PEGASUS COLLECTIVE · THE ENGAGEMENT, EXPLAINED

You built the machine. What makes it yours?

Seven steps, one page. You never buy the whole journey up front — each step tells you whether the next is worth it. Scroll to walk the full engagement, or jump straight to any step below.

↓ SCROLL TO WALK THE JOURNEY, OR PICK A STEP
00

Diagnostic

Self-serve · five minutes
FREE
↳ preview of Assess & Prepare (Step 2)

An online self-assessment across your people, your organization, and your technology — a printable read of where you're strong and where you're behind. It's built to surprise: most companies are further behind on their people and organization than their technology, the opposite of where budgets go. No call, no email wall.

HOW THE THREE MOVE HERE
H — is scored on two independent scales: how ready your people are to change, and how much the status quo is actually costing you.
O — is scored across eight organizational-readiness areas, 1–10 each, out of 80.
T — is scored across eight platform areas on the same 1–10/80 scale, so the two numbers sit side by side.
DECISIONIs the gap real enough to talk about? If yes, book a call.
Go deeper — what the read actually measuresShow less

H — two scales, not one number

Four questions on readiness for change and four on need for change, scored independently and read together into one of four bands: Primed, Ready-unconvinced, Willing-but-unequipped, or Stuck.

O and T — eight areas each, same scale

Each area is scored 1–10 against published anchors, totaled out of 80. 56–80 reads Ready (O) or Production-ready (T); 33–55 reads Foundational work first; 8–32 reads Survival risk (O) or Not yet safe to widen (T).

O · 1
Organizational Drag
How many hands a decision needs before it moves.
O · 2
AI Elevation
Whether AI has an owner above team level, or is scattered experiments.
O · 3
Work Architecture
Whether work is organized around whole jobs, or tasks an agent could take.
O · 4
Firm Boundary Design
Whether people can move to new work freely, or are locked to fixed roles.
O · 5
Decision Autonomy
How much runs without a human signature.
O · 6
Network Structure
Whether the structure lets work move sideways, or funnels every call upward.
O · 7
Reinvention Cadence
Whether you audit and retire your own workflows, or old steps just accumulate.
O · 8
Tacit Knowledge Accessibility
Whether operational knowledge lives in systems, or in people's heads.
T · 1
Leakage Prevention
Whether AI is blocked from data and systems it isn't authorized to touch.
T · 2
Workload Identity
Whether each agent acts under its own identity, with every action recorded.
T · 3
Reversibility
Whether a wrong action can be undone and the prior process restored.
T · 4
Human Accountability
Whether a named person is accountable for what the AI decides.
T · 5
Intelligence Stack Maturity
Whether AI runs on one shared operating stack, or scattered tools.
T · 6
Data Legibility
Whether business knowledge is accessible in data and code, or locked in people's heads.
T · 7
Autonomy Discipline
Whether clear limits exist on how far AI can act on its own.
T · 8
Observability
Whether you can test AI, log what it does, and catch when it drifts.

The read is deliberately directional, not flattering. Most companies discover they've been funding the platform while the organization is the actual wall. The full scored instrument — the Readiness Score proper — comes in the Roadmap.

01

Discovery Call

Half an hour
FREE
↳ before Backcast begins

A conversation that opens with your diagnostic already in hand. We walk through what your weakest area means and what a first move looks like, and ask about your size and systems, enough to shape a first project before you've spent a dollar. This is where we explain what the paid steps cost.

HOW THE THREE MOVE HERE
H — is leadership ready to help people rise, not just buy tools?
O — is there a real mandate and a first workflow worth reshaping around?
T — what shape is the platform in — ready to build on, or is there groundwork first?
DECISIONA low-key gut-check, both ways — is this worth a closer look?
Go deeper — what the call is really forShow less

The half-hour is about clarity

We open already holding your read, so we spend the time helping you understand what your result means for you, and what a sensible first move looks like. No pitch deck, no pressure to decide on the call.

A light first check

One person on our side gets a general sense of fit — a real organization, someone with the standing to sponsor a first project, and a gap worth exploring. It's a conversation, and it isn't the only look you'll get: the Roadmap is where fit gets tested for real. An honest "not quite yet" here costs you nothing.

You leave knowing roughly what a first step would cost and involve — enough to decide without pressure.

02

Roadmap

A defined engagement
PAID
↳ Backcast & Define + Assess & Prepare

A structured read across all three dimensions and a plan: your destination, your full Readiness Score, and the shape of your journey — first project named, the waves after it roughed in.

