Provide Giselle-managed model options

[shige]

Summary

I would like to see Giselle provide its own managed model options alongside the current BYOK (Bring Your Own Key) approach.

[giselles-ai]

Finished running flow.

Step 1
🟢	On Issue Created Status: Success Updated: Dec 10, 2025 2:46am
Step 2
🟢	Issue Analyzer Status: Success Updated: Dec 10, 2025 2:48am
Step 3
🟢	Create Issue Comment Status: Success Updated: Dec 10, 2025 2:48am

[giselles-ai]

Issue Classification

• Type: Feature request
• Priority: Medium
• Suggested Labels: feature-request, enhancement

Summary

The user is requesting Giselle to provide its own managed model options in addition to the current BYOK (Bring Your Own Key) approach. This would mean Giselle hosting/managing models rather than requiring users to supply and manage them themselves.

Recommended Action

• Acknowledge and categorize the request as a feature enhancement to Giselle’s model deployment capabilities.
• Gather clarifying details to scope the feature, such as:
  • Which models or model families should be offered as Giselle-managed
  • Regions and data residency requirements
  • Security controls (encryption at rest/in transit, key management, access controls)
  • Model lifecycle (updates, versioning, rollback, deprecation)
  • SLAs, uptime, and monitoring/observability
  • Billing and pricing model for managed options
  • Data handling and privacy implications (training data, inference data)
• Propose an MVP scope for initial implementation:
  • A limited set of Giselle-managed models or a standard hosting option
  • Provisioning of managed model endpoints via UI/API
  • Basic lifecycle management (deploy, update, rollback) and authentication
  • Observability (usage metrics, logs)
  • Clear distinction between BYOK and GIS-managed options in the UX and docs
• Create an accompanying design/PRD (product requirements document) and add to the backlog as an epic or initiative.
• Draft acceptance criteria and success metrics, e.g.:
  • Ability to deploy a Giselle-managed model endpoint
  • Security/compliance requirements met for the MVP
  • End-to-end lifecycle management supported
  • Transparent cost estimation for managed models
• Engage security/compliance teams early to assess risk and governance implications.
• If possible, solicit user feedback or a small pilot with customers to validate demand and requirements before committing to a full rollout.