Note

Directional analyst estimates from public information & operating-model analysis — not measured robot performance. Read methodology

Methodology

How the Atlas works

We start with how work happens, then map where humanoids may fit, what evidence exists, and what proof is still missing.

01 · Method

One method, every industry

Each industry is decomposed into operating cells, scored directionally, checked against evidence, and tied to proof gates.

01

Industry

We start from a real sector and its operating reality — not from a robot looking for a job. Each industry is decomposed before anything is scored.

02 · Decompose

From sector to testable work

The Atlas narrows a broad industry into a bounded unit of work that can be assessed, piloted, and eventually proven.

Industry

Logistics & fulfillment

Operating model

3PL inbound → storage → outbound flow

Operating cell

Inbound carton decanting

Robot job

Move & decant standardised cartons to a conveyor

Readiness band

Directional fit — first / next / watch / no-go

Proof gate

Sustained pick rate, zero safety incidents per shift

One worked path: Logistics & fulfillment, narrowed to a single provable cell.

03 · Cell

Anatomy of an operating cell

A cell is the smallest useful unit in the Atlas: a specific work context, task bundle, robot job, blocker, and proof gate.

Operating cell · example

Inbound carton decanting

Input / receiving · Warehouse inbound dock · 3PL / fulfillment

Robot job

Move and decant standardised cartons onto a conveyor or tote.

04 · Score

Directional fit score

Fit estimates suitability. Evidence tracks proof. They are separate.

input

Value

input

Feasibility

input

Strategic fit

÷

Risk

output

Opportunity

Value × Feasibility × Strategic fit ÷ Risk = Opportunity. Public pages speak in bands only — exact weights and sub-scores stay internal.

Directional fit

0–100

An estimate of fit — not a deployment claim.

Readiness band

First · Next · Watch · No-go

Where the cell sits today.

Evidence level

Separate

Never folded into the score.

Fit estimates suitability. Evidence tracks proof. They are separate.

05 · Evidence

The evidence ladder

Evidence tracks what has been publicly demonstrated — a separate axis from fit, read with the same hex ring everywhere in the Atlas.

  1. Scaled rollout

    Deployed and scaled across multiple sites.

  2. Measured operation

    Operating with publicly reported KPIs.

  3. Named deployment

    A named, ongoing deployment at a real site.

  4. Pilot

    Named customer pilot in a real operating environment.

  5. Controlled demo

    Shown in a staged or controlled environment.

  6. Vendor claim

    Claimed by a vendor without independent confirmation.

  7. Conceptual

    Analytical reasoning only; no public demonstration.

  8. No public evidence

    Nothing publicly demonstrated or reported yet.

06 · Alternatives

Humanoids vs alternatives

Every cell is compared against purpose-built automation, process redesign, and human-only operation.

Humanoid likely wins

Human-scale spaces, mixed objects, several adjacent tasks for one body.

Mobile manipulatorCobot workcell

Depends / mixed

Outcome hinges on speed, safety validation, or integration specifics.

AMR + armProcess redesign

Alternative likely better

A purpose-built machine clears the task faster, safer, or cheaper.

AMRFixed automationDroneSpecialized robotHuman-only for now
07 · Trust

Trust protocol

Public pages show directional analysis. Exact weights, raw scores, source mapping, and analyst notes stay internal.

What we claim

Directional analysis based on public information and operating-model research. Public pages speak in readiness bands and evidence levels.

  • Where a humanoid form factor may fit, cell by cell
  • What has been publicly demonstrated, on a separate evidence axis
  • What proof gate would still have to clear at a real site

Submit a correction →