Showing Enterprise AI Platform, slide 1 of 7.
01 — Governed publishing
Live assistant
Turn approved knowledge into a client-ready assistant.
Publish governed knowledge once, embed the experience in WordPress or another channel, and update the assistant without rebuilding the site.
- Versioned public widget contract
- Citations, history, feedback, and sessions
- Knowledge and behavior remain in Django
AI Publisher
Use the live, Django-published AI Publisher assistant. Knowledge, behavior, sessions, and citations remain governed outside WordPress.
02 — Structured data
Ready for client testing
Let teams ask governed questions of business data.
Test a read-only data assistant pattern for databases, CSV files, and governed datasets without turning the result into another spreadsheet interface.
- Synthetic sample questions for client testing
- Future AI Data Gateway connection
- Replaceable analytics-agent or workflow slot
Structured Data Analyst
This testable assistant uses the current Publisher runtime to explain the structured data experience. It is not connected to a database yet; the dedicated AI Data Gateway tool or analytics workflow can replace this template later.
03 — Customer support
Ready for client testing
Help customers self-serve from approved knowledge.
Test a support assistant pattern for product questions, order context, policies, escalation, and response drafting with honest action boundaries.
- Approved knowledge and honest answers
- Safe escalation instead of invented actions
- Replaceable support-system workflow slot
Customer Support Specialist
This testable assistant uses the current Publisher runtime to explain a governed support experience. It cannot see orders or take customer actions; a dedicated support agent or workflow can replace this template later.
04 — Customer support page
Embedded live page
Put the published support page inside any website.
Use the AI Token Guard product support page as a portable embedded experience, so visitors stay on the current site while the Django-published page owns the assistant, page content, and runtime behavior.
- Website page URL for visitors
- Django-published support page remains authoritative
- No provider keys, secrets, or AI runtime moved into the host site
05 — Enterprise data
Ready for client testing
Ask business questions across controlled data boundaries.
Test a read-only assistant pattern that can later connect to sanitized company data through approved, auditable access boundaries.
- Read-only testing posture
- Permission-aware future connectors
- Auditable requests and outcomes
06 — AI engineering
Ready for client testing
Turn a software request into evidence-backed delivery.
Test the planning, implementation, QA, security, review, evidence, and human approval path without mutating a public repository.
- Plan-to-review workflow visibility
- Evidence at every delivery gate
- No repository mutation in this test path
07 — Enterprise integration
Ready for client testing
Connect tools and knowledge without losing control.
Test how permissions, tool boundaries, auditability, and human approval become part of an enterprise AI product.
- Tool and permission boundaries
- Human approval for consequential actions
- Replaceable MCP architecture-agent slot
Governed by architecture
One clear boundary from presentation to runtime.
Presentation, safe configuration, loading, and consent-aware analytics hooks.
Versioned assistant identity, public widget, sessions, messages, and citations.
Knowledge, tools, policies, limits, behavior, and auditable outcomes.
Build the specialist next