Fractional CTO / CPO / tech-product builder

What it is. Embedded technical leadership at one to two days per week. Strategy, architecture, key hires, and the trade-off calls that need a second senior brain in the room. The seat above the IC layer where the consequential decisions get made — and where coding agents extend that seat into selective hands-on: codebase audits, architectural scaffolding, course corrections.

When it fits. You have a product in market, a team between three and fifteen, and you need leadership-tier presence without committing to a full-time hire. Or you're between CTOs and the work cannot wait.

When it doesn't. You need someone writing code five days a week. You want a full CTO replacement rather than a CTO peer in the room. The team has no leadership-tier vacuum to fill.

First two weeks. Reading the system. Talking to engineers and product. Producing a written read on what's working, what's at risk, and the three things that will compound over the next quarter if we get them right.

Cadence. Three months minimum, ongoing.

Technical advisor

What it is. A lighter version of the fractional CTO seat — a couple of hours per month, four hours per week maximum. Architecture and product judgment, direct discussions with the founders, a sounding board on the consequential calls. Coding agents make limited expert-level hands-on possible at this commitment level — codebase audits, architectural scaffolding, quick course corrections — kept minimal. The second senior brain in the room when one is needed.

When it fits. You already have one or two people building, and you want an outside read to make sure the foundations are right. No over-engineering, no premature abstractions, the right calls on the architecture and the product as it takes shape. Direct collaboration with the founders matters — the discussions about what you're building are the work. It's still too early to commit beyond a few hours a month.

When it doesn't. Things have gotten complicated enough that the org needs embedded leadership — that's the fractional CTO shape. You want someone to actually build it for you — that's the founder sprint. You don't have clarity yet and want a consultant to do business strategy on your behalf. You don't want discussions, you want a deliverable — product needs discussions, and if you don't want them, this shape isn't it.

First month. Reading what's being built, talking to the people building it, surfacing the foundations that will compound or break. By the end of the first month, a written read on what's solid, what's drifting, and where my time has the most leverage going forward.

Cadence. One month minimum. A couple of hours per month, four hours per week maximum.

Complex project lead

What it is. Owning a stuck or complex initiative inside a bigger client, end-to-end, as a senior freelance — strategy plus operational plus tactical, in the same person. Migrations, transformations, transitions, cross-functional unblocking. The work that fails when you split it across a consultant for the strategy and an IC for the execution.

When it fits. Bigger clients where the project crosses teams or stacks, and the actual blockers are people, decisions, and dependencies — only secondarily technical. Someone has to read the system, sequence the work, and move the org around it.

When it doesn't. The project is straightforward execution and you already have an internal owner. Or it's pure deliverable consulting with no operational surface — a deck, a recommendation, no movement. Or you're a founder building from zero — that's the sprint shape, not this one.

First two weeks. Mapping the moving pieces and the actual blockers. Producing a sequencing plan, a written commitment of what gets shipped in the first month, and the surface that lets the rest of the team see the work happening.

Cadence. Three to six months, full-time as a freelance.

Founder / builder sprint

What it is. A packaged, zero-to-one build with a founder or early-stage team. AI-native by design — agents do the bulk of the execution under my judgment. Strategy, product direction, and a clean handoff that comes with everything: the working product, the decisions documented, the harness left behind, and the founder briefed to operate it.

When it fits. You're a founder or early-stage team. You don't yet know what to build, or you know but don't have the muscle to build it well the first time. The bottleneck is judgment, not headcount. You'd rather pay for judgment up front than for months of compounding waste from a wrong build.

When it doesn't. You don't want AI coding agents doing the build. If you want a traditional dev shop, hand-coded everything, or a contractor who treats agents as forbidden — this isn't the right shape. Also doesn't fit if you already have a strong technical co-founder, or if you're a bigger client looking for cross-team execution — that's the project-lead shape, not this one.

First two weeks. Founder discovery, narrowing scope, first executable version. By end of week two you have a working prototype and a clear go / no-go for the next phase.

Cadence. One to three months. Not full-time. Packaged by deliverable, immediate presence.

AI-agent training

What it is. Hands-on training to bring a founder, a technical operator, or a technical team to top-tier proficiency with Claude, Claude Code, and the surrounding agent harness. Not a slide deck — the actual harness, set up live, in your own workflow.

When it fits. You want to stop being blocked on technical work, or you want your team to compress months of learning into days. You're serious about commanding agents yourself rather than outsourcing the question of what they can do.

When it doesn't. You want a reusable curriculum to deliver later without me. You want a recorded course, a kit, a slide deck. The training is custom and live; what stays with you afterward is your own working repo and notes, not a redistributable package.

First session. You leave with a working harness in your own repo, the muscle memory to use it, and a written reference for the parts that take time to internalize.

Cadence. Half a day to two days, depending on depth.