HOW THE THREE MOVE HERE
H — which people rise to govern the first workflow, and what that opens up for them.
O — how the structure around that workflow gets reshaped for agents to carry it.
T — what the platform becomes so it can carry that workflow — designed in step with the new structure, not after it.
DECISIONYou know your first project and roughly what the whole thing looks like — enough to start, or to stop having invested very little.
Go deeper — the destination, the full score, and the deployment modeShow less

The destination, named

We stand at the finished organization and work backward. The output is a signed Destination Architecture — your purpose, written plainly enough that an agent could tell what to build and what not to; your intelligence engine; and your organizational form, including how agent decisions stay chained to a named human.

Five conditions have to hold for that architecture to count as done: agent-first workflows, a system that improves from its own results, ownership of your own models and data, judgment built into every layer, and human flourishing as a hard requirement. Miss one, and this step isn't finished yet, however good the rest looks.

The full Readiness Score

The Diagnostic's directional read becomes the real instrument: all sixteen areas, scored and banded. The honest number matters more than the flattering one — a high score with nothing deployed is the most expensive place to be in this framework.

The route: four waves, easiest first

Wave 01
Foundation
The platform groundwork agents need to reach the work.
Wave 02
Readiness
The organization catches up — governance floor, people rising.
Wave 03
First Operation
The first workflow runs for real, under governance — the pilot.
Wave 04
Governed Scale
The rest migrates; change becomes the operating rhythm.

Deployment mode — set by headcount, not preference

A large organization resists a transplant; a small one doesn't. The Roadmap sets which mode fits: Direct Mode (50 or fewer — the whole company, in place, no twin needed), Light Edge Mode (50–500 — one Edge Twin around your highest-coordination workflow), Full Edge Mode (500+ — the twin built outside normal reporting lines, on CEO or board budget), or Mandatory Edge Mode (government/public sector — the only path, adapted for procurement and compliance).

The Roadmap gives you the shape. The deeper Blueprint, alongside the Pilot, draws the actual route.

03

Pilot

One workflow · parallel run · the real work
PAID
↳ Build & Prove

We build your first project: the Intelligence Stack, stood up for one workflow. It runs alongside your existing way first — the new version proposes, your people still decide, and the gap between the two is the training signal. Proof is the override rate falling over time.

Safety floorBefore any part of the system acts on its own, all four GOVERN / ASSURE pillars clear a 3 — every action tied to a named person, every change recorded, nothing that can't be undone.
HOW THE THREE MOVE HERE
H — your people step up into governing — the override rate falling is them rising.
O — the reshaped structure gets tested for real: what the agent may propose, what the organization permits.
T — the platform carries it in step with that structure — identity, record, integrity on every action.
Go deeper — the Intelligence Stack, and how "proven" is measuredShow less

The operating core

The Intelligence Stack runs the decision loop continuously — sense, interpret, decide, act, learn — with GOVERN/ASSURE running across all of it, never off.

Every agent ships with a spec

No spec, no agent. Before it runs, each agent gets a written spec — what it's for, how much it can decide alone, what it can access, and who's accountable if it's wrong — so authority is always visible, never assumed.

Parallel first

The new workflow runs on the same real inputs as the old one. It proposes; your people still act. Where the two disagree is the most valuable signal in the whole engagement, captured along with the reason.

Authority widens as it's earned

The rail matures in three stages — alert-only, escalation authority, kill-switch capability — each earned by the Four Pillars, never by the calendar. As the override rate falls, the workflow earns more authority, one tier at a time, with rollback always in place.

See the pilot in full — start with one workflow →
04

Blueprint

Runs alongside the Pilot
OPTIONAL
↳ Extract + Diagnose & Strip

The deep version of the Roadmap — a full read of your people and technology, and the sequencing of every wave after it. Its signature move is reconciliation: where two parts of the business disagree about who controls what, that disagreement is the finding. Most clients run it alongside the Pilot, so the Pilot has real evidence to build on instead of a guess.

Go deeper — Extract, Diagnose & Strip, and reconciliationShow less

Extract — get the judgment while it's still in the room

An organization's real advantage is mostly tacit — carried by whoever has done the work for years. The first agent any person meets is an interviewer, gathering how they actually work, and what comes out gets written into a one-page manifest per workflow: what it is, who says so, and what happens if it's wrong. If a reason can't be stated, the data doesn't move.

Diagnose & Strip — score every task, skip no steps

Every role touching the workflow gets broken into tasks, each scored 1–5 for agent-readiness. 4s and 5s go to agents now, 3s pilot in the parallel run, 1s and 2s stay human — and headcount reduction doesn't count as architecture until the absorption math is actually done.

Where the two reads of the same area disagree, that gap is the finding:

Technology says
"Operating knowledge is codified and machine-readable."
Organization says
"What people actually know still lives in heads and message threads."
Resolved → The platform is legible; the judgment that keeps it accurate was never captured. Extraction becomes the immediate task, slotted into an early wave.

