This Company Spends 75k on AI tools
Summary
One company allocates $75k annually per engineer on AI tooling—a 50/50 split between Claude and Codex with emerging preferences for specialized tools like Cursor for frontend work. This reveals how forward-thinking tech teams are operationalizing AI across development workflows rather than treating it as experimental.
Key Takeaways
- Allocate $75k per engineer annually for AI tooling across multiple specialized tools—not a single solution. This represents a significant but justified investment in developer productivity.
- Portfolio approach wins: combine generalist tools (Claude/Codex for logic) with specialist tools (Cursor/Composer for frontend visual work). Matching tool to task type improves outcomes.
- Model selection is actively shifting—Claude dominated 5 months prior, but new Codex iterations are now competitive. Continuously evaluate emerging models rather than locking into single vendors.
- Devon excels for bug triage and minor tweaks at scale. Use AI tools for high-volume, lower-complexity tasks (bug fixes, visual adjustments) to maximize ROI on tooling spend.
- Team experience integration matters: evaluate tools based on Slack/communication layer quality, not just core functionality. Better team workflows compound productivity gains.
Related topics
Transcript Excerpt
What percent of developer salary do you spend on tooling? >> I mean the run rate's at least 75k per engineer. >> Wow. Split between Devon, Cloud, Codeex, and then Town. We use Devon a lot for a lot of bugs that come in for a lot of like simpler little things or little like visual tweaks. We are kicking off Devon cuz we just find the team experience and Slack's really really good. We have a few people who use cursor for like more visual the front end. The model's really fast, right? Composer is really fast for front end. And then I would say it's probably 50/50 right now between Codex and Claude. And that's obviously changed a lot. Like I think five months ago I would have said it was mostly Claude. But the new Codex Codex is really good.…
More from 20VC
- Instinct Worth 30BN? | Microsoft Cuts 27,000 Jobs While Meta Taps Equity Markets for $95B CapEx
- Satya Nadella on Open vs Closed Models: Who Wins | Why Companies Need to Own Their Own Intelligence
- Inference Will Be $10TRN Market: What No One Sees | Software Investors are F***ed | Naveen Rao
- The New VC Game | CoreWeave Sell-Off Creates Panic | Revenue Multiples Out of Control
- The Future of Inference Market & Model Routing: Commodity or Increasing in Value? | LangChain CEO