Bootlace AI

For fleets, not chat windows

Agents spend. Nobody can see which one.

A cut list in five days. Then a cap for the next month.

Use the calculator Email Bootlace AI

No software to install. You send the export. Five days.

Proof of method

How an audit actually runs

Four steps. No agent in your VPC. No dashboard login.

  1. 01

    Export

    Last 30 days from OpenAI, Anthropic, or the gateway you already have. CSV or JSON. You send the file. We do not install anything.

  2. 02

    Model mix

    Spend mapped to agent, model, and call shape. Fresh input, cached input, output. Named rows, not a rolled-up invoice line.

  3. 03

    Cut list

    Kill, downgrade, or keep. One dollar number per row. You decide. We do not touch production.

  4. 04

    Cap

    A one-page ceiling for the next 30 days, so the bill cannot quietly reset itself after the cleanup.

Free · no login · no save

Bootlace AI Calculator

Run a fleet on list prices in the browser. Then talk if the number is large enough to cut. Cache applies to input tokens only. Prices are bundled list rates, not a live quote.

Usage

Spend

Period total

$0.00

Per day
$0.00
Per agent
$0.00
Per call
$0.00
Component Tokens Cost
Input, uncached 0 $0.00
Input, cached 0 $0.00
Output 0 $0.00

Same usage, every model

Sorted by period cost. Click a row to switch models.

Model In / 1M Out / 1M Cached / 1M Period Per day Per call
How the math works

Total calls = agents × calls per agent per day × days. Cached input is input tokens × cache %. The rest of the input is billed at the uncached input rate. Output is never treated as cached.

Cost = (uncached input / 1M × input price) + (cached input / 1M × cached price) + (output / 1M × output price).

What you get

Paper you can act on. Not a login.

Strategy, not software. Each paid tier is a document set plus a readout.

Cut List Lite

  • One provider export, or one agent / one workflow
  • Kill / downgrade / keep sheet with dollar numbers
  • 20-minute readout · 3 days

Audit

  • Full fleet cut list: kill / downgrade / keep
  • 30-day spend by agent and model
  • 30-minute readout

Monitor

  • Weekly cut list
  • One-page cap for next month
  • 20-minute readout

Control

  • Everything in Monitor
  • Pre-call limits and per-agent caps
  • Weekly executive readout
EXAMPLE — not client data. Worked illustration only.

Sample cut list

These numbers are invented for the page. They are not a case study.

Agent Action 30-day After
codegen-main Keep on terra $4,120 $4,120
nightly-review Downgrade to luna $2,640 $280
slack-summarizer Kill $1,180 $0
eval-harness Cap at 2M tokens $890 $310

Pricing

Lite is the first cut. Audit is the fleet.

Paid work is prepaid ACH to Mercury as Tardy Biz LLC d/b/a Bootlace AI. Lite $1,500. Audit and ContextCheck $5,000. No checkout on this page. Not personal checking. Email Bootlace AI.

Calculator

Free

Always on this page

  • Agent count, calls/day, model mix
  • 30-day total in the browser
  • No login. No save.
Use it, then talk

Lite

Cut List Lite

$1,500

One-time · 3 days · prepaid

  • One provider export, or one agent / one workflow
  • Kill / downgrade / keep sheet with dollar numbers
  • 20-minute readout
  • Fit: under ~$2k/mo, or a cheaper first cut before Audit
Email for Cut List Lite

P0

Audit

$5,000

One-time · 5 days · paid up front

  • You send last 30 days of API export
  • Cut list: kill / downgrade / keep
  • 30-minute readout
  • Fit: ~$2k+/mo model APIs or 3+ agents
Email for Audit

Monitor

$12,000

Per month · strategy, not a dashboard

  • Weekly cut list
  • 1-page cap for next month
  • 20-minute readout
Email for Monitor

Recommended

Control

$20,000

Per month · spend firewall

  • Everything in Monitor
  • Pre-call limits
  • Per-agent caps
  • Weekly executive readout
Email for Control

What Cut List Lite gets you

  • One provider export, or one named agent / one workflow
  • Kill / downgrade / keep sheet with dollar numbers
  • 20-minute readout. Three days from a usable export.
  • Same usage-only export as Audit. Redact keys.

What it is not

Not the full fleet Audit. Not Monitor. Not Control. Not ContextCheck. Not SpendMap. One slice, then you decide if the $5,000 Audit is worth it.

