AI Knowledge Exploration · Powered by VISML
One topic. Four ways in.
Knowledge that comes to life.
Write a topic into a Rumima mind map and AI helps turn it into something you can explore: a simple explanation for a young learner, a clear one for a teenager, the engineering principles for a student, and the hard questions for the expert — all connected, all in the same map.
VISML — Using AI to Make Knowledge Visual and Come to Life.
Example · a jet-ski expedition
A trip along the coast becomes four connected subjects.
Planning a jet-ski expedition touches GPS navigation, electrical repairs, underwater photography and marine exploration. In a mind map they stop being separate chapters: the GPS needs the battery, the camera needs the light, the route needs the tide. Pick a topic and a level to see how each node grows.
QnA,tech16 — GPS navigation
For teenagers
GPS navigation
Each GPS satellite broadcasts its position and an extremely precise time. The receiver measures how long each signal took to arrive, turns that into a distance, and combines the distances from at least four satellites to fix latitude, longitude, altitude and its own clock error.
How a topic comes to life
You write the map. AI adds the explanations beside it.
Nothing about mind-mapping changes. You add nodes the way you always do; the learning levels you choose decide what the AI writes next to them.
Write a topic
Add a node — “GPS navigation”, “Fuses & corrosion” — anywhere in your map. Its neighbours give the AI context.
Choose the levels
Set the map, or a single node, to tech12, tech16, tech21, nerd or a combination such as tech16 + nerd.
Save — RSGPUW does the rest
On save, Rumima desktop sends the node and its context through RSGPUW to a language model on a remote GPU, in the background.
Explore, edit, learn
The answer lands as a child node labelled QnA,tech16. Nerd questions fold open one by one and export to Anki flashcards.
Your edits win. When a topic changes, only its explanations regenerate — and a generated node you have edited is never overwritten. Choosing a learning level for a map means its node text is sent to the GPU service to generate them. AI can be wrong: treat generated nodes as a starting point, and check anything safety-critical, such as electrical repairs, against a qualified source.
Four ways to explore the same knowledge
Not four products. One map, seen from four depths.
The same node can carry all four. A family can start at tech12, a student can switch to tech21 on the same topic, and an engineer can go straight to the nerd questions.
Younger learners
Make knowledge understandable to younger learners.
Teenagers
Explain technical concepts to teenagers.
University level
Explore university-level concepts and engineering principles.
Deep dive
Investigate low-level technical details through advanced questions, answers, formulas, and references.
Powered by VISML
The parts that work together.
VISML — Using AI to Make Knowledge Visual and Come to Life. These components are intended to work together as one interactive learning environment.
VISML
The overarching vision of making knowledge visual, interactive, and understandable through AI.
Rumima
The visual mind-mapping and collaboration environment where people organize, explore, and share knowledge.
RSGPUW
The remote GPU processing bridge connecting applications to AI inference resources.
LLM & JEPA
Complementary AI technologies supporting explanations, contextual understanding, and knowledge exploration.
How a request flows
The language model writes the explanations. JEPA is planned as a context helper — finding related nodes and gaps — and is not a source of facts.
More about the philosophy: About VISML.
Honest status
What works today, and what is still being built.
AI knowledge exploration is new. Here is where each part stands.
| Capability | Status | Notes |
|---|---|---|
| Rumima mind maps, live collaboration, Anki and draw.io export | Available | The foundation everything else builds on. |
| tech12 · tech16 · tech21 · nerd generation in Rumima desktop | Early testing | Runs in the background on save; needs a connection to an RSGPUW GPU service. |
| Choosing learning levels on Android | Early testing | The phone sets the levels and shows the results; a desktop generates them. |
| RSGPUW remote GPU access for Rumima users | In development | Access model and availability not yet announced. |
| JEPA context selection (related nodes, gaps) | Planned | Research stage. Today, context comes from a node’s parent, siblings and children. |
| The jet-ski explorer on this page | Illustration | Hand-written sample content, not live output. |
Powered by VISML
Make knowledge visible, understandable, and alive.
Start with the mind map — Rumima is available today. AI knowledge exploration is arriving step by step; ask us if you would like to take part in testing.