Trust LeagueThe method and the workbench

The twin

The Trusted Living Twin

Your business, measured before artificial intelligence (AI) and after it, with the agents, vendors and gaps found, on one record

The Trusted Living Twin is MortarCloud.ai's software. It reads your organization's knowledge graph and presents it as pages a leader, a board or an insurer can read: the register, the gaps, the baseline and the picture they make together. The graph is the record of your business: measured before AI touched it, with the agents, vendors and gaps found, on one record, and a named person beside each gap or the admission that there is none. The twin lets you trust that your AI is paying back, or shows you that it is not, and backs either answer with that dated record. Its use during an engagement is part of the engagement, a client license afterward is optional and opt-in, on terms agreed at that time, and we do not host client twins.

Hollister Reed LLP is a fictional firm. The demonstration linked here was built for a firm that does not exist, so that you can see the method at work from end to end.

What you own, what MortarCloud.ai builds, and how it is built

Your organization's knowledge graph is yours. It is the record of your goals, people, processes, systems, data, agents and vendors, it stays in your environment under your control, and it stays yours when you change vendors. Today the graph is a versioned file.

The Trusted Living Twin is MortarCloud.ai's software. It reads the graph and presents it, and it is not the graph itself. Today it reads the versioned graph and renders it as pages, its use during an engagement is part of the engagement, and it extends with each engagement. A client license afterward is optional and opt-in, on terms agreed at that time.

The graph is built by federated knowledge intake. The people who hold the knowledge contribute it as part of their ordinary work, an accountable owner accepts each change, and the graph keeps growing after the engagement ends. Each contribution passes through three steps: AI-assisted extraction from the documents and exports you provide, review by people of everything extracted, and a deterministic build, in which the same reviewed inputs always produce the same versioned graph. The twin's pages are generated from that graph. The Trusted Living Agent Framework is the method that governs all of it.

What the graph holds, and the twin shows

The pre-AI baseline
The firm's mission and goals mapped to its actual performance, frozen and dated before AI touches the business, and re-measured against that date whenever the firm asks. The measures are ordinary on purpose: revenue, gross margin, operating expense, headcount, revenue per employee, customer count and churn, and the cycle time of the three core processes the firm names as the ones it lives by.
The register
Every agent and tool found running in the firm, each with a Tail Number, the identifier the framework paints on an agent the way a registration is painted on an aircraft, and each with an owner's name or the honest word that there is none.
The vendor ring
Every vendor found in what the firm already pays for, so that each dependency is seen before it is relied on.
The gaps and the plan
Every gap tagged to the pillar of the framework it falls under, and for each one the person who owns closing it or the admission that nobody yet does.
Before and after
The questions the firm is asked by its insurer, its customers and its board, with what it could say when the baseline was taken and what it can say now, each answer linked to its evidence or marked as unanswered.

What you see

The twin is a set of pages drawn from the graph, one record, and the footer of each says so. Two numbers sit at the top of every page: the trust posture, one reading drawn from the five pillars, and the results since baseline, the seven measures as change against the day they were frozen. Beside them is the verified fraction, the share of everything the record holds that has been checked against a second source rather than taken on someone's word. Every entry is drawn by how it is supported, so a reader can tell at a glance how much of the picture is known and how much is asserted.

  1. Who and why. The firm's stated goals, the drivers that serve them, its departments with the person accountable for each, and the things it spends money on that serve no stated goal.
  2. The register as found. Every agent with its Tail Number and its owner, beside the ring of vendors.
  3. The gaps. Each tagged to its pillar.
  4. The baseline. The seven measures, each with the date it was frozen and the source it was read from.
  5. The plan. Every gap with its owner or no owner, and what closes it.
  6. Before and after. The questions, the answers then and now, and the evidence behind each.

Two one-page documents fall out of each re-measurement. The board summary is one printed page: the two numbers, the verified fraction, the five pillars and the five drivers. The receipt carries the questions an insurer asked, what the firm could say at the last renewal and what it can say at this one, with the evidence beside each answer. Neither is written by hand.

Which of the ten questions the record answers

A leader should be able to answer ten questions about any AI in the business. The record answers four of them directly, with evidence rather than assurance, and the pages below show the evidence.

10. If a provider, a model or a person changed tomorrow, what would the business keep?
The record itself, in the firm's own environment. The vendor ring shows each dependency as a dependency, so that a leader can see what leaving would cost before deciding whether to stay.
02. What has it changed, at what full cost, and what can honestly be credited to it?
The baseline is the before that nobody keeps. The twin shows every result as change since the date it was frozen, with the record behind it. Whether AI caused the change is a judgment people make with that record in front of them.
01 and 03. What business outcome does it serve, and which AI is in scope?
The register and the vendor ring hold the agents, tools and vendors found, on one record, connect each to a goal the firm has captured, and show the spend that serves no goal.
05. Who owns it, and what is it allowed to cause to happen?
Every agent carries a number that is never reused and that names its owner, and the register shows the one whose owner has left the firm. The passport beside each number records the agent's purpose, its autonomy level and its human checkpoints, or marks the gap where none is on record.

The twin shows all of this surface by surface, and it marks every entry by how it is supported. It does not say what caused a result, because it does not know what caused it and says so; the person who sponsored the work signs the attribution, or declines to, and the page records which. A projection is never measured.

See it work

Hollister Reed LLP is fictional. It is an invented regional accounting and advisory firm, and its twin is built from end to end so that you can look at the thing rather than take a description of it. It is the only public twin. The five-question screen draws your first sketch of your own firm's twin, one question for each pillar, and nothing you answer leaves your browser.

Where the twin goes next

What this page describes above is what the twin does today. The twin grows with each engagement.

If you want this

The record is built in Ground, MortarCloud.ai's first engagement, around one consequential business decision, normally one or two departments and a bounded source set, and it records the AI already in use as found. Ground is described on mortarcloud.ai. Step Zero captures four answers: what the business is for, what it stands for, who decides, and how you will know it worked. A business that already runs AI keeps what it has built, and the twin fills in as each engagement adds to the record.

A client's graph is the client's own, and the record and the views built for it are confidential. The twin reads the client's own record, its use during an engagement is part of the engagement, and a client license afterward is optional and opt-in, on terms agreed at that time. MortarCloud.ai hosts no client twin and keeps no client data after delivery, and no client's twin appears on this site; the Hollister Reed demonstration is fictional, and it is the only public twin.

The essay beside this page says what a baseline is, what it makes possible, and what a firm holds once it has one: What AI Changed.