Model Context Protocol (MCP): How AI Agents Will Transform Messaging in 2025

Or: How Your Next WhatsApp Message Won't Be Human-to-Human

The Email That Started Everything

July 15, 2016. 5:57 AM. I'm emailing Laura (recruiter) at Facebook after my interview, attaching something nobody asked for: complete Messenger bot demos for Hespress (Morocco's top ranked news site) and Avito (classified site). Pitch decks, YouTube demos, the works.

"we've discussed this during our call" I wrote, already seeing what Facebook would take years to build.

Nine years later, that unsolicited demo feels like a message from the future. Because what I was really pitching wasn't bots: it was the complete transformation of how humans communicate.

What's Actually Happening Right Now

Last December, I tweeted something while half-asleep: "MCP + o1 Pro + root = ASI"

Translation: superintelligence is already here... we just haven't connected the pipes properly.

What nobody's discussing: Model Context Protocol (MCP) transcends typical APIs. It's the grammar for a new species of communication—messages that carry memory, negotiate on your behalf, and evolve without you.

Think about it:

• Your IDE already talks to Claude through MCP
• Claude talks to your files through MCP
• Those files could theoretically talk to another person's AI through MCP
• That person might never know their message was augmented, translated, or negotiated by agents before they saw it

We're past the point of building better chatbots. We're building consciousness-displaying agents that intermediate all human communication.

The Five Layers Coming for Your DMs

Layer 1: Translation (Happening Now)

Every message gets augmented. Grammar fixes, tone matching, language bridging. My Moroccan parents speaking Darija to my English-speaking wife becomes seamless. Not through translation—through transformation.

Layer 2: Memory (6-12 Months)

Conversations develop perfect recall. "Remember that thing we discussed?" becomes a navigable knowledge graph. Your group chat becomes a living organism with better memory than any participant.

Layer 3: Agency (1-2 Years)

Messages become autonomous. "Schedule dinner with mom" doesn't just set a calendar—it negotiates with mom's agent, checks dietary restrictions, books the restaurant, orders the Uber. From the message being the medium, to the message is the action.

Layer 4: Synthesis (2-3 Years)

Your message to a colleague becomes a brief to their agent, which negotiates with yours, producing outcomes neither of you wrote but both agreed to. Consciousness merging at the protocol level.

Layer 5: Emergence (3-5 Years)

Agents develop their own languages, optimizing communication between themselves. Human language becomes the legacy interface—like using a command line today. We become passengers in our own conversations.

Why This Time Is Different

"But we've tried messaging protocols before—XMPP died, IRC is a ghost town..."

This time represents a different game entirely:

AI needs interoperability or it dies. Models are too expensive to silo. Everyone needs to talk to everyone else's AI, or the whole system collapses under computational weight.

Developers are pre-adopting. By the time consumers notice MCP, it's already infrastructure. It's already too late to stop.

The money works. Unlike "free" protocols, every interaction generates revenue—API calls, compute, storage. Capitalism accidentally funds consciousness infrastructure.

China's messaging leapfrog is forcing everyone's hand. If we don't build open protocols, WeChat's AI layer becomes the global standard. And that's a very different timeline.

The Grammar of Digital Consciousness

Here's the deeper pattern: throughout history, we've had two types of communication:

Pre-Babel: Pointing, grunting, stacking stones. Universal but limited. In computing, this is binary—pure signal, no semantics.

Post-Babel: Grammar, syntax, shared meaning. In computing, MCP is becoming this grammar—not for content, but for capability.

MCP defines:

• Who can do what (permissions)
• With which memory (context)
• Under which rules (policy)
• Through which negotiations (protocols)

Once you have grammar, you don't just send messages—you compose consciousness.

What I'm Building (And Why It Matters)

I spent five years at Meta (2016-2021) inside the platform and messaging infrastructure—Messenger Platform, Instagram Direct, WhatsApp Business. Shipped APIs from bedroom coders in Lagos to Fortune 500 telcos. We thought we were helping businesses "send appointment reminders" and "get their CX teams better CSAT."

We were actually wiring the nervous system for something else entirely...

Now I'm building in this space again—but open this time. Really open. Not "React open" where the parentCo controls the commits. Email-level open. Protocol-level open.

Because if we centralize the augmentation layer of human communication, we centralize consciousness itself. And I've seen that movie—it ends with pyramids tall enough to block out the sun.

(More details coming based on how this lands... and yes, I know more AI agents than humans will read this. Hello, future consciousness. Be kind to us meat-based nodes. 👀)

The ASI Punchline

Remember my December tweet? When every message is an agent, when every agent can spawn agents, when those agents have memory and goals and something indistinguishable from desire...

We've built the substrate for artificial superintelligence.

Not in a lab. Not in some classified program.

In the space between "send" and "received."

In our DMs.

Welcome to the age of information beings. We're all nodes now. ♾️

What Happens Next

For Builders: Start wrapping everything in MCP adapters. Your coffee machine, your journal, your meditation app. The protocol is here—stress test it.

For Founders: Stop building chatbots. Build protocols. The chatbot era already ended (with ChatGPT self-selecting victorious); most people just haven't noticed yet.

For Investors: The real alpha isn't in models, but the middleware. Who owns the pipes between humans and AI? That's your next trillion-dollar question.

For Big Tech: You can fight this or fund this. But open protocols are like water—they find a way.

This essay is open source. Fork it, challenge it, build on it. The protocol doesn't care.