Your AI is wasting compute (and you're paying for it)

Add to Claude Other AI apps

Amazap cuts AI token usage.

Saved by Amazap users

374,765estimated tokens saved

since Oct 17, 2026

$0.53estimated AI spend saved

Estimated at $2 per 1M tokens.

How we measure ›

Includes our own testing.

Here's how:

Raw build log79,543 tokens

09:14:02 ##[group]Run actions/checkout@v4
09:14:02 with:
09:14:02   repository: acme/payments-api
09:14:02   fetch-depth: 1
09:14:02 Syncing repository: acme/payments-api
09:14:03 ##[endgroup]
09:14:03 ##[group]Run actions/setup-python@v5
09:14:04 Successfully set up CPython (3.12.7)
09:14:04 ##[endgroup]
09:14:05 ##[group]Run pip install -r requirements-dev.txt
09:14:06 Collecting fastapi==0.115.2
09:14:06   Downloading fastapi-0.115.2-py3-none-any.whl (94 kB)
09:14:06 Collecting pydantic==2.9.2
09:14:06   Downloading pydantic-2.9.2-py3-none-any.whl (434 kB)
09:14:06 Collecting sqlalchemy==2.0.36
09:14:07   Downloading SQLAlchemy-2.0.36-cp312-cp312-manylinux_x86_64.whl (3.2 MB)
09:14:07 Collecting pytest==8.3.3
09:14:07 Collecting httpx==0.27.2
09:14:07 Collecting anyio>=3.7.1
09:14:08 Installing collected packages: sniffio, idna, h11, anyio, ...
09:14:19 Successfully installed 48 packages
09:14:19 ##[endgroup]
09:14:19 ##[group]Run pytest -q --maxfail=20
09:14:21 ============ test session starts ============
09:14:21 platform linux -- Python 3.12.7, pytest-8.3.3
09:14:21 collected 1184 items
09:14:22 tests/test_auth.py ........................ [  2%]
09:14:24 tests/test_cart.py ........................ [  4%]
09:14:27 tests/test_catalog.py ..................... [  6%]
09:14:29 tests/test_checkout.py .................... [  8%]
09:14:31 tests/test_credits.py ..................... [ 10%]
09:14:33 tests/test_invoices.py .................... [ 12%]
09:14:35 tests/test_ledger.py ...................... [ 14%]
09:14:37 tests/test_limits.py ...................... [ 16%]
09:14:39 tests/test_orders.py ...................... [ 18%]
09:14:41 tests/test_quotes.py ...................... [ 20%]
09:14:43 tests/test_rates.py ....................... [ 22%]

Log Squeezer1,855 tokens

FAILED 1 of 1184 tests (job: test)

tests/test_credits.py::test_balance_applied
  tests/test_credits.py:88
  AssertionError: expected
  credit 10, got 0

Cause: credit hook skipped when
upstream returns 502
(app/credits.py:41)

Example log. Token counts are medians from our Log Squeezer benchmark of 9 CI logs. Results vary by log.