AI-pipeline consulting — games · creative teams · SMB
The AI pipeline your artists will actually use.
I’m a sixteen-year AAA production artist who builds AI tooling from the artist’s side of the desk — including a platform my own team used every day and other teams picked up on their own. I find where AI fits your pipeline, then build it. On your hardware, if your IP demands it.
The receipts
Built one. Shipped it. Watched it spread.
Inside a AAA studio I designed and built an internal AI concept platform for my own art team. No mandate, no rollout program — it spread because it made the work better.
- 5 of 6
- artists used it daily, unprompted
- 1–2 → 15–20
- same artist, same timebox
- “pick a favorite”
- up from “fix the comp”
- Zero
- other teams adopted it on their own
Engagements
Start small. Scale on evidence.
AI Readiness Audit
Two weeks inside your pipeline. You get a report on where AI actually fits, where it doesn’t, and what it will cost — whether or not you hire me to build it.
Pilot Build
One working tool aimed at one real bottleneck, shipped into production hands — the fastest way to find out what adoption looks like in your shop.
Pipeline Build
The full system: tooling, integration, training, and the adoption work that makes it stick. Scoped from the audit, never before it.
Fractional AI Director
Ongoing direction for teams that need the judgment without the full-time hire. Roadmap, vendor calls, guardrails, hands on the tools.
Managed Agent
For smaller operations: an AI “digital employee” installed against a real workflow — set up, monitored, and maintained so you never have to think about it.
Every engagement starts with the same thirty minutes.Book the call
The part legal will ask about
Nothing has to leave your building.
At most companies, AI adoption doesn’t die with the users — it dies in security review. I build pipelines on open-weight models running on your own hardware when the IP calls for it: modern image and language models on a single workstation GPU, measured, in production shape. “The data never left our network” ends a lot of conversations that “trust the vendor” cannot.
14B
65 tok/s
$0
0 bytes
How it works
Three steps. No mystery.
The call
Thirty minutes. You describe how work moves through your team; I tell you honestly whether AI helps — including when the answer is “not yet.”
The audit
I embed in the actual pipeline — files, handoffs, review loops — and map where hours are actually lost. No survey decks. Working sessions.
The roadmap
A written plan ranked by payoff: what to automate, what to leave human, what it costs. The report is yours. Build with me, or don’t.
Artist first.
That’s the point.
Sixteen years shipping AAA game art — franchise work at EA scale — before stepping into leadership of a concept team and teaching myself to build the tools we needed. Most AI consultants have never sat through an art review. I’ve run them.
That matters because pipeline tools fail socially before they fail technically: artists can smell a tool built by someone who doesn’t respect the craft. I’m the artist in the room — the tooling I build keeps every creative call human.
I also run BentBoneLab, a small Blender-tools label — I ship and support my own commercial software.
Next step