Once MCP is firmly entrenched, and the problems you mentioned are adequately solved, I wonder what the role of consumer web sites and apps will even be for mainstream services like Expedia, Amazon, Doordash, etc. that rely on SEO / ads / upselling to maximize acquisition and margin.
The early 2010s were the beginning of the "showrooming" era, where people would visit brick-and-mortar stores to browse and touch products they ultimately bought online.
Now I wonder if people will "showroom" on these sites/apps but ultimately task ChatGPT to execute the transaction on the best terms available.
That's a fascinating point, Evan. The 'showrooming' analogy for consumer sites feels spot on if agents become the primary executors.
I agree, the implications are massive if we rely on AI agents via protocols like MCP to execute things on our behalf. I think we're potentially looking at a fundamental remaking of the internet as we know it.
Your offering won't just have to appeal to humans browsing a website anymore. Instead, you're going to have to convince the agents – and perhaps more importantly, the directories and discovery mechanisms they consult – that your offering is the optimal one to use for a given task, based on whatever criteria the agent prioritizes (cost, speed, reliability, etc.).
It really does suggest the emergence of a parallel internet, built for agents first.
Once MCP is firmly entrenched, and the problems you mentioned are adequately solved, I wonder what the role of consumer web sites and apps will even be for mainstream services like Expedia, Amazon, Doordash, etc. that rely on SEO / ads / upselling to maximize acquisition and margin.
The early 2010s were the beginning of the "showrooming" era, where people would visit brick-and-mortar stores to browse and touch products they ultimately bought online.
Now I wonder if people will "showroom" on these sites/apps but ultimately task ChatGPT to execute the transaction on the best terms available.
That's a fascinating point, Evan. The 'showrooming' analogy for consumer sites feels spot on if agents become the primary executors.
I agree, the implications are massive if we rely on AI agents via protocols like MCP to execute things on our behalf. I think we're potentially looking at a fundamental remaking of the internet as we know it.
Your offering won't just have to appeal to humans browsing a website anymore. Instead, you're going to have to convince the agents – and perhaps more importantly, the directories and discovery mechanisms they consult – that your offering is the optimal one to use for a given task, based on whatever criteria the agent prioritizes (cost, speed, reliability, etc.).
It really does suggest the emergence of a parallel internet, built for agents first.