Private Equity FunCast · Oct 21, 2026 · 59m · 51m saved
With Dan Bender, Founding Partner at Code & Co. · Kirby Montgomery, Head of North America at Code & Co.
Less than 30% of the AI tokens consumed today go to frontier models, Dan Bender, founding partner at Code & Co., told Private Equity FunCast, alongside Kirby Montgomery, the firm's head of North America.
Private equity firms track sales pipelines and finance to the decimal. Engineering has stayed the department nobody can measure, and Dan and Kirby argue that every proxy tried so far, from lines of code to token spend, measures the wrong thing.
"Less than 30%, 28% of tokens consumed are being consumed by frontier models now. The rest is open weight models." — Dan Bender
Code & Co. runs technology due diligence for private equity buyers and audits the engineering teams in ParkerGale's portfolio, so Dan and Kirby see what a team claims about AI and what it can prove.
The full episode is covered here so you can skip it. 59 minutes of audio, 8 minutes of reading.
Here are the 9 lessons that matter.
Key Takeaways
Lines of code, story points and token counts are all bad proxies for engineering progress
The best work is often deleting code, not adding it
Tech diligence has moved earlier, and it now opens with whether a company's AI claims are real
Only 28% of tokens go to frontier models, a sign teams are routing work to cheaper models
DORA release metrics fit some companies, not all, especially legacy and regulated software
Saying "we're training a model" is a red flag when the team has not tried fine-tuning, RAG or MCP first
Once a number decides pay, it stops telling the truth, so measure to learn, not to punish
1. The Unmeasured Department
Jim Milbery, Technical Partner at ParkerGale, who hosts the show, opened with the gap his firm sees across its portfolio.
Private equity can measure sales and finance precisely, but not engineering
"We can see sales pipeline, we can see finance down to the decimal... So, why can't we measure engineering?" — Jim Milbery
Dan said engineering has always been hard to scope, and the gap shows up when a deal's plan meets the code.
Dan Bender sees a gap between what investors plan and what engineering can deliver
"There's often a disconnect between the commercial strategy, the value creation hopes and plans, and the engineering realities." — Dan Bender
