Are You Loco Enough?

LocoLab is not for everyone. It is for people who want to understand what local AI can actually do — and are willing to get their hands dirty finding out.

If any of these sound like you, welcome to the lab.

💰 The Budget Rebel

You refuse to pay per-token for something your own hardware can do. You’ve done the math on API costs and it offends you. You’d rather spend a weekend setting up a local stack than sign up for another subscription you’ll resent.

Your entry points:

LocoPuente — a full local AI service stack on hardware you own and control
LocoLLM — a routed specialist model that runs free on consumer GPUs
AI Landscape — honest comparison of local vs cloud options, including cheap API paths

🤖 The Tinkerer

You want to understand how LLMs actually work by cracking them open and rewiring the internals. Reading about fine-tuning is not enough. You want to train a real adapter, measure whether it helped, and understand why.

Your entry points:

LocoLLM — adapter training, evaluation harnesses, and a router you can improve
LocoBench — systematic benchmarking infrastructure to measure what you built
Getting Started — technical foundations: inference, VRAM, quantisation, the full stack

🔬 The Researcher

You need reproducible local inference for experiments. You want to test whether specialist routing actually beats a generalist on scoped tasks — and publish honest results either way. You are not interested in vibes.

Your entry points:

LocoBench — VRAM-tier benchmarking with real hardware, real cards, honest baselines
LocoConvoy — multi-GPU parallelism experiments on consumer PCIe hardware
LocoAgente — agentic scaffolding research: can small models think in loops?
Research — active and planned studies across the lab

🏫 The Educator

You teach AI, computing, or a professional discipline and want a real project your students can contribute to. Not a toy demo. Real infrastructure that grows with every cohort. You also want rehearsal environments where students practise professional skills before they face the real thing.

Your entry points:

LocoEnsayo — AI-populated rehearsal environments: security audits, requirements gathering, difficult conversations
LocoLLM — a teaching and research framework students build by contributing adapters, benchmarks, and routing improvements
Why Local AI — the case for local AI in education and institutional contexts

🔐 The Vault

Your data does not leave your machine. Period. Medical notes, legal research, personal journals, proprietary code, student assessment work — local inference is not a convenience, it is the only acceptable path. You do not need to be convinced. You need the stack to work.

Your entry points:

LocoPuente — local AI services for institutions and individuals who cannot or will not use cloud inference
AI Landscape — why “private by policy” is not the same as “private by architecture”
Why Local AI — data sovereignty, compliance, and the structural argument for local inference

⚙ The Scrapper

You know the best gear does not make the best work. A $150 secondhand GPU and sharp training data might just surprise you. You are assembling capability from what is available, and you want to know exactly where the floor is.

Your entry points:

LocoBench — floor-representative benchmarks: worst card per VRAM tier, honest baselines
GPU Inventory — the actual secondhand fleet running these experiments
Economics of Local Training — what local AI actually costs to build and run

Not sure where to start?

If you are new to local AI entirely, Getting Started covers the technical foundations without assuming prior knowledge. If you want to understand the broader landscape before committing to anything, AI Landscape gives an honest comparison of every option — including the ones that are better than LocoLab for your use case.

Loco by name. Serious by intent.