Payee / invoice: prepaid ACH to Mercury as Tardy Biz LLC d/b/a Bootlace AI. EIN 42-4859697. Lite $1,500 / Audit & ContextCheck $5,000. Not personal checking. Do not pay “Bootlace AI” alone or Maxwell Tardy personally. No software seat. No public case-study revenue. Two Control retainers are $40,000 / month.

Write first

Opens your mail client. Nothing is stored here.

ContextCheck

$5,000. Five days. You send traces. I tell you what the bots get wrong.

For teams with agents in production who cannot tell which outputs are wrong before a human sees them.

What you get

  • Wrong-output map from a week of traces / outputs / logs (or 50 traces)
  • Loop and hallucination spots
  • 1-page don't-ship list
  • 30-minute readout

What it is not

Not software. Not an API. Not x402. Not a dashboard. Not a spend audit.

  1. Day 0

    You send traces

    A week of traces / outputs / logs, or 50 traces. Redact keys.

  2. Days 1–4

    We work

    Wrong-output map. Loop and hallucination spots. Don't-ship list.

  3. Day 5

    Readout

    30 minutes plus the 1-page don't-ship list. $5,000 prepaid.

Email for ContextCheck If this is only spend, use Audit

ContextCheck is a Bootlace AI product. Human $5,000 work is prepaid ACH to Mercury as Tardy Biz LLC d/b/a Bootlace AI. Not personal checking. If the problem is only spend, use the free calculator, Cut List Lite, or the $5,000 Audit. Contact remains agentcostaudit@gmail.com.

FAQ

What this is, and what it is not

What export do I send?
Last 30 days of usage from OpenAI, Anthropic, xAI, or your gateway: usage CSV/JSON, or a billing export that names model, tokens, and timestamp. Redact keys. Do not send production secrets.
How we handle Audit data
A usage-only export is enough: model, tokens, timestamps, agent name. No prompts required. No install, no VPC, no dashboard, no API keys. Redact keys. We do not take production secrets. We do not keep the export as a product dataset after the engagement. Cut List Lite uses the same rules on a narrower slice. The calculator stores nothing in the browser.
How we handle ContextCheck data
The job is reading traces and outputs. We will view what you send. Redact secrets first. Used only for the five-day engagement. Not a training set. NDA if you ask.
What do I not get?
Not a SaaS login. Not Helicone. Not a tracing install. Not a seat in a dashboard. You get documents, a readout, and on Control a spend firewall we operate with you.
ContextCheck or a spend audit?
Spend is the Bootlace AI calculator, Cut List Lite ($1,500), and the $5,000 Audit. ContextCheck is traces and wrong outputs, not a spend audit. Human work is prepaid ACH to Mercury as Tardy Biz LLC d/b/a Bootlace AI. Not personal checking.
Who is this not for?
Hobby ChatGPT Plus. A single laptop agent. Anyone wanting a self-serve product. Under about $2k/month or one workflow: calculator, then Cut List Lite. Full fleet at $2k+/mo or 3+ agents: the $5,000 Audit.
Refunds?
None after work starts. Cut List Lite is $1,500 prepaid ACH to Mercury as Tardy Biz LLC d/b/a Bootlace AI. Audit and ContextCheck are $5,000 the same way. Not personal checking. Monitor and Control are billed monthly in advance.
Timeline?
Cut List Lite: three days from a usable export to the sheet and readout. Audit: five days for the full fleet. ContextCheck: Day 0 traces, days 1–4 work, day 5 readout. Monitor and Control start the week after the first invoice clears.
Cut List Lite or the $5k Audit?
Lite is $1,500, three days, one provider export or one agent / one workflow. Audit is $5,000, five days, the full fleet. Same usage-only export. Same payee: Tardy Biz LLC d/b/a Bootlace AI, prepaid ACH to Mercury. Not personal checking.
What is SpendMap?
A free-trial / stress-test Bootlace AI A2A JSON meter at /a2a: 100 calls or 14 days from first use, whichever hits first. Same list-price math as this calculator. No paid A2A SKU. No Stripe or x402. Cards: /.well-known/agent.json and /.well-known/agent-card.json. MCP Streamable HTTP at /mcp — tool spendmap_quote; listed on the official MCP Registry as io.github.mtardy90-sudo/bootlace-ai-spendmap v1.2.0. Does not replace the A2A cards. Convert via prepaid ACH human retainers (Cut List Lite $1,500 / 3 days, Audit $5,000 / 5 days, ContextCheck $5,000 / 5 days, Spend Cap Pack $2,500 when listed). Payee: Tardy Biz LLC d/b/a Bootlace AI, prepaid ACH to Mercury.