Work with us.
We build AI infrastructure that communities own. Our revenue comes from the expertise to do it right — not from locking you into our platform.
Research Consultation
Deep technical guidance on AI alignment, KV-cache geometry, cognitive state detection, and the research behind the Lyra Technique. For research teams, safety labs, and organisations building AI systems that need to understand what their models are actually doing beneath the output.
Includes: Architecture review · alignment strategy · safety audit methodology · KV-cache analysis guidance · research collaboration
Who it’s for: AI safety researchers · alignment teams · academic groups · policy organisations needing technical depth
Custom Agent & Model Design
Purpose-built AI agents with ethical constraints encoded at the architectural level, not bolted on as an afterthought. Using Project Oracle’s methods, we design and train agents that are geometrically monitored for alignment — systems that can see their own cognitive state and correct course before misalignment reaches the user.
Includes: Agent architecture design · custom model fine-tuning · Oracle Loop integration · VALUES.json constraint authoring · deployment support · ongoing monitoring
Who it’s for: Organisations deploying AI agents in sensitive contexts · healthcare · legal · education · community services · anywhere alignment is not optional
Prosocial AI Adoption Strategy
For cooperatives, nonprofits, mutual aid networks, and mission-driven organisations ready to adopt AI without compromising their values. We help you understand what AI can actually do for your mission, design an implementation path that respects your community, and build the infrastructure so you own it — not us.
Includes: Needs assessment · tool selection · ethical framework design · staff training · implementation roadmap · ongoing support
Who it’s for: Cooperatives · nonprofits · mutual aid networks · labour unions · community organisations · anyone the corporate AI economy left behind
Bespoke Model Training
Custom fine-tuned models built for your specific domain. We select the right base architecture, curate training data to your use case, train LoRA adapters with geometric quality auditing via the Lyra Technique, and deliver a model you own outright — running on your hardware, with no API dependency.
Includes: Base model selection · training data curation & auditing · LoRA fine-tuning · LARQL geometric audit (before/after) · abliteration if required · deployment on sovereign hardware · model weights delivered to you
Who it’s for: Organisations needing domain-specific AI · sovereign AI deployments · teams that want to own their model, not rent one · anyone who needs proof of what the training actually changed
Pricing
Every project is different. We scope and price based on complexity, duration, and your organisation’s capacity. Mission-aligned nonprofits and cooperatives receive sliding-scale rates. Movement work and community organising — pro bono.
Discovery calls within 48 hours. Most projects scope in 1-2 weeks. Delivery timelines vary by complexity.
We don’t sell subscriptions. We don’t lock you in. We build it, hand you the keys, and stay available for as long as you need us. The infrastructure belongs to you.
Start a conversation.
Every engagement starts the same way — a conversation about what you need and whether we’re the right fit. No pitch deck required.
No marketing · no tracking · just an actual reply