Casey

Engineer · AI usage, harness practices and where spend could have gone further.

Demonstration report. This engineer and their usage are synthetic.

This page is for the engineer and their manager to find cheaper defaults and harness fixes. It is never used to rank people.

Period

Summary

FY2027 to date

AI spend

$1,758

146 sessions · $12 each

Savings opportunity

$395

22% of spend, at list price

Practices on target

3 of 7

Focus: rtk coverage, Knowledge MCP use

Cache hit rate

81%

Team 82% · org 82%

FY2027 to date · estimated at list price

$395 of $1,758 spend

Largest saving first

Improve prompt caching$227
Cache hit rate 81% against an 85% target.
Keep stable context at the front of prompts and avoid switching models mid-session, so 85% of input reads from cache.
Right-size models$94
42% of routine spend ran on Opus: code research $90 of $225, documentation $90 of $197, ops $38 of $87, presentation $11 of $38.
Run docs, decks, ops and read-only research on Sonnet instead of Opus.
Keep context lean$56
9% of input tokens were past 100K of context.
Compact or /clear between tasks instead of carrying more than 100K tokens of context.
Route commands through rtk$18
61% of 1,364 Bash calls ran through rtk; 2.3M tokens of command output went into context.
rtk compacts test, build, git and search output before it reaches the model (about 70% smaller).

Compared with Content Pipeline and the organization

Practice	Casey	Team	Org	Target	Recommendation
rtk coverage	61%Needs attention	62%	56%	≥ 80% of Bash calls	Install the rtk hook so test, build, git and search commands return compact output.
Knowledge MCP use 4 findings saved back	30%Needs attention	30%	30%	≥ 40% of sessions	Search the knowledge MCP before re-researching a codebase, and save findings back with remember.
Cache hit rate	81%Needs attention	82%	82%	≥ 85% of input	Keep CLAUDE.md and tool lists stable, avoid mid-session model switches, and resume sessions instead of restarting.
Context past 100K	9%On target	9%	10%	≤ 10% of input	On target
Opus on routine work	42%Needs attention	57%	44%	≤ 25% of routine spend	Default routine sessions to Sonnet and keep Opus for harness, design and hard debugging work.
Sessions using subagents	55%On target	55%	50%	≥ 30% of sessions	On target
Tool error rate	2.1%On target	2.4%	3.0%	≤ 4% of tool calls	On target

Is the engineer moving toward the targets?

Month	rtk	Knowledge	Cache	Context	Model fit	Subagents	Savings opportunity
Feb 2026	–	–	61%Off target	10%Needs attention	–	13%Off target	$13
Mar 2026	–	–	65%Off target	7%On target	62%Off target	25%Needs attention	$43
Apr 2026	–	–	64%Off target	11%Needs attention	62%Off target	42%On target	$33
May 2026	24%Off target	–	72%Needs attention	11%Needs attention	55%Off target	53%On target	$50
Jun 2026	42%Needs attention	8%Off target	78%Needs attention	10%On target	53%Off target	46%On target	$91
Jul 2026	67%Needs attention	27%Needs attention	74%Needs attention	7%On target	40%Needs attention	50%On target	$68
Aug 2026	66%Needs attention	35%Needs attention	78%Needs attention	10%Needs attention	38%Needs attention	44%On target	$49
Sep 2026	66%Needs attention	35%Needs attention	87%On target	8%On target	30%Needs attention	67%On target	$33
Oct 2026	75%Needs attention	37%Needs attention	85%Needs attention	10%On target	27%Needs attention	66%On target	$15

FY2027 to date

By model tier, by month

Routine types (documentation, presentation, ops, code research) rarely need Opus

Cache reads are billed at a fraction of uncached input; a high share is what good caching looks like

Model	Requests	Uncached input	Cache reads	Cache writes	Output	Spend
Opus 5.5	2,359	12.1M	125.2M	8.8M	3.3M	$182
Opus 5	2,733	17.4M	137.3M	9.9M	3.9M	$315
Opus 4.8	2,686	25.1M	111.3M	8.7M	3.6M	$325
Opus 4.6	1,744	27.6M	75.4M	6.6M	2.6M	$283
Opus 4.5	59	895K	1.7M	164K	85K	$8
Sonnet 5.5	2,328	12M	120.2M	8.4M	3.8M	$95
Sonnet 5	4,657	29.4M	218.7M	15.8M	6.3M	$308
Sonnet 4.6	1,987	22.9M	85.1M	6.9M	2.8M	$162
Sonnet 4.5	32	549K	1M	100K	40K	$3
Haiku 5.5	2,687	15M	140.9M	10M	4.9M	$7
Haiku 4.5	2,636	19.7M	123.7M	9.2M	3.6M	$61
GPT Realtime (meetings)	80	730K	3.7M	286K	126K	$8
Total	23,988	183.4M	1144.3M	84.7M	35M	$1,758

Spend and tokens come from LiteLLM gateway logs and imported agent session history, priced at API list rates.
Savings are estimates: routine Opus work repriced at the Sonnet available at the time; uncached input moved to cache reads up to an 85% hit rate; rtk removing 70% of unfiltered command output, which stays in context for about half the session's remaining turns; and 50% of input past 100K context avoided by compacting. Levers overlap, so the total is an upper bound.
rtk and the knowledge MCP are scored from their rollout (rtk May 2026, knowledge MCP Jun 2026).
Generated Oct 20, 2026, 7:46 AM.