About
What we are here to do.
Make an organisation's whole knowledge reachable and connected, through one contract that does not change, on hardware it controls.
The purpose
Three words are doing the work.
- Reachable
- One address. One set of commands. A caller connects once and can search code, documents, schemas, live databases and live APIs without learning a special case for each.
- Connected
- Not a list of results. A real map, where a mobile screen links to the API it calls, which links to the database table that API reads. Search tells you what is relevant. A map tells you what depends on what.
- Contained
- One program inside your own network. The index is ordinary files on disk. No step of it has to send your content to an outside service, so data that is not allowed to leave does not have to.
Who
A small team, building one thing.
RootNet is a brand and a product house. Myco is the product line under it. There is no large team behind this and the engine was built by a very small one.
That shapes what we build. A small team cannot cover every system in the world, so the useful thing to build is not a hundred connectors but a contract other people can write a connector against, and a core that does the hard part once.
Everything is measured against three goals, and they are deliberately few: ship something a regulated organisation can run entirely inside its own network and pay for, build distribution through an open core, and stand up the hosted edition only once the self hosted one is worth running.
We put the hosted edition last on purpose. Hosting other people's source code and database credentials is a serious trust and liability surface, and it is the most expensive thing on the list to build.
Principles
Five rules, and each one costs something real to break.
These are architectural rules rather than values. They are on this page because they decide what the product can and cannot become, and because the third one is the one that costs us the most.
- P1 The contract facing the caller never moves
- The set of commands, and the shape of the answer they return, are frozen and versioned on their own. Everything underneath can improve indefinitely without breaking anybody who built on it.
- P2 Everything specific lives at the edges
- Knowledge about a particular kind of source lives at the outside, in the parsers. The core never learns about a particular customer or a particular system, so a kind of source we have never seen is added at that edge without moving the contract in P1.
- P3 A source cannot advertise what it cannot do
- A live answer is never served from a cache and labelled live. If a system is unreachable we refuse and say so rather than returning something stale and calling it current.
- P4 Self contained, and self host first
- One thing to deploy. No separate database. No external search service. The model runs locally by default. Nothing has to leave the network.
- P5 The index gets better under a frozen contract
- Better parsers, better precision, better linking, all shipping underneath. The improvement arrives with no migration.
Scope
What we are not building.
Saying no early saves everybody a lot of time later.
- A general purpose vector database. We do not want to sell or host one.
- A coding agent or an editor. We serve agents. We are not the agent.
- A horizontal knowledge assistant chasing the general office document search buyer. That fight is expensive and already crowded.
- A general write and mutate engine over your systems. Read only, and only the actions you have explicitly permitted.
The name
We checked the story before we built a brand on it.
We picked RootNet and Myco because of a story about forests: that trees are linked underground by a fungal network and use it to communicate. Before spending anything on the name, we went and read what the science actually says.
The short answer is that the structure is real and well documented, and the popular story about what the structure does is contested. So we named ourselves after the structure. This section is here because the checking is the interesting part, not the biology.
What is solidly established
- A hypha is a long, branching, thread like filament, and a mycelium is the network those filaments form. So the word already means, quite literally, a branching network.
- A mycorrhiza is a partnership between a plant's roots and a fungus. The fungus passes mineral nutrients to the plant and the plant passes carbohydrates back. Around 90% of land plants take part.
- When the threads of a mycorrhizal fungus connect the roots of more than one plant, that is a common mycorrhizal network. The definition is not disputed. Even the paper that pushed back hardest on the rest of the story opens by stating it plainly.
- Material really can move between trees through fungi. Carbon labelled in the air around one seedling turned up inside a neighbouring seedling of a different species, having travelled through roots and fungal tissue.
Sources: Britannica on mycorrhiza, and Simard et al., Nature, 1997, the original carbon transfer experiment.
What is contested
In 2023 a review in Nature Ecology and Evolution examined the three claims that popular writing repeats most often. This is what it found, and it is the half of the story that usually gets dropped.
| The claim | Verdict | Why |
|---|---|---|
| These networks are widespread in forests | Insufficiently supported | Field results vary too widely, have alternative explanations, or are too limited to generalise from. |
| Resources transferred through them increase seedling performance | Insufficiently supported | The same reasons. |
| Mature trees preferentially send resources and signals to their own offspring | No peer reviewed evidence | None published at all, for the claim most often repeated. |
The same review found that unsupported claims about these networks in the scientific literature doubled over twenty five years, driven by a bias toward citing positive results, and concluded that knowledge about them is "presently too sparse and unsettled to inform forest management". Karst, Jones and Hoeksema, Nature Ecology and Evolution 7(4) 501-511, 2023
A mycelium is a branching underground network of fungal threads that physically connects the roots of separate plants, and material can move across it. That is established, documented and safe to say.
That trees use the network to communicate, to recognise family, or to care for their young. It is a good story and it may yet turn out to be partly true. The evidence today does not support it, so we do not build marketing on it, and you will not find those phrases anywhere on this site.
Why the structure is the right analogy anyway
| In a forest | In Myco |
|---|---|
| Separate trees, each its own organism | Separate systems, each its own silo |
| Fungal threads physically connect their roots | Typed links connect entries across sources |
| Material moves along real, physical connections | Answers travel along real, provable links |
| Around 90% of land plants take part | Almost every system in an estate can be connected |
| The network is underground and invisible | The index and the map sit underneath, out of sight |
The row about material moving along real connections is the one worth repeating. In a forest those connections are physical: a thread either reaches a root or it does not. Our structural links work the same way. Each one is an exact fact a parser proved from an import statement, a call site or a database reference, and it points at a specific line. We do not invent a link because two things look similar.
RootNet is the root network. Myco is short for mycelium, the thing that actually does the connecting.
Where we are
And to be clear about the stage.
Myco is not generally available yet. If this is the shape of the problem you have, write to us and tell us about it.
There is one address and it reaches the person who built it: rutvik@flareware.app. There is no contact form and no chat widget, on this page or anywhere on this site.