Understand the economics
Learn what API keys authorize, compare token cost and latency, monitor usage, and keep provider spend visible.
Open-source AI capability laboratory
Learn what AI can do by changing the controls yourself.
Go beyond ordinary chat. Compare models and instructions, work with documents and tools, build agents, run verification loops, and rehearse coding or automation safely—on your own deployment.
- Guided, contextual, or expert mode
- Bring your own provider keys
- Cloud and local models
- Private, self-hosted progress
Controlled comparison
Private experimentQuestion
Review this plan for missing risks and tests.
Side APlain
Fast first response
Side BVerify & Improve
Missing rollback test found
Privacy assumption corrected
Acceptance check added
Audited before final answer
ModelSame modelSame model
InstructionsPlainVerify & Improve
What changesFirst answerAudited answer
A flight simulator for AI
The controls stay visible. MaiaChat teaches them in the right order.
Most assistants hide model choice, instructions, tools, memory, permissions, and cost. MaiaChat keeps them inspectable so a learner can change one variable and feel the difference.
Explore the capability path- 01
Ask
Chat, models, and providers
- 02
Ground
Documents, memory, and live evidence
- 03
Act
Tools, files, media, and connected services
- 04
Build
Coding workspaces, diffs, tests, and approvals
- 05
Repeat
Loops, orchestration, workflows, and schedules
Rehearse before local power
Practice file changes, commands, diffs, tests, checkpoints, and rollback before an agent touches a real laptop project.
Turn success into a routine
Move from a successful prompt to a specialist agent, bounded loop, reusable skill, workflow, schedule, or notification.
A category between courses and agent tools
Learn, compare, and act in the same platform.
MaiaChat combines structured learning with the runtime people eventually need. You do not graduate into a different product when the lesson becomes real work.
| Capability | MaiaChat | Typical course or playground |
|---|---|---|
| Learn with real controls | Models, prompts, tools, agents, loops | Usually explanation or chat only |
| Compare fairly | Same task, model/profile controls, cost and latency | Separate tabs or aggregate scores |
| Practice real work | Documents, files, coding, automation, channels | Course sandbox or single-purpose playground |
| Keep control | Self-hosted, own keys, local-model support | Vendor-hosted ecosystem |
| Teach your organization | Admin-authored local lessons | Fixed public curriculum |
Start with one experiment. Grow into the full platform.
Choose guided or expert mode, learn why each control matters, and keep using the same workspace when chat becomes agents, coding, or automation.
Own your provider keysSelf-hosted and open sourceGuidance can be switched off