What is a self-hosted presentation API?

A self-hosted presentation API runs the generation service in infrastructure your team controls. Applications can send prompts, documents, or structured content to create editable PPTX or PDF files without relying entirely on a managed presentation SaaS.

When self-hosting is a good fit

• documents must remain in a controlled network;
• the organization requires a specific cloud, region, or on-premise environment;
• teams need local models such as Ollama or an approved provider gateway;
• custom integrations or identity controls require infrastructure access;
• volume makes predictable infrastructure economics valuable;
• the organization can operate the service reliably.

Managed Cloud remains the simpler option when fast integration, managed upgrades, and lower operational overhead matter most.

Reference architecture

Internal application
  → gateway / load balancer
  → authenticated Presenton API
  → job workers
  → text model + image provider + optional search
  → database and protected object storage
  → monitoring, backups, and audit logs

Draw the complete data path. A locally hosted API can still send content to an external text model, image service, or web-search provider.

Deploy with Docker

Presenton provides a Docker-based self-hosting path. Pin a tested image version, use persistent volumes, store configuration in a secret manager, and place the service behind HTTPS and authentication. Start with the documented system requirements rather than guessing CPU, memory, storage, and optional GPU capacity.

Do not expose a development container directly to the public internet. Use a controlled network boundary and restrict administrative surfaces.

Choose model and media providers

Document which components receive user content and what they retain.

Security checklist

• require authentication for the browser and API;
• rotate API keys and separate environments;
• apply tenant authorization to presentations and files;
• restrict outbound network access where practical;
• scan and limit uploads;
• encrypt storage and transport;
• redact secrets and private content from logs;
• apply retention and deletion policies;
• patch images and dependencies on a tested schedule.

Operate the service

Monitor request rate, queue depth, generation latency, error rate, model failures, storage growth, export failures, and resource saturation. Back up the database and required persistent assets, then test restoration. A backup that has never been restored is only an assumption.

Create a staged upgrade process with representative prompts, documents, templates, and output comparisons.

Keep self-hosted and Cloud routes separate

Presenton Cloud currently uses a /api/v3 route family. The self-hosted API uses its own base URL and /api/v1 routes. Request fields and operational assumptions differ. Follow the documentation for the deployed edition and keep the base URL and version in configuration.