According to @tryramp's AI index, U.S. business AI adoption crossed 50% for the first time this year.
Sector adoption rates have been climbing since 2023. Gradual, broad, predictable. Spend per employee for tech companies has 2x'ed year-to-date to $66. Finance hit $36.
The adoption curve is a slow S-curve. The spend curve is a hockey stick. Companies aren't just trying AI anymore. They're deepening into it fast, and the cost structure is following. This is now an operating budget with a line item. The answer isn't picking one model and hoping it works for everything.
Different tasks need different models, and the companies that figure out smart routing early will spend a lot less getting a lot more done.
Peak of Inflated Expectations: 'I shipped an app in a weekend'
Trough of Disillusionment: 'Nobody opened it twice'
Slope of Enlightenment: realizing the hard part was never the code
@levie the layer also needs a good harness. @coworkerapp we've also found that it's not just routing to the right model but the instructions and task structure have to adapt to the model too, otherwise you're just switching engines without changing the fuel.
My 3 biggest takeaways from @lennysan's latest podcast with @benedictevans:
1/ We're in 1997 for AI
The internet was real and exciting, but Google, Amazon or Facebook didn't even exist yet. Most of AI's potential is still unrealized or not imagined.
2/ Job replacement fears are exaggerated
History shows automation creates new roles, not just eliminating old ones. There will be friction, but we are good at adapting.
3/ Foundation models face a commodity squeeze
AWS/Azure became plumbing. Value shifted to SaaS apps built on top. We're already seeing it as Microsoft and Uber are pulling back on AI spend. Pricing pressure is shifting to the customer's favor.
The implication: throwing the most expensive model at every task isn't a strategy. @coworkerapp automatically routes to the right model for every task. Frontier performance, without the frontier bill.
@MadisonMills22 We’ve spoken to a lot of companies that are thinking about costs @coworkerapp. A routing layer (right model for the task) coupled with a company context layer allows them to unlock much more for 80% less cost.
@MadisonMills22 as companies explore more "healthy swing" - my take is that there's going to be less of a tradeoff between jobs and AI spend. We've implemented multi-modal routing (including open-source models) at @coworkerapp and found costs to be 80% less. also 😂: https://t.co/P74ntMcRko
Token costs are exploding
Companies have to choose: cut AI spend or cut people?
Coworker is the 3rd option
> Same frontier AI
> Chat, Cowork, Code
> 5X more tokens per dollar
Live today.