Provider Models and Fallbacks

Selection criteria

Provider and model choice should be based on the work SophMate will perform: read-only Copilot answers, WooCommerce summaries, Theme Assistant design review, Image Studio assets, agents, workflows, and support drafts may have different cost, latency, and capability needs. Do not choose a model only because it is the newest or cheapest.

Fallback planning

Document whether the team should pause work, retry, switch provider, reduce workflow scope, or escalate when a provider fails. Pair provider choices with budget and usage controls, security and key rotation, and troubleshoot provider connections.

Production review

Review provider behavior after billing changes, model deprecations, high-volume campaigns, agent launches, workflow failures, and latency spikes. The provider setup tutorial covers first connection testing, but production teams should keep a written model ownership note.

Quick reference

• Use this page when changing provider, budget, role, mail, queue, cron, source, cache, or WooCommerce context settings.
• Do not store secrets, credentials, purchase codes, or private customer data in documentation notes or screenshots.
• Key decision: whether provider access, data handling, fallback behavior, and credential ownership are safe for the workflows being enabled.

Scope limits

• Do not use this page to approve a provider for sensitive data without reviewing provider policy, billing owner, fallback behavior, and key rotation responsibilities.
• This page does not replace provider, host, mail, billing, security, or legal policy review where those owners are responsible.
• Use it to frame configuration decisions, not to store secrets or justify broad access.

Owner and cadence

• Primary owner: site administrator with provider, billing, and security responsibility.
• Review cadence: after provider, mailbox, role, budget, security, WooCommerce, or integration changes.
• Escalate when provider choice affects privacy, cost, latency, model access, workflow reliability, or customer-facing output quality.

Access and data boundary

• Provider evidence should include account owner, model access, billing state, rate limits, fallback owner, and policy decision without exposing API keys.
• Restrict provider keys, billing, budgets, roles, mail routing, queues, cron, cache, and integration settings to accountable administrators.
• Document setting intent and validation results without storing API keys, mailbox passwords, purchase codes, raw logs, or private customer records.

Production checklist

• Document model purpose, capability, budget, latency, privacy expectation, fallback behavior, and owner before team rollout.
• Re-test provider behavior after billing changes, model deprecations, workflow failures, and high-volume campaigns.
• Document who owns provider credentials, budget limits, role access, notification routing, and ongoing review.
• Keep configuration changes behind administrator access and review them after plugin updates, staff changes, or incidents.

Acceptance checks

• The team knows whether to pause, retry, switch provider, narrow scope, or escalate when provider behavior changes.
• Provider/model choices are connected to budgets, key rotation, diagnostics, and support routing.
• A second administrator can explain why each high-risk setting is enabled and who may change it.
• No production credential, support mailbox, or notification path depends on an unmanaged personal account.

Failure modes to test

• Test invalid keys, revoked keys, missing billing, model denial, rate limits, fallback provider failure, and provider-policy mismatch.
• Test invalid credentials, exhausted budget, unmonitored mailbox, wrong role mapping, queue delay, cron failure, and cache-dependent behavior.
• Confirm configuration failures are visible to an owner without exposing provider keys, mailbox credentials, or raw logs.

Evidence to capture

• Capture provider account owner, allowed model, billing state, rate limits, data-handling decision, and fallback owner without storing API keys.
• Record changed setting, previous value category, new value category, owner, approval reason, and review date without storing secrets.
• Capture provider, budget, role, mailbox, queue, cron, cache, and WooCommerce evidence relevant to the setting being changed.

Decision record

• Decision field to include: provider account owner, allowed model, billing or quota state, fallback owner, key rotation status, and data-handling approval.
• Record the configuration decision, setting owner, previous setting category, new setting category, validation result, approval reason, and next review date.
• Include whether the change affects provider spend, privacy, mail delivery, queue or cron behavior, WooCommerce context, or user permissions.

Stop or rollback path

Pause provider-dependent workflows when credentials, billing, rate limits, model access, data handling, or fallback behavior are unclear. Resume after diagnostics pass and affected workflows have an owner-approved restart note.

Monitoring window

• Monitor the next operating cycle for unexpected spend, blocked work, missed notifications, queue delays, or changed provider behavior.
• Compare the setting change against audit records, diagnostics, and user reports before making another related change.

Expansion criteria

• Provider-dependent workflows can expand only after model access, billing, rate limits, data handling, and fallback behavior are accepted.
• Widen the setting change only after audit records, user reports, diagnostics, and cost or delivery signals match expectations.
• The owner can explain how to reverse or narrow the change without searching through unrelated settings.

Common mistakes

• Switching providers for cost without checking model access, quality, latency, privacy, and fallback behavior.
• Using personal provider keys, personal mailboxes, or broad administrator access because it is faster during setup.
• Changing budgets, roles, notifications, or integrations without recording the owner and review reason.

Common questions

Who should approve provider or credential changes?

A site administrator or provider owner should approve the change after model access, billing, data handling, fallback behavior, and affected workflows are understood.

Who should own this decision?

The site administrator should own the decision, with provider, billing, security, mail, or hosting owners involved when their systems are affected.

What should stop the rollout?

Stop when a setting changes spend, access, privacy, mail, queue, cron, cache, provider behavior, or WooCommerce behavior without validation evidence.

Related operations

• Start with Connect an AI Provider.
• Pair configuration work with Roles and Permissions.
• Review Approval Controls before enabling write-capable modules.
• Use Cost Allocation and Client Billing Review before client or team billing reviews.
• Use Cache Queue and Performance before scaling workflows or alerts.
• Use Scheduled Task and Cron Reliability before relying on recurring work.
• Use Data Residency and Provider Policy Review before sending sensitive context.
• Use Provider Rate Limits and Retry Planning before high-volume workflows.