Back to Blog

AI Agents Can Now Send RCS (But Can't Test It)

· Jenny

AI Agents Can Now Send RCS (But Can't Test It)

Last month, Infobip launched an MCP server that lets AI agents send RCS messages. Big news, right?

Here's what's weird: Those AI agents have no way to test what they're sending.

Unlike human marketers who can wait 8-16 weeks for carrier approval, AI agents are supposed to iterate fast. They're built to test, learn, and optimize. But with RCS, they hit a wall.

🤖 The AI agent writes a message
📱 It gets sent through the MCP integration
⚠️ But there's no preview, no device testing, no validation
📬 The end user receives something the agent has never seen

This is the paradox of AI agent messaging: We trust AI to write personalized messages at scale, but we have zero visibility into how those messages actually render.

The Challenge Isn't AI Capability. It's RCS Infrastructure.

AI agents need:

  • Device-by-device preview (Samsung, Pixel, iPhone all render differently)
  • Flow validation before submission
  • Instant feedback on rich card rendering
  • A/B testing capability without carrier wait times

Sound familiar? It's the same problem brands face—but multiplied. AI agents generate hundreds of campaign variants. Testing each one manually is impossible.

The platforms (Infobip, Klaviyo, Sinch) help you DESIGN campaigns. But nobody built the testing layer.

That's where we see the gap—and the opportunity.

What AI Agents Actually Need

Traditional brands might run 1-3 campaign versions. AI agents? They generate hundreds of personalized variants. And device fragmentation means the same RCS renders differently on Samsung, Pixel, and iPhone.

The multiplication problem: each variant Ă— each device = exponential testing needs

AI agents need:

  • Instant device-by-device preview (not 8 weeks)
  • Rich card validation (buttons, carousels, media rendering)
  • Flow testing (button clicks, user journeys)
  • A/B testing without carrier submission
  • Real-time rendering feedback

The Platform Gap

What existing platforms provide: Klaviyo, Infobip, TrueDialog, Sinch help DESIGN campaigns.

What's missing: testing, validation, preview capabilities.

Why platform vendors haven't solved this: They focus on differentiation rather than infrastructure.

The opportunity for testing-first solutions? Huge.

RCS X: The AI Agent Testing Layer

RCS X enables instant device preview for AI agents:

  • Flow validation before MCP submission
  • Batch testing hundreds of variants
  • Integration with AI agent workflows

Whether you're building customer service bots, sales assistants, or AI negotiators—test before you deploy. Because in AI agent messaging, what you don't know will hurt you.

What do you think: Is AI agent messaging ready for production, or is the testing gap a dealbreaker?

#RCS #AIMarketing #AgenticAI #MarketingTech #RCSX