Two ways to give your AI chatbot the right knowledge.

Should Velaro host your content, or connect to the AI agent you've already built? A plain-language comparison for business readers.

What this page is for

If your company has already invested in a Microsoft AI agent - or is about to - you have a decision to make when you roll out customer-facing chat. Do you duplicate that knowledge inside Velaro, or do you connect Velaro to the agent you already have? Both paths work. Here is how to decide, without getting into the weeds.

Option A

Velaro ingests your documentation

One pipeline. Velaro manages content and answers end-to-end.

What it looks like

  • You point Velaro at your sources: website, PDFs, SharePoint/OneDrive, KB articles, videos
  • Velaro indexes everything and keeps it fresh automatically
  • Customer chats are answered by Velaro's AI layer using your content

Why teams pick this

  • One system to log into - content, bot, analytics, and handoff all live in Velaro
  • Tight control over what the bot can see - if you don't connect a source, it's invisible
  • Flat monthly pricing - your bill does not go up when the AI gets better
  • No dependency on another team's uptime

Trade-offs

  • Content is maintained in two places if your company also runs a separate Microsoft knowledge base
  • You don't get to use tokens you've already committed to buying from Microsoft
Option B

Velaro connects to your Microsoft AI agent

Your agent reasons. Velaro handles everything the customer touches.

What it looks like

  • Your Microsoft AI agent (Copilot Studio, Azure AI Foundry, or Azure OpenAI Assistants) exposes an MCP endpoint or REST API
  • Velaro's workflow engine calls your agent for answers
  • Velaro still owns the widget, conversation memory, handoff, reporting, and compliance

Why teams pick this

  • One source of truth - content lives where your IT team already manages it
  • You burn tokens you've already committed to, instead of paying twice
  • Governance (internal vs. external, approval workflows) stays inside Microsoft where your team already runs it
  • Your IT team keeps full control of the reasoning layer and model choice

Trade-offs

  • If your Microsoft agent goes down, the customer-facing bot is affected - Velaro configures a safe fallback for this
  • Slightly longer integration window while the MCP/REST connection is built and tested
  • Governance must be enforced inside your Microsoft agent, not in Velaro

Which one is right for you?

Option A (Velaro-hosted) is the right pick when your customer-facing team owns the content, when you want the simplest operational footprint, and when predictable flat pricing matters more than reusing committed tokens.

Option B (Connect to your agent) is the right pick when your IT team has already built an internal AI agent, when you have a committed Microsoft token pool, and when you want one knowledge base serving both internal and customer-facing channels.

Both paths deliver the same thing to your customer: a conversational AI experience that hands off cleanly to a live agent when needed. The difference is who owns the content pipeline.

Common questions

Which Microsoft services can Velaro connect to?

Any Microsoft AI agent that exposes an MCP endpoint or a REST API. That includes Copilot Studio, Azure AI Foundry Agent Service, and Azure OpenAI Assistants. The exact integration pattern depends on which one you use; we confirm that on the technical kickoff call.

Can we start with Option A and move to Option B later?

Yes. Option A is often the faster launch path. Once your Microsoft agent matures, we can swap the source without changing the customer experience - the widget, the handoff, the routing, and the reporting all stay the same.

What about token costs?

For customer-facing retrieval, token spend is small and predictable - a typical question consumes less than a cent of model time. The real cost driver is content maintenance. That's why the decision comes down to who owns the pipeline, not who burns the tokens.

Does Velaro charge per AI resolution?

No. Velaro charges by conversations and data volume. Competitors charge per AI resolution (Intercom Fin at $0.99, Zendesk at $1.50, HubSpot at $0.50) which means your bill spikes as the AI gets better at resolving cases. With Velaro, your bill does not change because your AI improved.

Want to see both paths walked live?

A 30-minute call, focused entirely on your scenario.

Book a personalized demo