AI Model Playground
Built-in model access out of the box. Compare outputs, tune parameters, and save prompt templates — no setup required.
Lab Maneuver gives every user their own AI workspace — model playground, Jupyter notebooks, RAG knowledge base, and hard token limits. One platform. Full control. No surprise bills.
No Kubernetes expertise needed. No DevOps team required.
Set up your organisation, configure token limits, choose which labs and models are available, and define user tiers.
Upload users under your org — individually or in bulk. Each user gets an isolated workspace with their own quota, ready immediately.
Users log in and access built-in Bedrock models across all labs — chat, RAG, notebooks, agents, and more. Optionally connect your own API keys per org.
From prompt engineering to production RAG pipelines — secure, scalable, and cost-controlled.
Built-in model access out of the box. Compare outputs, tune parameters, and save prompt templates — no setup required.
Per-user Jupyter notebooks with auto-scaling, persistent storage, and pre-loaded AI/ML libraries. Productive from day one.
Upload PDFs, docs, and text — auto-chunked, embedded, and indexed. Ask questions and get answers with cited sources. Private per user.
Token quota system with real-time tracking, monthly resets, and configurable tiers. No surprise bills ever.
Org-level isolation, RBAC, per-user data separation, and JWT authentication built in.
Build teams of specialised agents — Researcher, Writer, Reviewer. Chain them in sequence or run in parallel. Watch each agent's output in real time.
Upload images and ask questions about them. Extract text from screenshots, describe charts, identify objects — powered by Bedrock vision models.
Built-in Bedrock models work out of the box. Need more? Connect OpenAI, Anthropic, or any custom API endpoint per org — your keys, your costs.
Every user gets a private PostgreSQL schema — browse tables, run queries, do vector search, and connect from JupyterLab or any SQL client using personal credentials.
Per-user and org-level token consumption by lab, by day, and by model. Full visibility for admins and users.
Design prompt templates with {{variables}}, run them against any model, and save for reuse. Every run is logged with token count and output.
Connect GitHub, Notion, Slack, Tavily web search and more via Model Context Protocol. The AI automatically picks the right tool based on your question.
Connect via LTI 1.3 or REST API. Users are live in minutes, not days.
Launch from Canvas, Moodle, Blackboard or any LTI-compliant LMS. Students auto-provision — no separate login needed.
POST /lti/launch
iss: canvas.instructure.com
sub: student_12345 → auto-provisioned ✓
Provision users, set quotas, manage sessions and pull analytics — all from your own backend or automation tools.
POST /api/users/provision
PUT /api/users/:id/quota
GET /api/analytics/usage
Start free, scale as your team grows.
For individuals & small teams exploring AI
For teams & organizations at scale
Building this yourself takes months. We've already done it.
| Feature | DIY / Cobbled tools | Lab Maneuver |
|---|---|---|
| Per-user token quotas | ✗ Build it yourself | ✓ Built-in |
| Jupyter notebooks per user | ✗ Manage Kubernetes | ✓ Auto-provisioned |
| RAG vector store | ✗ Separate service | ✓ Included |
| LMS / LTI integration | ✗ Custom dev work | ✓ One config |
| Multi-tenant security | ✗ Easy to get wrong | ✓ Org-level isolation |
| Bring your own LLM | ✗ Re-wire everything | ✓ Per-org API keys |
From research labs to fast-moving startups — here's what they say.
"Cortex replaced 4 different tools. The RAG engine with web fallback is incredibly smart — it just works."
"The cost controls saved us from a $12k surprise bill. Per-user quotas with real-time tracking is a game changer."
"Jupyter notebooks with pre-loaded ML libraries means our data scientists are productive from day one."
Practical deep-dives on cost controls, RAG pipelines, and deploying AI infrastructure.
Shared API keys and no usage limits are a recipe for surprise invoices. Here's how per-user token quotas work.
Read article → University / LMSWhat the JupyterHub setup looks like, what breaks, and how to avoid it.
Read article → Research LabHow to make your team's knowledge base queryable by an LLM.
Read article →One platform. Every user gets their own workspace, token budget, and full AI stack. We'll help you every step of the way.
No credit card · No DevOps needed · No setup headaches