No single-surface consultant can produce that — it takes both reads to see the seam. Reconciliation is how the Blueprint turns two scores into one sequenced route.

05

Scale

The remaining waves
PAID
↳ Build & Prove, continued — wave by wave

Once the Pilot proves, we run the next workflows — easiest first, one or two at a time, each proven before the next begins. Continuous change becomes how you operate, not a project that ends.

HOW THE THREE MOVE HERE
H — more people move up into governance as more work migrates to agents.
O — the structure keeps re-shaping around everything agents now carry.
T — the platform widens what it safely permits as trust is earned.
DECISIONEach wave sized and agreed before it starts.
Go deeper — how the people actually riseShow less

Elevation, first

Most transformation stories have one people-plan: reduce headcount. Ours is different — as agents take the volume, the question becomes whose judgment expands to fill the space that opens up. We map whose judgment grows, keep the human layer working the exceptions (their reasons are the highest-value signal the system produces), and make who-decided-what visible.

People who ran a process by hand become the ones who govern the system that now carries it. That's the rise, made concrete.

06

Rewire

You own it now
YOU OWN IT
↳ Rewire & Evolve

The old org chart — a latency map of how information waits — gets replaced with pod-based networks organized around capability. The framework, the sequenced route, and the discipline of proving-before-trusting are all yours now. Your people run the reviews. We step away.

Go deeper — the fade, and what you keepShow less

The engagement is designed to end

The goal was never dependency on us. The mature state is a standing loop your people run: spot the next workflow ready to move, prove it in parallel, migrate it, repeat — self-disruption as a standing capability.

The governed core, the route, and the review discipline are yours, with no vendor lock. The moat is what a competitor can't buy: your way of operating and your people's judgment, fused into the engine.

THE SAFEGUARD · GOVERN / ASSURE

One rail. It never turns off.

GOVERN / ASSURE runs underneath every step, from Backcast through Rewire — never a phase you pass and leave behind. It logs every decision, holds the guardrails, and owns the kill switch. Four pillars make it concrete:

Guardrails WHAT EACH AGENT MAY DO

Authority granted per agent, task by task, and enforced by the system — so overreach simply isn't possible.

Segregation HOW AGENTS WORK TOGETHER

One agent can ask another for help — but cannot borrow its authority to do it. Overstep is caught at the boundary between roles.

Oversight THE HUMAN WATCH, EXTENDED

Monitoring agents watch for violations in real time and surface to a human — never act alone, so the watch scales with the work.

Accountability THE DECISION, OWNED

Every consequential move is specified, logged, and defensible before it happens — chained to a named human owner.

Stage 1 · Alert-only — observes, flags, humans act Stage 2 · Escalation authority — can pause and route to review Stage 3 · Kill-switch capability — can halt an agent class on its own

Authority isn't granted up front — each stage is earned as the Four Pillars prove out inside the Pilot.

A deployment gate, not four features. Each pillar is scored 1 to 5, and the rail is capped at its lowest score — no new class of agent reaches production until every pillar clears 3. A strong score in one can't cover for a weak one elsewhere.

The Discovery Call's gut-check and the Roadmap's signed mandate are the human side of this; the rail is what they're judging once work begins.

WHAT EACH STEP ASKS OF YOU

Two free steps to your first honest read.

From the Roadmap on, they're paid engagements — walked through on the Discovery Call before you commit to anything. We size the work with you, never off a price list.

DiagnosticFREEFive minutes, on your own
Discovery CallFREEHalf an hour — and where we explain what the paid steps cost
RoadmapPAIDA defined engagement to get your plan and first project
PilotPAIDThe first real build — scoped with you, no surprises
BlueprintOPTIONALRun alongside the Pilot if you want the full sequencing up front
ScalePAIDPriced and agreed wave by wave, never as one lump commitment
RewireYOU OWN ITNo fee — this is the handoff, and the point of the whole engagement
WHAT YOU'RE LEFT WITH

A governed organization you own — and people who now operate higher.

The goal was never dependency on us. As the pilots prove and the waves migrate, the whole thing becomes yours: the governed operating core, the sequenced route, the discipline. Your people run their own reviews now, spending their time on judgment and creativity while agents carry the heavy lifting. We fade.

Anyone can buy the same engine now. The materials are available to anyone; what makes it yours is the structure you raise on your own foundation, and the people who govern it. Owning your intelligence is owning your differentiation. Renting it is renting your commoditization.

START WITH WHERE YOU STAND

The HOT scan reads all three dimensions.

Five minutes, your scores, no email wall. The gap is rarely where the budget went.

Run the HOT scan → ~4 MINUTES · NO EMAIL