Enterprise AI Spend Benchmark
A possible benchmark for how organizations allocate AI spending across inference, software, implementation, training, and human review.
The more defensible data product is not a synthetic benchmark. It is a benchmark of actual organizational behavior.
Participating companies could contribute anonymized observations about AI spend, workflows, review burden, throughput, error rates, and realized savings.
Possible cuts
The benchmark could eventually compare:
- industry;
- company size;
- function;
- workflow type;
- spend per employee;
- inference versus software versus implementation cost;
- human review time;
- measured output gain.
The publication can create the network around the dataset before the dataset is large enough to become a product.