Working Notes on Agent Systems/Brad Zhang

@teach_fireworks / X longform

AI Agent brings unprecedented opportunities to small teams. I remember last year I reposted an interview with

AI Agent brings unprecedented opportunities to small teams. I remember last year I reposted an interview with Nadella. He accurately predicted that Agent would explode this year...

July 28, 2025 · 2 min read

AI Agent brings unprecedented opportunities to small teams. I remember last year I reposted an interview with Nadella. He accurately predicted that Agent would explode this year, and previous business models and implementations would undergo huge changes. I posted a paragraph of his judgment last year that all business logic will be transferred to these intelligent agents, and these agents will perform CRUD operations on multiple databases. So they won't differentiate what the backend is, they will update multiple databases, and all the logic will be in the AI ​​layer. When he talks about business logic, he basically means leveraging the core data in the database to actually do things, whether it's updating a CRM record, or sending emails based on different criteria in the database, or any piece of software that sits on top of the database. All this business logic will be compressed into the Agent, and the Agent will be able to do anything, you just tell it what you want. Manus appeared not long after, and I think looking back in the future, this has become a landmark event for AI Agent. In particular, Manus is neither a traditional major manufacturer nor a product of the Silicon Valley model, but a work from a Chinese team. Let’s look back at the Manus R&D cycle revealed by the Butterfly Effect, and combined with the fact that they have not launched an AI browser, I boldly speculate that they should have predicted and actually implemented the amazing universal Agent effect internally. The inspirations I got are as follows:

  • With excellent underlying basic model capabilities, all living beings are equal. It is no longer the past model of stacking money and people looking for VCs.
  • Anyone who can deeply understand the upper and lower limits of the basic model, has a lot of excellent engineering practice experience, can understand the real demands of users in the AI ​​era (even when the users themselves cannot explain clearly), and coupled with correct predictions of technology trends, can quickly succeed. This may seem difficult, but it is also the breakthrough point for small teams, and it has been proven time and time again: traditional giants are still very strong, but huge business empires and models are also a limitation, a double-edged sword, it depends on how flexible the elephant can turn around, so it is difficult to pull back. It is difficult for the elephant to turn around flexibly, but they will do everything possible to maintain the glory and share of the past.

Visual summary

Article argument map

Generated from the post's content graph

FORMATTOPICCAPABILITYMARKETcoverscoverssignalssignalssignalssignalssignalssignalsFORMATlongform noteTOPICagentsTOPICretrievalCAPABILITYagent workflowCAPABILITYdeveloper toolingCAPABILITYtechnical writingCAPABILITYproduct surfaceCAPABILITYtechnical distributionMARKETAI startupMARKETopen-source builders
Mermaid outline
flowchart LR
  format-long_post["longform note"]
  topic-agents["agents"]
  topic-retrieval["retrieval"]
  capability-agent-workflow["agent workflow"]
  capability-developer-tooling["developer tooling"]
  capability-technical-writing["technical writing"]
  capability-product-surface["product surface"]
  capability-technical-distribution["technical distribution"]
  market-ai-startup["AI startup"]
  market-open-source-builders["open-source builders"]
  format-long_post -->|covers| topic-agents
  format-long_post -->|covers| topic-retrieval
  format-long_post -->|signals| capability-agent-workflow
  format-long_post -->|signals| capability-developer-tooling
  format-long_post -->|signals| capability-technical-writing
  format-long_post -->|signals| capability-product-surface
  format-long_post -->|signals| capability-technical-distribution
  format-long_post -->|signals| market-ai-startup

Visual structure

Essay structure map

Built from summary and key paragraph positions

AI Agent brings unprecedented opportunities to small teams. I remember last year I re...THESISAI Agent bringsunprecedentedopportunities to smallteams. I remember lastSIGNALAI Agent bringsunprecedentedopportunities to smallteams. I remember lastOPERATOR1. With excellentunderlying basic modelcapabilities, allliving beings areIMPLICATION2. Anyone who candeeply understand theupper and lower limitsof the basic model,
Mermaid outline
flowchart LR
  thesis["AI Agent brings unprecedented opportunities to small teams. I remember last year I reposted an interview wi..."]
  signal["AI Agent brings unprecedented opportunities to small teams. I remember last year I reposted an interview wi..."]
  operator["1. With excellent underlying basic model capabilities, all living beings are equal. It is no longer the pas..."]
  implication["2. Anyone who can deeply understand the upper and lower limits of the basic model, has a lot of excellent e..."]
  thesis -->|frames| signal
  signal -->|develops| operator
  operator -->|lands in| implication

Source: View the original post