Buy the operation.
Not just the robot.

Write requirements in operational language

Describe the task before selecting a platform. Include the site, permitted routes, operating hours, required observations, response workflow and constraints around people or vehicles. Separate mandatory requirements from desirable features so a persuasive demonstration cannot displace the original need.

Ask suppliers to state exceptions explicitly. A response of “supported” is incomplete if a function needs an additional license, continuous connectivity or a building integration not included in the quote. Request the proposed configuration, software dependencies and evidence supporting each important claim.

Compare against alternatives, including improved lighting, fixed cameras, access-control changes and revised patrol procedures. The decision is not simply one robot versus another. A less mobile solution may meet the requirement with fewer operational dependencies.

Understand what the commercial model includes

An outright purchase can provide asset ownership, but software, connectivity, support and monitoring may remain recurring costs. Leasing spreads equipment payments while introducing term, return-condition and early-exit obligations. A managed subscription may bundle more services but make the operation dependent on the provider’s continuing availability.

Do not assume “robot as a service” has a standard definition. Establish who owns the hardware, who operates it, who repairs it and who pays for replacement batteries or accidental damage. Identify whether a replacement unit is guaranteed during extended repairs.

There is no universal price suitable for these guides: configurations, regions and service scopes differ, and no current quotes have been verified here. Obtain written, itemized proposals for the same requirements and compare tax, currency, contract duration and excluded work consistently.

Build a lifecycle cost sheet

Use a common evaluation period and list one-time implementation costs separately from recurring costs. Include staff time rather than treating existing employees as a free resource. Mapping, network improvements, privacy review and operator training can materially change the implementation effort.

Test assumptions about utilization and downtime. A cheaper proposal can become expensive if staff must repeatedly recover the machine. Conversely, a comprehensive service fee may include work that another quote leaves to the customer. Compare equivalent responsibility, not just the headline payment.

Make the pilot falsifiable

Agree pass criteria before equipment arrives. Define required mission completion, usable evidence, intervention logging and safe behavior during predictable faults. Choose thresholds based on your site’s risk and workload, not an arbitrary industry figure.

Record a baseline using the existing process, then observe the proposed workflow over representative operations. Include operator handovers, route obstructions, charging and approved loss-of-connectivity exercises. Have the relevant stakeholders witness critical tests rather than relying solely on a supplier’s summary.

Keep unresolved conditions visible. A mild-weather pilot cannot establish winter performance. Document exclusions, corrective actions and retest requirements. The contract should state whether pilot failure allows cancellation, modification or an extension without an unexpected long-term commitment.

Put responsibilities into the contract

Distinguish acknowledgment time from restoration time in service commitments. A quick support reply does not mean the robot is operational again. Establish support hours, escalation paths, spare-parts availability, update practices and the treatment of recurring faults.

Review data ownership, processing roles, hosting locations, subprocessors, retention and incident notification. Ask whether recordings or operational data can be used to train supplier systems, and require clear contractual limits where necessary. Ensure security updates and end-of-support notice are addressed.

Confirm insurance responsibilities and obtain qualified legal advice on liability allocation. Safety documentation, approved operating limits and maintenance obligations should match the actual configuration rather than a generic product brochure.

Plan the last day before the first

Ask how to export incident records, revoke vendor access, erase local storage and remove integrations. Establish what remains usable if a subscription ends or the supplier stops supporting the platform. Proprietary maps and cloud dependencies can make switching harder than moving the physical machine.

A sound purchase leaves the organization with a documented, supportable service and a credible exit. If the supplier cannot explain either, resolve that uncertainty before expanding the deployment.