LAUNCH UPDATE

Planned launch:

Checkout stays paused until activation.
167:53:06Remaining time before launch
50% off your first month

Starter $9.50, then $19/month. Growth $24.50, then $49/month.

Reserve launch offerCreate a free trial account before launch. No card or payment required.
tracewell.
AI COST TRACKING FOR TEAMS & AGENCIES

Know your AI costs.
Protect your client margins.

Tracewell tracks reported AI costs, checks automation results, and helps you review what to charge each client. One workspace for the AI features and workflows your business runs.

Plans from $19/month · API keys included

↗ For agencies, software teams, and businesses running AI
Send data with API keys. Ask about it in chat.
  • AI providersOpenAI · Anthropic · Gemini · Your models
  • Automation platformsn8n · Make · Zapier · Custom code
  • Chat workspacesCodex · Compatible ChatGPT setups · MCP
Send usage and run metadata from your backend or HTTP steps. Read reported data in Codex or compatible ChatGPT setups. Setup required; no automatic account discovery.
t.NS
Studio workspace Fictional demo
THE BIG PICTURE

A little clarity.
A lot of control.

This month
Known AI spend$217.40Across 3 instrumented providers
Automations monitored031 needs attentionKnow when the work actually happens
Spend, in perspective Reported cost · USD
$15$7$010-1410-1710-2010-2310-26
Research digest is overdue

Expected every 24 hours · Research pipeline

Less checking. More knowing.Usage + outcomes, together.
01 / OBSERVE THE WHOLE PICTURETRACEWELL
One view. Whatever your stack.
OpenAIAnthropicn8n↗MakeYour own code

Send events from any provider or workflow. No automatic account discovery.

WHAT TRACEWELL DOES FOR YOUR BUSINESS

Your AI operations,
explained in one place.

Use Tracewell alongside your existing software. Start with the part of your business you need to understand.

Running AI in your app?

Report usage from OpenAI, Anthropic, Gemini, or your own models. Compare known costs by project and review budget alerts.

See the usage dashboard

Delivering automation for clients?

Send run reports from n8n, Make, Zapier, or custom code. Review failures, overdue runs, and whether the expected output arrived.

See workflow monitoring

Deciding what to charge?

Set a markup or review a client charge after delivery. Compare it with reported AI costs while keeping the original cost separate.

Explore the cost planner
HOW TO GET STARTED

One project. A scoped key. Your first report.

Create a project, keep its ingest key in your backend, and send usage or run metadata. Then review the dashboard or connect a read-only Codex or compatible ChatGPT workspace.

Read the connection quickstart →
YOUR TOOLS. ONE CLEAR VIEW.

Connect the tools you already use.

Choose how your data gets here, and where you review it.

Automation platforms

n8n · Make · Zapier · Custom code

Send run events through the API or an HTTP step to monitor failures and missed schedules.

Read the connection quickstart →

Chat workspaces

Codex · Compatible ChatGPT setups · MCP

Connect project-scoped read access to ask about tracked usage and automation outcomes.

Read the connection quickstart →

Setup is required. Tracewell sees only reported events; it does not automatically connect provider accounts or read private chats.

PRODUCT DEMONSTRATION

See the work. Understand the cost.

A short look at client projects, AI usage and automation outcomes.

Product demonstration · Sample dataEnglish narration

Read video transcript (English)

Your AI tools are running.

But what are they costing, and what are they delivering? Follow a run, inspect a failure, and see which costs are known, estimated, or still unknown. Set project budgets, review the record, and give the next decision a better foundation. Connect through documented APIs and project-scoped access with setup requirements made clear. From the first event to the next review, understand the work behind your automation.

Know what your AI delivers.

LESS NOISE. MORE SIGNAL.

See costs, results,
and client charges together.

Find the expensive project, the failed workflow, or the client charge that needs review—without piecing together separate spreadsheets.

01 / UNDERSTAND

Track AI costs by client.

Break down costs by client, project, provider, and model. Add your own markup or review a client charge after the outcome. Original costs stay separate; unknown prices stay unknown.

Explore usage
02 / VERIFY

Know whether workflows delivered.

A workflow ran. Did it do enough? Track failures, minimum record counts, and expected run windows from the events you send.

See automations
03 / ACT

Review missed runs and budget alerts.

Review in-app alerts for missed runs and budget thresholds. Get the context you need to investigate before small gaps grow.

See the signals
SMALL INTEGRATION. USEFUL PERSPECTIVE.

Connect your backend.
Keep your existing tools.

Send metadata from your code, webhook, or automation platform. Ask about tracked costs from Codex or another compatible chat with the included read-only MCP plugin.

Read the quickstart
// Usage metadata. No prompts. No provider secrets.
POST /api/v1/usage
Authorization: Bearer <project_key>

{
  "eventId": "brief-2026-10-23",
  "provider": "your-provider",
  "model": "your-model",
  "costUsd": 0.042
}
EXPLORE BEFORE YOU CONNECT

Try the workspace.
See what you could track.

Explore Tracewell

Explore sample data without signup. Plans are $19 or $49 per month; checkout remains on hold.