Boston just built a front door for AI agents to talk to government, and it might be the first real answer to the chaos of machines pretending to be people on public websites.
The Signal
Machine traffic is already hitting government services hard, and nobody's built the infrastructure to handle it properly. AI agents are scraping pages, guessing at forms, and treating government portals like they're consumer websites. Boston CIO Santi Garces isn't waiting for the chaos to get worse. The city is implementing the Model Context Protocol (MCP), a spec from Anthropic that lets AI agents query government resources through a structured, governed interface instead of pretending to be browsers.
Here's why this matters: Boston is starting with open data, which is smart. Low stakes, high value. Instead of agents scraping the city's data portal and hoping they parsed the CSV correctly, they can query directly through MCP. The city pushes computation into the portal itself, which means better reliability and actual control over what agents can access. This isn't about letting ChatGPT fill out your parking ticket appeal. It's about building the pipes before every AI assistant in the world tries to simultaneously check if street cleaning is happening tomorrow.
The bigger picture: government interfaces were built for humans with browsers. That assumption is breaking. Some of this machine traffic is useful. Some is ambiguous scraping. Some could be actively harmful if agents start reserving scarce public services or flooding systems with automated requests. Boston is building the middle path between blocking all agents or letting them run wild.
The Implication
Watch this model. If MCP or something like it becomes replicable infrastructure that other cities can deploy, we're looking at the early architecture of agent-to-government interaction. The cities that build governed layers now will control how AI accesses public resources. The ones that don't will spend the next five years playing whack-a-mole with bot traffic.
Source: Fast Company Tech