Ben Keilman · Boston, MA

I help teams adopt AI — and I build the tools myself.

Operations analyst by training. Self-taught AI builder by obsession. The rare combination behind every line of my resume — and the reason a rollout actually sticks.

Scroll

The problem everyone has

Companies are spending fortunes on AI. Most of it isn't getting used.

Buying the licenses was the easy part. The hard part is the people — getting a busy, skeptical, non-technical team to change how they actually work. That gap, between AI bought and AI used, is where the money quietly disappears.

What I actually do

Adoption is the deliverable. Installation was never the hard part.

I lead the human side — the trainings, the guides, the change management that move a tool from "we have access" to "we rely on it." And because I build the same kinds of systems myself, I do it as someone who has shipped, not someone reading from a script.

The unlikely part

I didn't come from computer science. I came from operations.

For nearly a decade I've worked inside a government agency, turning dense technical systems into the documentation and training thousands of staff actually use. Somewhere along the way I started building the tools too — teaching myself RAG systems, autonomous agents, and automation on Claude and the Anthropic API.

That path is the point. I've spent my whole career standing between complicated technology and the people who have to use it — which is exactly the job when the technology is AI.

Proof, at real scale

I've done this where mistakes get audited.

~0
staff I drive tool & process adoption across
0
page knowledge base I develop & maintain
0
days an agent I built ran autonomously in production
0
building on the Anthropic API, hands-on

Sole owner of enablement for an internal GenAI platform rolled out across the agency — guides, group trainings, 1:1s, road shows, train-the-trainer, and a community of practice. Adoption was the deliverable, not a side effect.

And I build the tools

Not slideware. Systems that run in production.

A self-healing error agent that reads its own code and fixes itself inside hard guardrails. A multi-source RAG over 11,000+ regulatory sections that answers plain-language questions with citations. A fleet of scheduled agents that share a workspace and accumulate knowledge over time. Real pilots, real freelance clients, real code.

See the work in detail →

Why it's rare

Most people do one of these. I do both.

The enabler

Runs the adoption — training, change management, the patient human work of getting people to actually use a new tool. Usually can't build.

The builder

Ships the AI systems — RAG, agents, automation. Usually can't (or won't) do the unglamorous work of driving adoption.

One person who runs the rollout and is a credible technical partner. Not a talker. Not just a coder.

And I can prove I move fast

I don't just say I'd get it moving. Here's exactly how.

The fear with any enablement hire is that nothing visible happens for months. So I wrote out my first 90 days — a plan built to land one measurable win in one team early, then scale a proven playbook instead of a hopeful one.

Read the 90-day plan →

Let's talk about what your team is trying to adopt.

Boston, MA · in-office or hybrid preferred, open to remote · government & regulated-industry native