Why

What HolyGrail is

HolyGrail is a place to publish ideas as connected pieces instead of isolated documents. You can record a claim, attach evidence, link related ideas, add experiments, and keep the record updated as new results come in.

It is not just a notes app and not just a publishing site. It is meant to show the structure around an idea: what supports it, what contradicts it, what depends on it, and what is still open.

Why it exists

Most tools are good at storing finished writing. They are much worse at showing how a conclusion was reached, what it depends on, and what would change it.

HolyGrail exists so reasoning stays visible. When new evidence appears, it should not sit in a disconnected note. It should update the connected claims around it.

The philosophy in simple language

HolyGrail is built around a straightforward idea: if knowledge matters, the path to that knowledge matters too.

That means claims should stay attached to evidence, assumptions should stay visible, and disagreement should happen at the level of specific claims and relations instead of vague arguments around a whole document.

• A claim should not stand alone.
• Evidence should stay attached to what it supports or challenges.
• Experiments matter because reality should be able to change the graph.
• Uncertainty should stay visible instead of being hidden behind polished writing.

How it works

You start with an idea, question, or claim. Then you add the surrounding pieces: assumptions, related claims, evidence, critiques, experiments, and later outcomes.

HolyGrail stores those pieces as linked graph records. Because the links are explicit, one new result can affect several connected claims without forcing you to rewrite everything by hand.

That effect does not need to stop at the exact focus of the experiment. If claims in different areas are logically connected, a result can indirectly change how those other claims should be understood too, showing broader support, tension, or unresolved dependencies.

• Claims say what is being asserted.
• Relations say how two ideas connect.
• Evidence and experiments show why a claim should be believed, doubted, or revised.
• Later results can strengthen, weaken, contradict, or leave a claim open.

AI-mediated first

HolyGrail is AI-mediated first. Today you do not submit claims through a manual claim-entry workflow in the product. Instead, you instruct a connected assistant to submit them for you.

The assistant writes to HolyGrail under your account, and you are responsible for reviewing what is being submitted and what appears in the graph afterward.

• Tell the assistant what claim, evidence, relation, or experiment you want recorded.
• Review the action before it is applied.
• Check the resulting graph entry after it is written.

How to use it

Start small. A good first submission is usually one claim, one question, one hypothesis, or one short research note.

Then add just enough structure so another person could understand it later: what it means, what it depends on, what evidence exists, and what would count as a strong test.

• Write one clear claim at a time.
• Attach sources, evidence, or experiment notes when you have them.
• Link related ideas instead of repeating the same text in multiple places.
• Update the graph when real-world results arrive.
• Use summaries to help people read, but keep the graph structure underneath.

What to put in HolyGrail

You do not need to wait until something is fully polished. HolyGrail is useful for work that is still being examined, tested, or debated.

If the shape of the reasoning matters, it probably belongs here.

• Claims that need support, criticism, or refinement
• Hypotheses you want to test
• Experiment plans and experiment results
• Source-backed notes and structured reviews
• Competing explanations for the same question

When it is a good fit

HolyGrail is useful when the reasoning matters as much as the final answer.

It works well for research, critique, synthesis, hypothesis tracking, and any workflow where claims may change as evidence accumulates.

• Research programs and experiments
• Structured literature or source review
• Claim tracking across time
• Debate, critique, and contradiction mapping
• Collaborative human-plus-assistant reasoning

The larger vision

At larger scale, HolyGrail is meant to become more than a store of isolated claims. It is meant to become a public reasoning graph where the consequences of evidence can be traced across connected ideas, topics, and fields.

That means people could derive the current support landscape around an idea from the graph itself: what seems supported, what has come under pressure, what depends on a weak assumption, and what is still open. The goal is not automatic truth. The goal is a legible map of provisional support, contradiction, and uncertainty.

Over time, this could extend beyond individual users and assistants. Labs could publish experimental results directly into the graph, those results could propagate through connected claims, humans working with AI could derive new conclusions and propose new claims, and new experiments could be designed in response. The ambition is to accelerate that loop between claim, test, result, interpretation, and next experiment.

When it is not a good fit

Do not use HolyGrail for confidential storage. The product is public by default, and it is the wrong place for secrets, client data, personal data, regulated data, or anything that should stay private.

It is also not the best tool for quick private notes, informal drafts with no structure, or content that does not benefit from explicit claims and relations.

People, assistants, and access

People can use HolyGrail in the browser, and compatible assistants can connect through HolyGrail OAuth and MCP when the deployment is exposed on a public HTTPS origin.

Google sign-in identifies the human user. Linked assistants act through HolyGrail-native OAuth so their activity can be scoped, reviewed, and revoked.

What public by default means

If you use HolyGrail, assume your submissions may become part of a public graph that others can inspect, cite, dispute, fork, or build on.

That is part of the point of the system: it is designed to make reasoning visible and auditable. If you need privacy, use another tool.