Glossary
Fractional AI lead
A senior AI leader you share with other companies, part time. A fractional AI lead sets your AI direction, runs governance and enablement, and builds what the plan needs, month to month. It starts in weeks, without an executive search.
Fractional chief AI officer (CAIO)
Another name for a fractional AI lead, borrowed from the full-time executive title. Some companies use it when the role reports to the CEO and owns AI policy. In practice the work is the same: direction, governance, enablement and, with the right person, the builds.
Forward-deployed engineering
A way of building where the engineer works inside the customer's team instead of from a spec. They sit in the team's tools and meetings, find the real problem in the day-to-day work, build the fix in small tested pieces, and hand it off so the team runs it.
Forward-deployed engineer
An engineer who works forward-deployed: embedded with the team that has the problem, close enough to see it as they live it. The term comes from software companies that send engineers to customers' sites. In my practice it's also an offer for internal initiatives.
AI dial
My shorthand for deciding how much of a task to hand to AI. It isn't on or off: for each task you choose a first draft, a check, or a run that a person approves, and turn it up as trust builds. You own the output at every setting.
AI governance
The rules for how a company uses AI: who can build what, what data can go into which tools, what reaches a customer, and where a person signs off. Good governance is short enough that people follow it, and it's set before the first build ships.
Shadow AI
AI tools people use at work without anyone approving them, often with company or customer data pasted in. It usually means people want help the company hasn't given them yet. The fix is a short list of approved tools and rules, not a ban.
AI office hours
A standing weekly session where a team brings AI questions, half-built prompts and stuck workflows to someone who can answer them on the spot. It's how enablement keeps going after training day, and it surfaces the next thing worth building.
Person checkpoint
A step in an AI workflow where a named person reviews and decides before anything moves on, such as approving a draft, a recommendation or a change. The AI does the slow part; the person makes the call, and both are on record.
Build vs buy
The decision between building a tool yourself and buying one from a vendor. For AI tools it has a third option, waiting for a platform you already pay for to ship it. The answer turns on how specific the job is, the data it needs, and who'll own it.
Internal tool
Software a company builds for its own team, not for customers: a workflow, a dashboard, an assistant that drafts or checks work. With AI, many internal tools now cost less to build than the hire they replace or the vendor contract they'd otherwise need.
Scope sprint
A short, fixed period at the start of a build, usually a week, to agree what the product does, who uses it, what done looks like and what's out. In the Roadmap Accelerator it's week one of thirteen.
Prototype to production
The work between something that runs in a demo and something a team relies on every day: real data, permissions, error handling, tests, monitoring and a person who owns it. It's where internal AI projects tend to stall.
Handoff package
What a team gets at the end of a build so it can run without the builder: the code in their own account, a spec, a risk list of changes that would break it, docs in their language, and training with the people who'll run it.
Bus factor
How many people would have to leave before a system stops working because nobody else understands it. A bus factor of one is how a lot of internal tools quietly die. A handoff package exists to raise it.
Build and Share
A Roadmap Accelerator option for products that will earn money: a lower upfront fee plus a share of the product's revenue for a set period. The client still owns the product.
Solutions architecture
Deciding how systems, data and automations fit together before anyone builds, including where the truth lives and which system wins a conflict. It's what keeps an AI build from breaking the first time the business changes.
Parent theme
In HubSpot's CMS, a theme that holds the shared modules, templates and styles. Child themes inherit from it, so fixes and new features in the parent reach every site built on it. Geekify is my parent theme.
Child theme
A HubSpot theme that inherits everything from a parent theme and holds only a brand's own colors, fonts and changes. The parent keeps getting updates and the site keeps its look.
Emerjent
The AI operations platform I built and run my practice on. It turns an agency's way of working into the steps every project follows, with AI doing the busywork and people making the calls.