NewCost per customer from OpenRouter and OpenTelemetry

See where your AI spend goes.

Know what every AI dollar earns.

Your providers send you a total. Lamdy breaks it down by model, key, customer and day, then lines it up with your Stripe revenue, so you see the margin and not just the bill.

Start free See how it works

Free up to $2,500 of AI spend a month. Read-only keys, no SDK, no proxy, no credit card.

lamdy.ai/dashboardSynced just nowSample data

AI spend

Revenue

Gross margin

Month-end forecast

RevenueAI spendLast 30 days

CustomerAI costRevenueMargin

Acme Co$4,120$19,800+79%

Northwind$3,050$12,400+75%

Nova Labs$2,900$2,100-38%

Brightside$1,780$8,600+79%

Reads billing from the providers you already use

OpenAI
Anthropic
Azure OpenAI
OpenRouter
Vercel AI Gateway
Cursor
Deepgram
ElevenLabs

The problem

Your providers send one number at the end of the month. It doesn't say which customers cost the most, which features are underwater, or whether AI spend is growing faster than revenue.

Lamdy turns that number into a ledger.

How it works

Set up in 5 minutes. No code changes.

1
Paste read-only keys
Connect OpenAI, Anthropic, Azure OpenAI, OpenRouter, Vercel AI Gateway, Cursor, Deepgram or ElevenLabs. Lamdy reads billing data only. It never sends requests or adds cost to your account.
2
See spend by provider, model, team and feature
Daily breakdowns, model comparison, trends and month-end forecasts, without building a single dashboard.
3
Connect Stripe. See which customers are profitable.
Gross margin per customer and per feature. A key per customer works as is; for a shared key, turn on OpenRouter Broadcast or OpenTelemetry, or send a usage event per AI call.

What you get

Every AI dollar, accounted for. From the provider bill to the customer who caused it.

What makes Lamdy differentSample data

Acme Co+79% margin

Revenue

$19,800

AI cost

$4,120

Northwind+75% margin

Revenue

$12,400

AI cost

$3,050

Nova Labs-38% margin

Revenue

$2,100

AI cost

$2,900

Brightside+79% margin

Revenue

$8,600

AI cost

$1,780

Know which customers are profitable. Revenue from Stripe next to the AI cost each customer drives. When every customer shares one API key, turn on OpenRouter Broadcast, point your OpenTelemetry traces at Lamdy, or send one small event per AI call. Lamdy prices each call at your real billed rates and never stores prompts.

OpenRouter Broadcastno code
OpenTelemetryVercel AI SDK and more
Usage events APIone request per call

Sample data

AI spend, last 30 days

OpenAI

$7,940

Anthropic

$5,120

Azure OpenAI

$2,860

OpenRouter

$1,540

Others

$780

Every provider, one dashboard. Total spend, daily trends and month-end forecasts across all your providers.

Sample data

gpt-4o$4.40 / 1K req

claude-sonnet-4$3.10 / 1K req

gpt-4o-mini$0.40 / 1K req

Switch gpt-4o to gpt-4o-mini on classification: save $1,240 a month

Compare every model. Cost per request and per token, and where a cheaper model would do the job.

Sample data

Budget threshold

80% of $8,000 monthly budget

sent

Spend spike

OpenAI up 42% day over day

sent

Forecast overrun

Projected to exceed budget by Sep 28

watching

Know before the bill arrives. Budget, spike and forecast alerts in Slack, Discord or email.

Sample data

API key …3f7aSupport Bot

Project prod-chatCustomer chat

Project internal-toolsInternal (unbilled)

Map spend to teams and features. Rules on the API keys, projects and models you already have. No code changes.

Sample data

Savings found this month

$0/mo

Switch gpt-4o → gpt-4o-mini for classification calls$1,240
Enable prompt caching on repeated system prompts$610
Remove 3 idle API keys with no traffic in 30 days$290

Find savings you're missing. Model switches, prompt caching, idle keys. Each with its monthly saving.

Questions Lamdy answers

Ask what the invoice can't tell you.

Which customer is losing us money?

