In a latest version of The Sequence Engineering e-newsletter, “Why Did MCP Win?,” the authors level to context serialization and change as a motive—maybe a very powerful motive—why everybody’s speaking concerning the Mannequin Context Protocol. I used to be puzzled by this—I’ve learn a variety of technical and semitechnical posts about MCP and haven’t seen context serialization talked about. There are tutorials, lists of accessible MCP servers, and way more however nothing that mentions context serialization itself. I used to be much more puzzled after studying by way of the MCP specification, wherein the phrases “context serialization” and “context change” don’t seem.
What’s happening? The authors of the Sequence Engineering piece discovered the larger image, one thing extra substantial than simply utilizing MCP to let Claude management Ableton. (Although that’s enjoyable. Suno, beware!) It’s not nearly letting language fashions drive conventional functions by way of a typical API. There isn’t a separate part on context serialization as a result of all of MCP is about context serialization. That’s why it’s known as the Mannequin Context Protocol. Sure, it gives methods for functions to inform fashions about their capabilities in order that brokers can use these capabilities to finish a job. But it surely additionally offers fashions the means to share the present context with different functions that may make use of it. For conventional functions like GitHub, sharing context is meaningless. For the newest technology of functions that use networks of fashions, sharing context opens up new potentialities.
Study quicker. Dig deeper. See farther.
Right here’s a comparatively easy instance. It’s possible you’ll be utilizing AI to jot down a program. You add a brand new characteristic, take a look at it, and it really works. What occurs subsequent? From inside your IDE, you’ll be able to name conventional functions like Git to commit the adjustments—not an enormous deal, and a few AI instruments like Aider can already do this. However you additionally need to ship a message to your supervisor and crew members describing the challenge’s present state. Your AI-enhanced IDE may have the ability to generate an electronic mail. However Gmail has its personal integrations with Gemini for writing electronic mail, and also you’d want to make use of that. So your IDE can bundle every little thing related about your context and ship it to Gemini, with directions to resolve what’s necessary, generate the message, and ship the message by way of Gmail after it has been created. That’s totally different: As a substitute of an AI utilizing a standard utility, now we’ve got two AIs collaborating to finish a job. There may even be a dialog between the AIs about what to say within the message. (And it is advisable verify that the end result meets your expectations—vibe emailing a boss looks like an antipattern.)
Now we will begin speaking about networks of AIs working collectively. Right here’s an instance that’s solely considerably extra complicated. Think about an AI utility that helps farmers plan what they’ll plant. That utility may need to use:
- An economics service to forecast crop costs
- A service to forecast seed costs
- A service to forecast fertilizer costs
- A service to forecast gasoline costs
- A climate service
- An agronomy mannequin that predicts what crops will develop properly on the farm’s location
The appliance would in all probability require a number of extra companies that I can’t think about—is there an entomology mannequin that may forecast insect infestations? (Sure, there may be.) AI can already do a superb job of predicting climate, and the monetary business is utilizing AI to do financial modeling. One might think about doing this all on a large “know every little thing” LLM (possibly GPT-6 or 7). However one factor we’re studying is that smaller specialised fashions usually outperform giant generalist fashions of their areas of specialization. An AI that fashions crop costs ought to have entry to a variety of necessary information that isn’t public. So ought to fashions that forecast seed costs, fertilizer costs, and gasoline costs. All of those fashions are in all probability subscription-based companies. It’s possible that a big farming enterprise or cooperative would develop proprietary in-house fashions.
The farmer’s AI wants to collect data from these specialised fashions by sending context to them: what the farmer desires to know, in fact, but in addition the situation of the fields, climate patterns over the previous 12 months, the farm’s manufacturing over the previous few years, the farm’s technological capabilities, the supply of assets like water, and extra. Moreover, it’s not only a matter of asking every of those fashions a query, getting the solutions, and producing a end result; a dialog must occur between the specialist AIs as a result of every reply will affect the others. It might be doable to foretell the climate with out realizing about economics, however you’ll be able to’t do agricultural economics for those who don’t perceive the climate. That is the place MCP’s worth actually lies. Constructing an utility that asks fashions questions? That’s positively helpful, however any highschool scholar can construct an app that sends a immediate to ChatGPT and screen-scrapes the outcomes. Anthropic’s laptop use API goes a step additional by automating the press and screen-scraping. The actual worth is in connecting fashions to one another to allow them to have conversations—so {that a} mannequin that predicts the worth of corn can uncover climate forecasts for the approaching 12 months. We will construct networks of AI fashions and brokers. That’s what MCP helps. We couldn’t think about this utility only a few years in the past. Now we will’t simply think about it; we will begin constructing it. As Blaise Agüera y Arcas argues, intelligence is collective and social. MCP offers us the instruments to construct synthetic social intelligence.
The business has been speaking about brokers for a while now—dozens of years, actually. The latest burst of agentic dialogue began simply over a 12 months in the past. For the previous 12 months we’ve had fashions that had been ok, however we had been lacking an necessary piece of the puzzle: the power to ship context from one mannequin to a different. MCP gives among the lacking items. Google’s new A2A protocol gives extra of them. That’s what context serialization is all about, and that’s what it permits: networks of collaborating AIs, every performing as a specialist. Now, the one query is: What’s going to we construct?
