Skip to main content

Your AI bill doubled.Can you saywhy? We can.

Here is how.

Every Claude Code, Codex, and Cursor session, split by project, person, task type, and model. Quesma then lists the changes worth making, and answers any question about your agent spend.

Talk to the founder See how it works
Product demo: the Quesma usage overview, cost breakdown by project, and suggested changes to agent usage.

When the bill doubles,the reasonslook like this:

Illustrative AI coding usage bill showing token usage and amount due

  • Marketing started using Claude Code

    A team you didn’t expect found a use for the tools.

  • Heavy work with Fable 5

    The default model got smarter, and pricier.

  • Security fixes before the audit

    The work ran hot for three weeks because you asked it to.

Fix the problems. Find good practices.

Cost tools only find things to cut. Quesma also finds problems recurring across teams, and the skills, tools, and routines used in one part of the company but nowhere else.

A failure worth fixing once

Three teams worked around the same broken integration environment.

Sessions by repository■ = one session

A skill worth sharing

deploy-preview worked for one team. Others had never seen it.

Weekly use of deploy-preview12 weeks → now
Platform12/12
Payments7/12
Web2/12
Datanot yet

Instructions agents misread

One repository rule sent every session down the same wrong turn.

web-app/CLAUDE.md · line 14
“Always run the full e2e suite before committing”
  • #4821 fix typo in tooltip → full e2e suite38 min
  • #4837 rename a variable → full e2e suite41 min
  • #4859 update README → full e2e suite44 min

31 sessions took the same wrong turn

Dependencies nobody decided on

context7-mcp spread until more than half the company relied on it.

Teams relying on context7-mcpFeb → Jul

Our track record

BinaryAudit benchmark artwork

BinaryAudit

Can coding agents find backdoors in compiled binaries using Ghidra and radare2?

See all benchmarks

Frontier-model training

We built training environments for difficult, multi-hour agent tasks. The work required reading full trajectories: prompts, tool calls, failures, retries, and results.

Independent audit

We audited Blitzy's 66.5% SWE-Bench Pro result, reviewing the environment, required changes, and final score.

Read the independent audit

Terminal-Bench 3.0

Quesma contributed to Terminal-Bench 3.0, a benchmark for testing agents on practical terminal tasks.

View the contributors

OpenAI-cited research

OpenAI cited our Baba Is Bench work while examining what makes agent evaluations succeed.

Read the citation

Latest from the blog

View all posts

Bring clarity to your AI coding budget

Early access opens soon. The open-source collector runs in your environment and the session data stays in your cloud. Contact us to be one of the first teams.

See how it works Talk to the founder