Skip to content

Backbone OS, the engine behind every deployment

Our product records who does each thing, with what permission and what changed. What can be undone has a way back, and what is definitive waits for the person in charge.

Backbone OS comes inside every system we deploy: AI works freely while your company keeps authority, truth and a record of every action.

Free above. Firm beneath.

AI needs room to solve problems, and your company needs what matters to stay under control. That is why every system is built in four layers.

  1. 01 · The base

    Your operation

    People, customers, suppliers, exceptions and the reason behind each practice. We start by understanding how the work gets done every day.

  2. 02 · They stay where they are

    Your systems

    Email, ERP, CRM, spreadsheets and documents. We get them ready and connect them around the outcome you want.

  3. 03 · Firm beneath

    Backbone OS

    It knows who acts and with what authority, what the company holds as true and what changed outside. Every action leaves a record and a way to undo or correct it.

  4. 04 · Free above

    AI

    It reads, proposes and acts within what the floor allows, and checks its own work.

Models change every few months. With this split, a better model is swapped in above and your company’s floor stays the same.

What every task keeps

Durable work
A task keeps its identity and state through failures, retries and handovers.
Proposal before effect
Anything that changes outside is proposed first and authorised to the measure of its consequence.
A record of every action
Who acted, within which scope and what outcome was observed.

What holds firm

Identity
Who acts, for whom and inside which company.
Authority
What it may read, propose or do, and where its scope ends.
Truth
What the company holds as true, with provenance and an owner.
Effects
What changed outside the system and the best path to recovery.

This is what a task looks like

A fictional order crosses four systems. Readings and proposals change nothing outside; reversible effects can be undone and definitive ones wait for your decision.

Order 1842 · fictional example1 task · 4 systems
  1. 08:42Reads the order: customer, stock and terms checked across email, CRM and ERP.reading · no effect
  2. 08:43Proposes the delivery address: the one from the last accepted order, with the source linked.proposal · no effect
  3. 08:44Holds stock for 30 minutes: a reversible effect, within its permission.
  4. 08:45Updates the status in the CRM: “confirmation in progress”, reversible.
  5. 08:46Sends the confirmation to the customer: a definitive effect, waiting for the responsible person's signature.
Backbone OS beneath every action
  • Who acts
  • With what permission
  • What is held as true
  • What changed

Fictional example: no data belongs to an existing customer or system.

2 reversible effects · 1 awaiting authorisationEvery step leaves a record
Fictional, interactive example; it does not represent a customer deployment.

Product state: working local beta, verified on 15 September 2026. The release candidate is still in development and currently comes inside each deployment.

Open standards the work converges on

MCP
Model Context Protocol: how AI connects to your tools and data, with permissions.
A2A
Agent2Agent: how agents from different vendors work together.
OAuth 2.1 and OpenID Connect
Who is who and what they may do, with your company’s identities.
OpenTelemetry
What happened at each step, visible in standard observability tools.
Agent Skills
The team’s judgement in a format you can take to another system.

We adopt them as they mature, and the state of each integration is stated with its date.

What you can check today

We separate working software, controlled tests and customer impact. Each one appears with its label.

Work that survives

Tasks keep their identity and state through retries, crashes and recovery.

Scope Verified in the product's local environment.

local beta

No duplicate effects

30 cases, 30 terminal records, 0 duplicate effects and 0 additional effects on replay.

Scope Invented data and controlled execution.

synthetic proof

Checkable output

Every proposal and execution keeps who acted, within which scope and what outcome was observed.

Scope Not yet attributed to a customer deployment.

local beta

Tell us which process you would like to see working.