Dan Garraway

Everyone has an AI that can do the work.Almost nobody is using it.

That is not a model problem. It is an interface problem.

Applied-AI product lead · Co-founded WIREWAX, sold to Vimeo

What I ship

I design the surface where an agent meets a person. When it speaks, what it shows, what it asks you for, what it just gets on with. I have had an agent running my own work since 2025, and what makes it work is not the model. It is the decisions about when it is allowed to interrupt me.

Since 2025

an agent running my own operation, unsupervised, through several generations. See it running

91%

of Vimeo's acquired enterprise book migrated. 200+ go-to-market staff enabled.

$5M

new product category in year one, about 3x plan.

BAFTA2 Webbys2 Clios

for designing the surface between people and machine-understood media.


Three ways in

Engagements

Start here

Agent Build

For
a team drowning in one workflow
You get
that workflow running as a production agent, with evals and a runbook, handed over
Shape
4 to 6 weeks, fixed scope

from$18,000

Scope an Agent Build

Embedded Product

For
a company shipping an AI product with no senior owner
You get
an AI feature or product from scope to shipped, embedded in their team
Shape
retained, 10 to 20 hrs/wk, 3 months minimum

from$14,000 / month

Talk about a retainer

Operating Model

For
an exec team whose decisions are the bottleneck
You get
decision and communication protocols, AI configured around the team, a rhythm that holds
Shape
4 weeks, then optional retainer

from$18,000

Talk about an operating model

One or two engagements at a time. US and UK hours.


Why

Change is my constant.

We've accelerated intelligence.

We haven't evolved how we work with it.

You can build almost anything. Deploy agents. Connect MCPs. Automate workflows.

The technology is no longer the hard part.

Whether anyone uses it is.

That gets decided at the interface.

Built AI and interactive products trusted by teams at Nike, Disney, BBC and Vimeo. Recognized with five international industry awards.

  1. 2005

    Video leaves television.

    Built one of the UK's earliest video-first web products, exploring streaming on the web years before online video became mainstream.

    Question Can video become an internet-native experience?

  2. 2009

    Video becomes interactive.

    Co-founded WIREWAX and built a browser-based video motion-tracking engine, connecting what a viewer wanted to products, information and action.

    Question Can video become a two-way medium?

  3. 2012

    Machines begin to understand video.

    Built production computer vision that tracked faces and objects frame by frame, with confidence scoring and human review, years before generative AI.

    Question Can machines understand visual context?

  4. 2015

    AI becomes a product.

    Took WIREWAX from bespoke innovation to a scalable SaaS platform, proving computer vision could become repeatable commercial software.

    Question Can emerging technology become a business?

  5. 2018

    AI enters real work.

    Built AI services that processed media at production scale, saving thousands of hours of manual work and proving AI's value inside enterprise workflows.

    Question Where does AI belong inside real organizations?

  6. 2021

    Intelligence becomes infrastructure.

    Joined Vimeo after the acquisition. 91% of the acquired enterprise book migrated, 200+ go-to-market staff enabled, a $5M new category in year one.

    Question How do large teams adopt new technology?

  7. 2026

    AI is everywhere.

    Everyone wants teams to use it.

    But, for what?

    Today the challenge is designing how humans and agents think, decide and operate together, so the capable system actually gets used.


Questions

Isn't this just AI consulting?

No. I ship. The deliverable is a system in production that your team uses, not a recommendation about one.

Why not hire an engineer?

Hire one. I work with them. Wiring the agent is the easy part now; deciding when it speaks, what it shows and what it does silently is what decides whether anyone uses it, and that is the part I do.

What do I actually get?

Agent Build: a working agent for one workflow, evals, a runbook, handed over in 4 to 6 weeks. Embedded: a senior product lead inside your team, 10 to 20 hours a week, shipping. Operating Model: how your exec team decides, with AI configured around it.

Who is this for?

Founders and operators at companies that have decided AI matters and are stuck turning that into something shipped. Usually Series A to C, usually one specific workflow that is eating the team.

Book 30 minutes


I take one or two engagements at a time.

If you were sent here by someone, tell me who. It helps.

dangarraway@gmail.com · linkedin.com/in/dangarraway