AI setup, built for game studios.
We help Unreal Engine teams adopt Claude Code and AI workflows in 4–6 weeks, with guardrails for C++, Perforce, and proprietary IP.
Privacy-first · ROI-tracked · Fixed-scope packages
The bottleneck
Your team wants to use AI. Setup is the bottleneck.
Most studios we talk to face the same friction. None of it is technical genius, but it eats months of senior time when nobody owns the rollout.
Reviews drowning in style nitpicks
Senior engineers spend hours flagging the same convention violations. AI output makes it worse without proper guardrails.
Proprietary IP risk
Devs paste engine code into personal AI accounts. Your code may be retained for 5 years and used for training. Most studios don't even know this is happening.
No Perforce, no integration
AI tools assume Git. Your team uses Perforce or Plastic. The "AI workflow" everyone talks about doesn't apply out of the box.
Costs spiral with no visibility
Token spend with no per-team budgets, no monitoring, no controls. Some studios have spent thousands before noticing.
Senior devs resist, juniors over-rely
Without a playbook, adoption fractures. Half the team avoids it; the other half ships AI-generated bugs to main.
No way to prove ROI
Leadership wants numbers. "It feels faster" doesn't get budget renewed.
Services
Four ways we work with your team.
Most clients start with an Audit, move into Setup, then keep us on a monthly Retainer. The MCP Perforce license is available standalone.
AI Readiness Audit
Map of current workflows, friction points, privacy tier recommendation, phased adoption roadmap, ROI estimate, and Shadow AI detection (devs using personal accounts on company code).
Start with an AuditSetup & Integration
Hierarchical CLAUDE.md adapted to your codebase. Custom skills per domain (gameplay, networking, UI, shaders). MCP Perforce configured. Permission settings (engine off-limits, protected paths). Onboarding guide. Cost-monitoring dashboard. Lead handover session.
Plan a SetupTeam Training
Prompting patterns for performance-critical C++. Code review of AI output (red flags, anti-patterns). Safe workflows. Hands-on sessions on your real codebase.
Train your teamOptimization Retainer
Skills and MCPs evolve with your codebase. Office hours for devs (4–8h/month). Code review of AI-related PRs. Integration of new Claude features as they ship. Quarterly ROI report.
Stay supportedAlso available: MCP Perforce License. €3,000/year up to 30 seats, +€80/seat above. Drop-in tool that lets Claude Code create, shelve, and submit Perforce changelists with your studio's conventions. Sold standalone or bundled with Setup.
Process
From first call to shipping AI workflows in 6 weeks.
A predictable process. No open-ended retainers. No "we'll figure it out."
- STEP 01 30 min · free
Discovery Call
We confirm fit, your stack, your constraints, and which tier of privacy you need. If we're not the right call, we'll say so.
- STEP 02 asynchronous
Pre-Audit Questionnaire
30 structured questions sent before any work begins. You fill it in; we arrive prepared. Most teams find pain points they hadn't articulated.
- STEP 03 1–2 weeks
Audit
Tech Lead interview, codebase walkthrough, workflow mapping, Shadow AI scan, privacy tier recommendation, phased roadmap, written report.
- STEP 04 3–6 weeks
Setup
CLAUDE.md hierarchy, custom skills, MCP Perforce, permissions, monitoring dashboard, onboarding doc. Module by module, not big bang.
- STEP 05 ongoing
Retainer
Skills evolve with the codebase. Office hours. PR reviews. New Claude capabilities folded in. Quarterly ROI report.
Privacy & IP
Your code never trains a model.
The biggest blocker we hear: "Our code is proprietary." Here's how we solve it, ranked by paranoia level.
Claude for Work / Enterprise
Commercial terms. Anthropic does not train on your data. Suitable for ~80% of studios. Fastest to deploy.
Claude API (Commercial Terms)
Direct API access under commercial terms. Same training protections, plus tighter log retention. Better fit for studios with custom tooling.
Zero Data Retention (ZDR)
Enterprise addendum. Logs processed only for real-time abuse detection, then immediately discarded. No chat content, metadata, or request details persisted. Right answer for AAA / licensed IP / heavy NDA work.
AWS Bedrock or Google Vertex AI
Claude runs on the cloud provider's infrastructure. Your code never touches Anthropic infrastructure. Maximum control, zero-friction compliance for studios already on AWS or GCP.
Critical to know. Claude Free, Pro, and Max are consumer accounts. By default they retain data for up to 5 years and may use it for training unless manually opted out. If your devs are using personal Claude accounts on company code, you have a leak. Detecting and shutting that down is part of every Audit.
Differentiation
Why a gamedev specialist, not a generalist AI consultant.
Generalist AI consultant
This practice
Pricing
Transparent pricing. No hourly billing.
Fixed-scope packages. 50% upfront, 50% on delivery. Anthropic API tokens billed at cost + 15% as a separate line item.
Custom scopes available for larger studios. Volume discounts on Retainers paid annually.
FAQ
Questions, answered.
Do you work with studios outside Portugal?
Yes. Most engagements are remote with one or two on-site weeks. Clients in EU, UK, and Nordic countries are the most common fit.
What if our team uses Plastic SCM, not Perforce?
Covered. Custom MCPs for Plastic on request, billed within Setup scope.
Can you sign our NDA?
Yes. Mutual NDA is signed before the Audit kicks off. Larger studios usually have their own template; we'll review and adapt.
What about Codex / GitHub Copilot / Cursor?
Our specialty is Claude Code and the Anthropic ecosystem, where we go deepest. We can integrate alongside other tools your team already uses, and we'll tell you honestly when another tool is the right fit for a specific use case.
How is ROI measured?
During the Audit we agree on 3–5 KPIs (review cycle time, PR throughput, time-to-onboard, etc.) and baseline them. Quarterly retainer reports track delta.
What if we already started rolling out AI internally?
Even better. The Audit doubles as a course-correction. We've seen teams 3 months in with the wrong tier of privacy, no Perforce integration, and no cost controls. Easier to fix than to start cold.
What does the typical engagement look like end to end?
Discovery call → questionnaire → Audit (1–2 weeks) → Setup (3–6 weeks) → Retainer (ongoing). Total active engagement to first results: 4–8 weeks.
30 minutes.
No pitch deck. Just a real conversation about your stack.
Bring questions about your codebase, your team, your blockers. We'll tell you whether we're the right fit. If we're not, we'll point you to who is.
No prep needed. Pre-audit questionnaire only after the call, if we agree to proceed.