Success rate
Can the robot complete the task often enough?
The evaluation starts with the pass bar the robot team actually needs.
Real-site robot evaluation
Evaluate robots on real sites before deployment.
Test policies and scenarios against capture-backed sites before field time. Robot teams pay for evaluations and optional data exports; site operators participate free.
Cycle time
Can it keep up with the site rhythm?
Blueprint frames the task against target timing, bottlenecks, and site drift.
Intervention rate
Where will people still need to step in?
Blueprint keeps assist points visible instead of hiding them behind a score.
Safety threshold
What still needs review before field exposure?
Safety stays scoped to the request and does not become a blanket validation claim.
What Blueprint sells
Real-site evaluations for robot teams.
Blueprint turns capture and pipeline evidence into a scoped evaluation workflow: what the site is, which scenarios matter, how policies behave, and which proof or data export the team needs next.
01
Real-site evaluation
A fixed-scope review for one site, robot task, policy/profile, target thresholds, and missing proof.
02
Scenario tests
Task variations, start states, dynamic conditions, object zones, failure cases, and observed-vs-inferred labels.
03
Site package
Capture-backed world model, walkthrough media, geometry when available, provenance, rights posture, and export limits.
04
Training exports
Exportable data for post-training, fine-tuning, regression checks, and site-specific model improvement.
How it works
Turn a real site into an evaluation plan.
Blueprint keeps the workflow compact: real site, robot task, scenario variation, policy run, exported evidence, and the next proof step.
Capture a real site
Start from a lawful capture, existing site package, or structured request for the facility in question.
Scope the evaluation
Set the site task, robot profile, thresholds, scenario variations, provenance, rights labels, and proof boundaries.
Run the robot policy
Use a manual browser session or a headless agent path to test the robot profile against site tasks.
Export the evidence your team needs
Package observations, scenario results, failure cases, and optional data bundles for training or fine-tuning.
Decide the next step
Proceed to a short pilot, request more proof, tune on the exported set, or hold until missing evidence clears.
Planning ranges
Two paid robot-team products. Operators are free.
Robot teams pay for evaluations first, with optional site packages and data exports. Operators can submit sites and define boundaries without paying Blueprint.
Policy Evaluation Set
$6,500 / site evaluation
Robot teams run one policy/profile across the site's task suite by manual browser session or headless agent.
Request policy evaluationSite Data Package
$3,500+ / site package
World model, scenario set, provenance, rights labels, and data exports for post-training or fine-tuning.
Request site dataSite Operators
Free
Operators can submit a site, define access and privacy boundaries, and participate without paying Blueprint.
Submit site freeProof boundary
Public samples show the workflow. Request packets prove one site.
Blueprint can look ready and polished without pretending a robot has passed deployment, safety, payment, provider, rights, or hosted-session checks that still need owner-system proof.
Sample
Public product samples
Stronger proof
Request packets
Samples show the product shape. Request packets prove one site with provenance, rights, thresholds, and gaps attached.
Sample
Generated or model-derived visuals
Stronger proof
Owner-system evidence
Generated outputs can support review, but simulator traces, action logs, robot trials, safety review, rights proof, and runtime artifacts own stronger claims.
Sample
Evaluation output
Stronger proof
Operational readiness
Policy-evaluation output stays advisory until the missing proof exists for that exact site, robot, task, and threshold set.
First request
Ask for one real site to evaluate before deployment.
Bring the facility, task, robot profile, target thresholds, timeline, and proof you already have. Blueprint routes the next step to a real-site evaluation, site package, capture ask, or proof blocker.
Request evaluationRequests do not grant package access, rights clearance, payment, fulfillment, or hosted-session availability by themselves.
Policy-evaluation output remains advisory until simulator traces, action logs, robot trials, safety review, rights proof, and runtime proof support a stronger claim.
Site operators can submit and govern a site for free; paid usage starts with robot-team evaluations or optional data access.
Generated imagery on the public site is illustrative, not customer or robot-trial proof.
Public Launch Ready copy is allowed. Operational Launch Ready claims still require proof from the system that owns them.