Agent First pilots are designed to prove one bounded workflow with approved rules, a measurable baseline, and a clear continuation decision. Evaluation can start with fictional or de-identified material. Proprietary production client information waits for Security Readiness sign-off.
One bounded workflow family and one primary Client Profile by default. Additional Profiles or integrations are separately scoped.
The live production pilot clock starts only after operational readiness and, where proprietary production data is involved, Security Readiness Gate sign-off.
Per pilot agent per month at the Starter rate, with a 10-seat minimum ($690 per month). No Growth-tier discount during the pilot. No automatic renewal.
Agent First can be demonstrated, configured, benchmarked, and tested with fictional or appropriately de-identified material. Before proprietary production procedures, client-specific operating material, sensitive production context, or regulated information enters a live pilot, Agent First completes a focused security architecture/application-security review with an experienced security-minded SaaS engineer or appsec consultant.
Pick the workflow, sponsor, users and baseline.
Collect approved policies, scripts, contacts, services and exceptions.
Build and test with approved, fictional or de-identified material where appropriate.
Rule owner reviews what Agent First should say and do.
Required before proprietary production data and production live pilot use.
Bounded use, weekly review, issue tracking and measurement.
Baseline, results, limitations, issues and next recommendation.
Continue, expand, revise and retest, or stop.
Rules-heavy, source-ready, measurable, sponsored, and bounded operations where the correct answer changes by client, location, account, product, service or exception.