Advisory · CI/CD and agentic engineering
Your pipeline is the bottleneck now. I ran engineering at the company that builds it.
Agents write code faster than your org can prove it works. The tooling vendors sell you the generation part. Nobody sells you the confidence. CI/CD is the single biggest bottleneck in engineering orgs today, and it's the one I know best: I was SVP Engineering at CircleCI. I help engineering leaders rebuild their delivery around it.
Book a conversation →The problem
Agents didn't break your delivery. They exposed it. I've watched a team go from 40 to 200 pull requests a week with the same eight engineers, and their deployments slowed down. A pipeline built for a few human changes a day, with slow builds, flaky tests, and a person at every gate, can't absorb agent-scale volume.
That isn't a tooling problem. It's a validation problem, and fixing it takes two things. Know what must be true to validate a code change with no human in the loop. Then make that thoroughness a property of the pipeline, not a phase staffed by people. After that, ship small chunks that are easy to validate, backed by a lot of production observation.
That's the work I do with teams.
What you get
Fixed fee
AI-delivery diagnostic
Four weeks inside your pipeline. I find where it breaks under agent-scale volume and hand you a plan to fix it.
Build times, flaky tests, manual gates, and review queues
Where a human is deciding something the pipeline should
A remediation plan with owners and sequencing
Scope set on the first call
Some clients run the remediation themselves. Most roll it into a retainer.
$5K–$10K / month
Advisory retainer
Fractional engineering leadership, working directly with your CTO or VP Engineering.
Find where your CI/CD breaks under agentic engineering
Redesign the pipeline so validation scales with generation
Make sure the org actually changes, because the tooling alone won't
Weekly working sessions, async access between them
If the problem is already clear, we can start here.
Who it's for
Engineering leaders at scaling companies where coding agents landed but delivery didn't speed up.
Teams where the pipeline is the constraint: stacked PRs, slow and flaky builds, long cycle times, engineers waiting on a human for something CI/CD should decide.
CTOs and VPs who need someone who's run this at scale, not a strategy deck.
Who it's not for
Teams that need more code written. You don't have a generation problem.
Orgs looking for an interim manager to sit in the chair. I deliver outcomes, not attendance.
Anyone shopping for the cheapest option. This is senior judgment, priced like it.
Why me
I was SVP Engineering at CircleCI, where CI/CD wasn't a platform concern, it was the product. That meant seeing from the inside how engineering orgs actually ship: where builds queue, where tests flake, where people become the gate. Few people have seen more pipelines up close.
Agentic engineering is that same problem at a volume no human review queue survives. I'm SVP Engineering at Ada now, and alongside that role I take a small number of engagements.
Where does the time come from? The same place I'll tell you to find yours. Agents write the code, the pipeline decides what ships, and my hours go to the judgment calls nothing else can make. I run the thesis on myself first.
The last few years I've been writing about what changes when code gets cheap: review as the constraint, validation as a property of the pipeline, small batches or nothing else matters. I only take engagements where that thesis applies.
The thinking behind this
It's all public. Read it before we talk, and you'll know whether we'd agree.
Questions
What does engineering advisory with tinycto cost?
The advisory retainer is $5K to $10K a month. The AI-delivery diagnostic is a fixed fee, scoped on the first call.
Why is CI/CD the bottleneck for agentic engineering?
Agents write code faster than most pipelines can prove it works. A pipeline built for a few human pull requests a day, with slow builds, flaky tests, and a person at every gate, can't absorb agent-scale volume, so cycle time goes up, not down. It's the single biggest bottleneck in engineering orgs today.
What is the AI-delivery diagnostic?
Four weeks inside your pipeline: build times, flaky tests, manual gates, and review queues, and every place a human decides something the pipeline should. It ends with a remediation plan with owners and sequencing. Some clients run the remediation themselves; most roll it into a retainer.
What does JP LeBlanc's CircleCI experience bring?
I was SVP Engineering at CircleCI, where CI/CD was the product. That meant seeing from the inside how engineering orgs actually ship: where builds queue, where tests flake, and where people become the gate. Agentic engineering is that same problem at a volume no human review queue survives.
Is this an interim or fractional CTO role?
It's fractional engineering leadership, not an interim manager in the chair. I work directly with your CTO or VP Engineering in weekly working sessions, with async access between them.
How does JP LeBlanc fit this alongside a full-time role?
I'm SVP Engineering at Ada and take a small number of engagements alongside that role. Agents write the code, the pipeline decides what ships, and my hours go to the judgment calls nothing else can make.
How does an engagement start?
With one conversation. You describe your delivery pain and get a straight answer on whether I can move the needle. If the diagnostic fits, we scope it on the call; if a retainer fits, we start there; if neither fits, you get the honest answer anyway.
How it starts
One conversation. You describe your delivery pain, and I tell you straight whether I can move the needle and how. If the diagnostic fits, we scope it on the call. If a retainer fits, we start there. If neither fits, you get the honest answer anyway.
Book a conversation →