I don’t understand the point of WebMCP. Any website that you could do interesting things with programmatically either already has an API or doesn’t want to provide programmatic access for business reasons. So who is this for?
I don't think it's that binary. For the majority of websites, adding programatic access didn't make sense because it would primarily have been for scraping data. No financial incentive for them to support this.
The dream today is that agents might prefer your website due to it's MCP and actually purchase something. i.e. 'hey claude go find me the cheapest places I can buy these headphones'
Scrolling through the sponsors list (it includes, among others, Shopify), I'd guess that they hope to reduce the friction between consumer AI usage and shopping.
Many people already default to agentic AI with sub agents to compare products and stuff like that, from Google, or from their favorite chatbot.
And sites that sell things definitely have an incentive to open up programmatically to this kind of usage, and drive it further to the point of selling via chat bot or other AI tool (e.g. speech thingy): people who do that are now a small minority.
And also, I think the AI labs, especially Google, are all aware of the "Dead Internet" problem in terms of their own business.
It's not because of ethics, but the utility of Gemini or GPT Deep Research or Claude or whatever else quickly approaches zero for consumers when they lose all ability to rank credibility of the sites they crawl.
Cynicism wouldn't help them, because consumers do have the pesky habit to notice after a while when they're being scammed, e.g. not just people frequenting HN by now have noticed that you won't find an impartial and valuable product comparison on the internet. For example, when it comes to comparing features of a vacuum cleaner or office chair or whatever else.
And for the AI companies, programmatically communicating with commerce site could level this field a bit again, allowing more credibility assessment, in order to use sources other than lists of affiliate links or "trusted sources" (with lists of affiliate links) when "assisting" the consumer as an ad broker.
And on top of that, while controlling randomly-shaped bloated SPAs via a browser agent, LLM, screenshots, a11y tree etc in a multi-modal way is awesome for developing web apps, it is too expensive to connect AI to shopping websites at scale for consumers.
I mean, Google always was at the forefront of doing things like that, for example bringing hotel reservations or food delivery into Google search by pushing sites into publishing structured data.
Now that we have tools to automatically handle unstructured product presentations, there's money on distributing these product presentations at scale, while still trying to build consumer trust.
If you care about web standards, please for the love of god do not participate in this challenge. “WebMCP” is an extension to the `document` global under `document.modelContext`. The `document` object exists to represent the visual and structural node tree of an application. The `modelContext` extension is essentially an API gateway, and this proposal staples a global RPC registry with structured JSON payloads onto the DOM. It ignores the standard Request/Response API, and it does all of its work on the main thread. It forces developers into manual state-synchronization hell just to maintain a facade for a screen-scraping LLM, meeting them halfway so increasingly capable bots can “book flights on your website.” The API makes no considerations for API versioning, so any WebMCP tools you write are carved in stone.
I have no idea who at Google is advocating for this, but it’s a half-baked idea that goes against 20 years of web standard work. At least put this under the `window` or `navigator` global, goddamn.
I don’t understand the point of WebMCP. Any website that you could do interesting things with programmatically either already has an API or doesn’t want to provide programmatic access for business reasons. So who is this for?
I don't think it's that binary. For the majority of websites, adding programatic access didn't make sense because it would primarily have been for scraping data. No financial incentive for them to support this.
The dream today is that agents might prefer your website due to it's MCP and actually purchase something. i.e. 'hey claude go find me the cheapest places I can buy these headphones'
I assume it’s for one-off cases that you encounter while agents scrape/browse with a goal in mind.
Scrolling through the sponsors list (it includes, among others, Shopify), I'd guess that they hope to reduce the friction between consumer AI usage and shopping.
Many people already default to agentic AI with sub agents to compare products and stuff like that, from Google, or from their favorite chatbot.
And sites that sell things definitely have an incentive to open up programmatically to this kind of usage, and drive it further to the point of selling via chat bot or other AI tool (e.g. speech thingy): people who do that are now a small minority.
And also, I think the AI labs, especially Google, are all aware of the "Dead Internet" problem in terms of their own business.
It's not because of ethics, but the utility of Gemini or GPT Deep Research or Claude or whatever else quickly approaches zero for consumers when they lose all ability to rank credibility of the sites they crawl.
Cynicism wouldn't help them, because consumers do have the pesky habit to notice after a while when they're being scammed, e.g. not just people frequenting HN by now have noticed that you won't find an impartial and valuable product comparison on the internet. For example, when it comes to comparing features of a vacuum cleaner or office chair or whatever else.
And for the AI companies, programmatically communicating with commerce site could level this field a bit again, allowing more credibility assessment, in order to use sources other than lists of affiliate links or "trusted sources" (with lists of affiliate links) when "assisting" the consumer as an ad broker.
And on top of that, while controlling randomly-shaped bloated SPAs via a browser agent, LLM, screenshots, a11y tree etc in a multi-modal way is awesome for developing web apps, it is too expensive to connect AI to shopping websites at scale for consumers.
I mean, Google always was at the forefront of doing things like that, for example bringing hotel reservations or food delivery into Google search by pushing sites into publishing structured data.
Now that we have tools to automatically handle unstructured product presentations, there's money on distributing these product presentations at scale, while still trying to build consumer trust.
If you care about web standards, please for the love of god do not participate in this challenge. “WebMCP” is an extension to the `document` global under `document.modelContext`. The `document` object exists to represent the visual and structural node tree of an application. The `modelContext` extension is essentially an API gateway, and this proposal staples a global RPC registry with structured JSON payloads onto the DOM. It ignores the standard Request/Response API, and it does all of its work on the main thread. It forces developers into manual state-synchronization hell just to maintain a facade for a screen-scraping LLM, meeting them halfway so increasingly capable bots can “book flights on your website.” The API makes no considerations for API versioning, so any WebMCP tools you write are carved in stone.
I have no idea who at Google is advocating for this, but it’s a half-baked idea that goes against 20 years of web standard work. At least put this under the `window` or `navigator` global, goddamn.