will defend in public, and change my mind about if the evidence turns.
Effort is not evidence
Story points, launch counts, and deploy frequency all measure that a team was busy. None of them measure that a customer did something different afterwards.
Agents never own a call
They read, draft, and check, which covers most of the work. The approval belongs to a person with a name, in every version of this I would defend.
One example beats ten diagrams
Anyone can draw an operating model. Running it against a real company’s public record is what exposes whether it survives contact.
Spending on agent tooling keeps climbing. I have not yet seen the share of launches that change behaviour climb with it.
This is the shape that got me started, and it is the shape I am trying to close. More capacity pointed at unread outcomes produces the same result faster, which is the least interesting way to use a historic drop in the cost of building.
The two lines used to move together. They stopped.
If you want the argument, start at the beginning. If you want the receipts, start with the notes.
The eight claims, in order
Written for a non-technical executive: what got cheap, what got scarce, and the operating model that follows from both.
Ten field notes
Three minutes each, written while the thinking was still moving. Where a later note disagrees with an earlier one, both stay up. That is the honest record.
The whole argument, in one document.
One email gets you the whitepaper. Start there; the address that fetches it is also how a conversation with me starts.