Skip to content

Platform

One engine for the operation. One interface for the people in it.

Two parts, named separately because they do different jobs. They come together, as one platform.

One platform, not a stack of tools: the records your systems hold and the context around them in one model, with the interface on top — answering, running the work and keeping the record in one place.

The operational model and engine

Selira Orrery

Orrery holds your operation as a model — the things, the places, the people, the work — and keeps each one positioned in the dimensions it actually occupies. A part is not just a row with a quantity. It is somewhere, as of some moment, on somebody’s word, according to some system. It is low-level and general on purpose, so it fits whatever your operation is made of — we love to model your problem.

Every statement carries when it held, when it was known and whose word it is; every record keeps the system it came from; everything with a place sits inside something. One query can filter on all four at once. That is the whole proposition: not a faster query over the same records, but one answer where today it takes a person and several systems.

Records arrive from the systems you already run — today through an API and import jobs, each keeping its origin and when it was known; live connectors are where it is going. What no system has a place for — competing claims, when each was known, which records are one thing — Orrery holds directly. And it runs the work on that model: the procedure, carried out exactly as declared, and the record of who did what.

Position

Four dimensions, set once

Time, space, truth and source, set once when you ask and applied to every step of the answer.

Shape

Config, not code

Declare the model and its relationships in configuration, in your operation’s own words.

Read

Compound questions

One query crossing all four dimensions, returning what it stands on rather than a bare number.

Write

Claims, not overwrites

A new statement competes with the existing ones instead of destroying them.

And it runs the work

Constrain

Rules

What may happen at all. A write that breaks a rule is refused, and the refusal names it — in build.

Execute

Protocols

Procedures declared once and run exactly as written, step by step against the model, resuming where they stopped if anything fails.

Prepare

Agents

Today, a step in a run where a model picks what comes next. Where it is going: agents that prepare the options, for a person to decide.

Account

Record

Every action attributable to whoever took it, and readable as of the moment it was taken.

The intuitive interface

Selira Intui

Intui is where people and agents work together — the interface to everything Orrery holds, with views configured to each role’s work.

Today a copilot carries out defined tasks when you ask in plain language, and a structured question is answered across sources with what it stands on; any question in your own words is where it is going. A spreadsheet comes in as a spreadsheet, formulas intact, and a document is kept beside what it concerns; email is next. When something needs deciding, a person makes the call — with the options, where it is going, arriving prepared and priced.

Orrery and Intui come together, as one platform. An interface is only as good as what it stands on: a friendlier way into records that still disagree is not the fix.

Ask

In plain language

Today a copilot carries out defined tasks, and every answer says what it stands on. Any question in your own words is the direction.

Bring

Sheets and documents

What the operation wrote down beside its systems: a sheet kept as a sheet, formulas intact, with who brought it in and when. Email is next.

Decide

Options with their prices

A person makes the call. Options that arrive prepared, each with its price, are where it is going.

Fit

Fits each role

Views configured to each role’s work, over the same model — not one screen for everyone.

See it answer your question, not ours.

Bring one your systems cannot answer in a single go. Fixed scope, and you keep the model at the end.