Mule Agent Service Policy Reference

Pricing Tiers (reference, by scale)

Deployment fee includes software license and implementation (environment planning, deployment, IM integration config, integration testing, acceptance). Actual pricing depends on team size, number of IM platforms, knowledge base volume and deployment complexity — per signed proposal.

The following are customer-managed (not included in any fee)

• Server / hardware: customer-provided; see the deployment page for reference config. We can assist with selection; hardware billed at invoice.
• LLM API costs: the model account is registered, topped up and owned by the customer. We do not recharge, bundle or mark up.
• IM platform enterprise qualification: enterprise certification fees for WeCom / Feishu / DingTalk (where applicable).
• Public resources: domain, SSL certificate, cloud bandwidth (as needed).

Annual service fee includes (unmetered, unlimited)

These are the "keep the system running" base services:

• Fault handling: troubleshooting and repair.
• Version upgrades: upgrade deployment after a new release.
• Routine inspection: periodic system health checks.
• Usage Q&A: administrator daily usage support.
• Account/permission routine changes: add/remove users, basic permission adjustments.
• Backup checks: periodic verification that backup strategy is effective.

Specific frequency and scope per signed proposal.

Work-hour package (metered service)

"Extra labor" is deducted from a work-hour package, covering: adding new IM platform integration, bulk knowledge base organization, permission structure re-setup, small feature optimization, server migration/expansion, data recovery assistance, etc.

• Quote before work: each request is assessed for hours (min 0.5 person-day), then confirmed before starting.
• Deducted as used: remaining balance is queryable, reported quarterly.
• No carryover: valid for the year, reset at year end.
• Overage top-up: purchase additional at per-person-day rate (rate per signed proposal).

Each tier's work-hour allowance (e.g. person-days for Starter / Standard / Enterprise) is specified in the signed proposal.

Customization principles

To avoid unbounded customization, we use a product-roadmap model:

• Feature requests are assessed and answered clearly: "do / don't / when".
• Accepted requests enter the product roadmap and are shared by all customers on upgrade, with no extra fee.
• Enterprise benefit: requests get priority assessment and scheduling.
• If dedicated development is truly needed (won't wait for a release): separate project, separate quote, deliverables owned by the customer.

Response times (targets)

These are targets, not contractual guarantees; actual per signed SLA. P0 = system fully unavailable, no user can work.

Cases outside fault handling

• Customer server hardware failure, hosting/cloud outage, customer network issues.
• Customer accidental data deletion (recovery assist if backup exists; no liability without backup).
• LLM API provider-side failures, rate limits, price changes.
• IM platform API changes requiring adaptation (billed as work-hours).
• Anomalies from customer modifying system files/database structure (recovery billed as work-hours).