4 ms·

> If you haven't spent at least $1,000 on tokens today per human engineer So a four person team should be spending close to $1M/year, double each engineer’s sa

> If you haven't spent at least $1,000 on tokens today per human engineer So a four person team should be spending close to $1M/year, double each engineer’s salary, on AI alone? To get the output of one junior engineer who smokes crack and has his memory wiped every twenty minutes?

direwolf20 9mo ago
Doubling? Try quadrupling outside of silicon valley. He is saying hire 4x as many engineers and make 3/4 of them AI. So much for the 10x productivity increase — that's 0.25x!
simonw 9mo ago
If that team is producing 4x what they would be producing without LLMs then spending 2x their salaries on tooling sees financially rational to me. (I know, that's a very big "if".)
- coder23853 9mo ago
  There are many ways "producing" can be quantified (LOC, PRs, features) such that 4x production does not correlate to a 4x in value of the product (quality, revenue).
- ricardobeat 9mo ago
  There is no mention of a productivity increase anywhere in this piece. 2x? Maybe, but at this price you can hire another very senior engineer, and have both be 50% more productive with a $200/month AI spend.
  - simonw 9mo ago
    Measuring productivity in software development is notoriously hard, but the demos I saw them give when the team of three people had been working together for three months showed more work then I would ever expect from a team of that size over that amount of time. They already had their own custom agent harness and digital twin imitations of Jira and Slack and Okta and a bunch of other custom tooling. It looked to me like way more than a 2x or 4x thing.
    - ricardobeat 9mo ago
      Opus can one-shot a simple Jira clone in under five minutes. I haven’t seen any of their work, but this “digital twin” story sounds like a marketing play. I agree that productivity is very subjective too, in my experience a team of three developers with a solid plan could build a lot in three months even before AI.
      - simonw 9mo ago
        The thing they're building here is specifically a clone of Jira that duplicates the official Jura API well enough that they can use it for integration testing without being held to Jira rate limits. They do that by running the official client libraries against their implementation to make sure the functionality is all covered.

direwolf20 9mo ago

Doubling? Try quadrupling outside of silicon valley. He is saying hire 4x as many engineers and make 3/4 of them AI. So much for the 10x productivity increase — that's 0.25x!

simonw 9mo ago

If that team is producing 4x what they would be producing without LLMs then spending 2x their salaries on tooling sees financially rational to me. (I know, that's a very big "if".)

coder23853 9mo ago
There are many ways "producing" can be quantified (LOC, PRs, features) such that 4x production does not correlate to a 4x in value of the product (quality, revenue).
ricardobeat 9mo ago
There is no mention of a productivity increase anywhere in this piece. 2x? Maybe, but at this price you can hire another very senior engineer, and have both be 50% more productive with a $200/month AI spend.
- simonw 9mo ago
  Measuring productivity in software development is notoriously hard, but the demos I saw them give when the team of three people had been working together for three months showed more work then I would ever expect from a team of that size over that amount of time. They already had their own custom agent harness and digital twin imitations of Jira and Slack and Okta and a bunch of other custom tooling. It looked to me like way more than a 2x or 4x thing.
  - ricardobeat 9mo ago
    Opus can one-shot a simple Jira clone in under five minutes. I haven’t seen any of their work, but this “digital twin” story sounds like a marketing play. I agree that productivity is very subjective too, in my experience a team of three developers with a solid plan could build a lot in three months even before AI.
    - simonw 9mo ago
      The thing they're building here is specifically a clone of Jira that duplicates the official Jura API well enough that they can use it for integration testing without being held to Jira rate limits. They do that by running the official client libraries against their implementation to make sure the functionality is all covered.