Controlled access
Requests are reviewed individually. Access is granted per person and can be suspended or revoked at any time — authorization is re-checked on every protected request, not once at the door.
Greister Technology — Private Platform
MERIDIAN
Intelligence, behind the glass. Invitation only
A private environment for exploring Greister's intelligence systems.
Built for controlled access, rigorous experimentation, and the next generation of machine intelligence.
Platform status — controlled access
01 — Scroll
01 Platform
One private environment for Greister's intelligence systems.
MERIDIAN is a controlled research surface. It exists to make selected Greister models usable — safely, precisely, and only for people who have been authorized to work with them.
It is not a public playground. Every visitor can request access; every request is reviewed; only approved users reach the protected workspace.
Rigorous experimentation
Submit validated inputs, run supported operations, and inspect exactly what a model returns. No decoration, no invented confidence — only real outputs, labelled honestly.
Proprietary intelligence
Built around models, not the other way around. Each system exposes only the capabilities it actually supports, and nothing about it leaks into the browser.
Intelligence is not defined by what it reveals, but by what it allows us to understand.
MERIDIAN — Product principle
02 Access
Requested, reviewed, and granted — in that order.
Access to MERIDIAN is deliberately human. An application tells us who you are and how you intend to use the platform; a reviewer decides; an invitation follows.
Step 01
Request
Submit the access form with your background, intended use, and reason. Your application is stored securely and enters the review queue.
Step 02
Review
A Greister administrator inspects the application and approves or rejects it. Every decision is recorded in an audit history.
Step 03
Activation
Approved applicants receive an invitation to create their account. Only then does the protected workspace open.
03 Intelligence
One platform. Many ways of observing.
MERIDIAN is model-independent by design. A forecasting system may accept candles, markets, and horizons; a different system may accept something entirely different. The workspace adapts to each model's declared capabilities instead of forcing every model into one template.
Models that are not configured or not currently available are labelled as such. If a server-side integration is missing, the interface says so plainly — and no output is invented to fill the gap.
Declared capabilities
Each model declares purpose, version, supported operations, accepted inputs, and availability. The registry — not the interface — decides what you can do.
Adaptive surface
A forecasting model exposes charting, horizons, and candle input. A different model reshapes the workspace around what it actually supports.
Honest rendering
Predictions, intervals, or confidence appear only when the model genuinely returns them. The chart draws a forecast only after a real response arrives.
The most useful discoveries often begin at the edge of what we already know.
MERIDIAN — Research notes
04 Research
Forecasts you can trace back to data.
When a forecasting model is connected, the workspace accepts validated candle data, applies only the horizons that model supports, and renders the output the model actually returns — overlaid on the data you submitted.
Every candle is checked before submission: structure, timestamps, numeric values, sequence length. Input is never silently altered, and failed requests are explained rather than disguised.
Structured candles
Timestamp, open, high, low, close, and volume when supported. Pasted as JSON or CSV, or loaded from a file — validated client-side, and again server-side.
Only what returns
Predicted points, directional calls, or forecast intervals — exactly the fields the model returns, with the raw response kept alongside for inspection.
05 Model access
Model access, without the guesswork.
Oracle ViSION and Goose are examples of Greister intelligence systems that may eventually be accessible through MERIDIAN. The platform does not assume they behave alike — and it will never list a model as available before that is actually true.
Every model offers a different way of observing the world.
MERIDIAN — On model independence
06 The future
Infrastructure first. Models continuously.
What will remain stable is the part that matters: authentication, controlled access, entitlements, audit history, and a workspace that adapts. Models will come and go on top of it.
This page describes what the platform is designed to do. Features that are planned rather than implemented will be enabled only when they are real.
The infrastructure
Access review, invitations, entitlement checks, revocation, and auditing belong to the core platform — not to any individual model.
The models
Each system is integrated through its own adapter with its own validated inputs and outputs. Adding a model never means rebuilding the platform.
The access
Authorization is re-checked on every protected request. Access granted is access that can be precisely and immediately withdrawn.