◆ Boutique AI consulting · Education technology

Make AI that earns a place in the curriculum.

Devfire helps EdTech companies modernize technology, strengthen engineering organizations, adopt AI-first practices, and ship AI features that teachers trust and students can rely on, minus the hype.

Start a conversation See how we work

Taking on two new partners this quarter.

01 The gap

Most EdTech AI fails on trust, not capability.

A tutor that confidently hallucinates in front of a thirteen-year-old. A copilot that ignores the curriculum it's meant to support. A dashboard no teacher asked for, grading them on metrics they never chose. The model is rarely the problem.

The problem is everything around the model: the technology it has to integrate with. The data and context it depends on. The retrieval, evaluation, guardrails, observability, and APIs that make it reliable. The engineering practices required to ship it predictably. And the people, tooling, and operating model needed to keep improving it once it's in production.

That's why becoming AI-first is bigger than choosing a model or buying a coding assistant. It can mean modernizing existing platforms, introducing AI-first engineering practices, measuring the impact of AI tooling, developing new approaches to hiring and technical assessment, and creating shared AI capabilities that can be reused across products instead of rebuilt feature by feature.

The gap between what a model can do and what a classroom will accept is. That gap is where we work.

02 What we do

Four ways we help.

  1. 01

    Strategy, readiness, & roadmapping

    We assess the technology, engineering team, delivery model, and AI readiness — then turn what we find into a practical roadmap for modernization, AI adoption, and predictable delivery.

  2. 02

    Responsible AI & evaluation

    We build the systems around the model: retrieval, evals, guardrails, observability, and the engineering practices that make AI reliable enough to trust in production.

  3. 03

    Design & prototyping

    We take AI ideas from zero to one — shaping the use case, building working prototypes, and designing reusable capabilities that can scale beyond a single feature.

  4. 04

    Implementation & enablement

    We work alongside your team to modernize the stack, ship production AI, adopt AI-first coding practices, improve hiring, and increase engineering capacity without simply adding headcount.

03 How we work

Fixed scope. Senior people. No body-shopping.

We work in clear phases with a small team. The people who scope the work are the people who do the work.

Diagnose 2–3 weeks

We embed with your team, audit your data and product, and find where AI earns its keep — and where it doesn't. You get a sharp, honest recommendation, not a deck.

Build 6–12 weeks

We design, prototype, and ship alongside your engineers. Evals first, features second. Every release gets more defensible than the last.

Hand off Ongoing

Documentation, runbooks, and a team that owns what we built. We stay on call, but we're never the dependency.

04 Selected work

Outcomes, anonymized.

18% → <2%

Factual hallucinations in a K-12 math tutor, cut with curriculum-grounded retrieval and a continuous eval harness.

~4 hrs / week

Returned to teachers in a three-district pilot of an AI feedback assistant built around their actual workflow.

Zero records

Student records exposed to model providers, after we built a FERPA-aligned data pipeline for an adaptive-learning platform.

Representative engagements, anonymized at our partners' request. We're glad to discuss specifics under NDA.

05 About

Between two worlds that rarely meet.

Devfire was founded on a simple frustration: too much AI in education is built by people who've never taught, and too much education software is built by people who've never shipped AI. We sit between those worlds — senior practitioners in machine learning, product, and learning science who've shipped production systems and spent real time in classrooms.

We stay small on purpose. It keeps the work senior, the communication direct, and the accountability absolute. When you hire Devfire, you get Devfire.

06 Meet the team

The people on the other end of the email.

Devfire is a handful of senior practitioners. No account managers, no junior staff learning on your dollar — the people who scope the work are the ones who do it.

Holly Garner

Holly Garner

CEO

Holly is CEO of Devfire. Her career has run through the middle of education technology — Pearson, ACT, Kaplan, Junior Achievement — the place where learning science, assessment, and real product decisions have to meet. She's spent it translating research and data into products people actually use, and decisions they can defend.

She also founded and exited an AI company built on automated scoring, and her name is on several patents in that space — so the "AI in education" conversation is one she's been in since before it was a marketing category.

She holds an M.Ed. and an MBA, did Ph.D. coursework in educational measurement and statistics, and sits on several EdTech boards. She started college on a Pell Grant, which is a polite way of saying she knows exactly what's at stake when education technology gets it right — and when it doesn't.

Adam Burke

Adam Burke

CTO

Adam is a technology and engineering executive with more than 20 years of experience building, scaling, and modernizing SaaS businesses. He has worked across 35+ SaaS companies, including more than a dozen in education and learning technology, assessing technology, engineering teams, delivery practices, and organizational risk.

His recent work has focused on making engineering AI-first — modernizing existing platforms, taking AI features from zero to one, creating reusable AI capabilities, rolling out and measuring AI development tools, and rethinking coding practices and technical hiring around the leverage AI creates.

Previously, Adam served as VP of Engineering for learning technology businesses within LTG and as a portfolio VP of Engineering at Cordance. At Devfire, he focuses on the technology, people, and engineering systems around the model that turn AI capability into reliable, predictable delivery.

07 Contact

Tell us what you're building.

Whether you're scoping your first AI feature or rescuing one that went sideways, we'd like to hear about it.

We read every message ourselves. Expect a thoughtful reply within two business days.