@benln Building an entire new paradigm for Agents on the internet, starting with commerce.
AFCP AI is building the Foundational infrastructure of Agentic commerce.
Great article by @joshelman on what compounds for consumers AI.
Agree with points 1 and 2 and great questions that help me in building as well. The way I see things, Point 1 is about providing Agents with magic abilities and point 2 is mainly about Agents becoming a part of people's daily lives as it improves, just like a friendship maturing like wine. I am building for such perspectives.
Question 3, in my opinion, is a tricky one to answer.
Here is why.
From my experience of building an Agentc infrastructure and learning Agents. I do not believe in a model-centric approach to Agents. So, even if the model or harness improves, there is little value I see in it's improvement, even if they are used in Agents.
In an Agentic infrastructure, the ecosystem matters, the end value in terms of usability to customers matter more. For example, an agent delivering personalized value is actively learning customers preferences. To solve for personalized value, There are rather many factors at play, rather than static compute of a model. It needs multiple factors like orchestration, memory, personalization, ecosystem and customer experiences to be solved for.
I do believe the world of Agents is a massive opportunity and in the future we will move away from a model-centric agent world to becoming Agent-centric. I call this the Agent-pilled or Agent-first world.
In this world, there are diverse and exciting ways in which an Agentic infrastructure provides value. The infrastructures which solve for an Agent-first world will be the winners.
I am solving for such a world while building our Foundational infrastructure for Agentic commerce.
Agree, For Agents, It's all about usability coupled with network, habit and magic like abilities. I am working on this in our foundational infrastructure building.
Are you seeing consumers founders ignoring model improvements? ...meaning not being model-first, yet using models at a smaller scale...
The general advice is to fight of fear, or kill it.
I have an antithesis to it.
I believe, One must find their fear again.
Fear is a friend. It is a good thing.
As Andy Grove rightly said Only the paranoid/fearful survive.
Fear can shape innovation, impact and creativity.
Eg:
While building our Agentic Infrastructure for commerce.
Some questions keeping me up are:
In what way the Shopper Agent we offer our clients cannot be relatable to their customers?
In what way can it go rogue, and become unreliable and unsafe?
In what way it cannot be in touch with reality?
These fearful questions are what keep me going, as I want the entire world moving onto building and using Agents.
Building apps, desktop, tablet or website solutions will then be made history.
Convert your website or app store to an Agent with our Platform.
Build a shopper Agent for your online store.
Your online store is no longer just a destination; it can now be an intelligent agent that actively sells for you.
Your customers get a digital friend for life.
Join us for our Beta trial.
https://t.co/JJP0G1IZTL
@aleshu_ber correct, Self learning with statistical paradigms is a myth. Models have bounds till which they can improve. These bounds in itself are en-tropical in essence.
@Dr_Gingerballs correct, Self learning with statistical paradigms is a myth. Models have bounds till which they can improve. These bounds in itself are en-tropical in essence.
Insightful read from @0xfishylosopher!
Thanks for sharing @mignano.
An Antifragile internet for the win :) And agents will end up being the ones to get us here despite the security breaches and loopholes they have been exploiting so far.
We overstate the negative effects of agents as they are now and keep understating their benefits.
For an antifragile internet to exist, no tech company alone must have the power to dictate AI and AI Agent safety.
Rather it must be a collaborative effort i.e. an Open Standard just like the pre-agentic internet.
Imagine a world where an enterprise, a business, or a customer of a business have that power and control over their agents, so they get safe, reliable and secure agents?
That is the problem I am solving whilst building the foundational infrastructure for agentic commerce.
It makes sense it is a consumer AI Agent, however I think it is not a true Agent in its essence. Agents in it's true essence are personalized, have an identity and can reliably get more efficient with time. Moreover, an Agent for a website or store needs to deliver value to their customers.
I believe the true impact of such Agents are that customers can have such Agents for their lifetime. I am building a platform to make this a reality for commerce.
@skye_venture@joshelman affordability is key. However this is directly correlated to how efficient platforms help Agents become. Not many are focused on such innovations, I am solving for efficient AI.
It's an exciting shift. The things Agents can enable us do is massive. I am working on such an impact of Agents for commerce. Agents for everyday life where people have more time to do things they love and also that such Agents can be accessible on any device, so people do not have build so much or at all.
@liadyosef Great points Liad, agree that browsers are not the only entry point. I believe Agents are the entry way to everything. Furthermore, Agents will work everywhere. I am building Agentic commerce for this shift.
There are 2 paradigms for creating agents at the moment.
The model-pilled approach and the Agent pilled approach.
The businesses building Agents on top of models are the ones who are going to struggle.
The one's building Agents with an Agent first approach are the ones who will solve for the future value.
We at AFCP AI are building for the later.