Charlie, your AI assistant inside CloudContactAI, doesn’t just answer questions — it takes action in your account. Under the hood, that’s powered by MCP (the Model Context Protocol), a standard way for an AI model to reach out and call a specific tool — look something up, update a record, send a message — rather than just generating text. Over the past few weeks we’ve expanded the set of tools Charlie can reach for, and added the ability for you to build your own.
Charlie Can Now Catch You Up on Your Inbox
Ask Charlie about a conversation and it can now actually go look. New inbox tools let Charlie list recent conversations, pull up a specific thread, and summarize what’s been said — so instead of scrolling back through a long SMS or email thread yourself, you can ask “what did this contact say last week?” and get a straight answer.
Charlie Can Search the Web, When It Makes Sense
Building on the web search capability we rolled out last month, admins can now choose which provider powers it — Tavily or Amazon Bedrock’s built-in search — from global settings. Charlie reaches for the web only when a question genuinely calls for current information, like up-to-date compliance guidance or industry best practices, rather than relying solely on what it already knows.
Build Your Own Agent on Top of Our Tools
This is the big one: you can now configure a custom agent with its own scoped set of tools that acts directly on your contacts from within a live SMS conversation. A contact can text in and your custom agent can, in the moment, look up their contact record, add them to a list, leave a note on their file, or update their opt-in/opt-out status — all without a human needing to jump in.
Every action a custom agent takes is scoped to the exact phone number that’s texting in and logged for audit, so you always have a clear record of what the agent did and why, even when it’s running unattended.
More Context, Better Answers
We also taught the underlying messaging tool to recognize the direction of an SMS conversation — inbound versus outbound — so when Charlie answers a question about message history, it understands who said what, not just that a message exists.
Why MCP Matters Here
MCP gives us (and eventually you) a consistent, extensible way to hand an AI agent new capabilities without rebuilding how it reasons from scratch. Today that means inbox lookups, web search, and custom contact actions. It’s the same foundation we’ll keep building on as Charlie takes on more of the busywork in your account.
Have feedback on any of these changes, or a feature you’d like to see next? Reach out to your account team — we’d love to hear from you.