To get into the real code base, AI software engineers must connect the boundaries of work orders, testing, and model routing, Cognition, and the Common Model Lab.

Scott's statement is clear: They will be more concerned about "what it looks like to build software end-to-end in real environments like Goldman Sachs and Mercedes-Benz." The difficulties here come from the existing code base, real human collaboration, ticketing system, local testing, permissions, processes and a bunch of messy details.

This is also the most difficult part of AI software engineer products: it needs to understand what the code base looks like today, connect to the system that the team is already using, so that the agent can run its own tests, verify the results locally, and then hand the output back to the engineering process.

He also mentioned that Cognition can do "Switzerland": help customers route to appropriate models, optimize price/performance, and recommend appropriate use cases.

This positioning is valuable because what enterprise customers really want is to complete tasks, and budgets should not be tied to a certain model.

Visual summary

Article argument map

Generated from the post's content graph

Mermaid outline

flowchart LR
  format-long_post["longform note"]
  topic-agents["agents"]
  topic-harness-engineering["harness engineering"]
  capability-agent-workflow["agent workflow"]
  capability-product-surface["product surface"]
  format-long_post -->|covers| topic-agents
  format-long_post -->|covers| topic-harness-engineering
  format-long_post -->|signals| capability-agent-workflow
  format-long_post -->|signals| capability-product-surface

Visual structure

Essay structure map

Built from summary and key paragraph positions

Mermaid outline

flowchart LR
  thesis["To get into the real code base, AI software engineers must connect the boundaries of work orders, testing,..."]
  signal["To get into the real code base, AI software engineers must connect the boundaries of work orders, testing,..."]
  operator["This is also the most difficult part of AI software engineer products: it needs to understand what the code..."]
  implication["This positioning is valuable because what enterprise customers really want is to complete tasks, and budget..."]
  thesis -->|frames| signal
  signal -->|develops| operator
  operator -->|lands in| implication

Source: View the original post