LamdySample data

Nova Labs costs more than it pays. Most of it is claude-sonnet-5 on support chat.

AI cost

$2,900

Revenue

$2,100

Margin

−38%

One key, shared by everyone?

Cost per customer, even on a shared key. No SDK to install.

Read the guide

request.jsonSending…

Create an ingest key
Write-only. It can add usage, never read your data. Revoke it in one click.
Point your sender at Lamdy
Turn on OpenRouter Broadcast, add an OpenTelemetry exporter, or post one event per AI call.
See cost per customer
Priced at the rates on your own bill. Retries count once. Prompts are never stored.

Pricing

Simple pricing. Scales with your AI spend.

Every plan includes all provider connections. Upgrade when you need alerts, margins, or more history.

Free

$0/mo

Everything you need to see where your AI spend goes.

Up to $2,500 managed AI spend
1 team member
All provider connections
Spend dashboard & model comparison
30-day data history

Start free

Growth

$25/mo

Get alerts before your next surprise bill.

Up to $6,000 managed AI spend
5 team members
Email, Slack & Discord alerts
Spend forecasting
Attribution rules
90-day history · CSV export

Start with Growth

Frequently asked questions

What is managed AI spend?

Managed spend is the total AI provider spend that Lamdy monitors across your connected accounts. It's what you pay OpenAI, Anthropic, etc. — not what you pay Lamdy. If your OpenAI bill is $3,000/mo, that's $3K in managed spend.

Do I need to change my code?

Not for spend. Lamdy connects to your provider accounts using read-only keys and reads billing and usage data only — spend by provider, model, project and API key, alerts, forecasts and savings need no SDK, no proxy and no code changes. Lamdy never sends requests to your AI providers or incurs costs on your account.

Do I need code changes to see margin per customer?

No, if each customer has their own API key or workspace: an Attribution rule maps it to the customer, and with Stripe connected you see that customer's margin. If customers share one key, Lamdy can split cost per customer from usage your app sends: turn on OpenRouter Broadcast, point an OpenTelemetry exporter at Lamdy, or send one small request per AI call. Prompts and responses are never stored.

Is my data safe?

Yes. Lamdy uses read-only credentials and never stores prompt or response content. We only read billing metadata: dates, models, token counts, and costs. API keys are encrypted at rest and never displayed after initial entry.

How is Lamdy different from observability tools?

Observability tools like Langfuse and LangSmith track request-level traces, latency, and prompt quality. Lamdy tracks dollars: what you spend, where it goes, and whether it's profitable. We connect AI costs to Stripe revenue to show margins. Observability tells you how your AI works. Lamdy tells you whether it's making money.

Can I use Lamdy with my AI gateway?

Yes. Lamdy reads billing from Vercel AI Gateway and OpenRouter, as well as OpenAI, Anthropic, Azure OpenAI, Cursor, Deepgram and ElevenLabs. Google Gemini, Together AI and Fireworks AI are coming with the Lamdy SDK.

What happens if I exceed my managed spend limit?

Nothing breaks. Lamdy will notify you and suggest upgrading. Your dashboard keeps working. We don't cut off your data.

About

Why I'm building Lamdy

I studied physics and chemistry before computer programming — I like being precise about what's actually true, not what looks true. That habit followed me into building with AI: I was running things across several providers, and more than once a key ran dry mid-project with no real sense of which provider was even worth paying for.

I didn't have a good answer to a simple question: is this AI spend working, or just adding up? So I built a way to see it — cost next to what it was actually producing. Lamdy is that tool, built for myself first, now open for anyone asking the same question.

Mahamat Adoum

Founder, Lamdy

X (Twitter)LinkedIn

Start free

If that sounds familiar, see your own numbers in five minutes.

Stop guessing. See your AI margins.

Free up to $2,500 of AI spend a month. Connect a read-only key and see your numbers in about five minutes.

Start free Read the docs

No credit card required

Why I'm building Lamdy

Mahamat Adoum

Founder, Lamdy

X (Twitter)LinkedIn

Start free

If that sounds familiar, see your own numbers in five minutes.

Know what every AI dollar earns.

Set up in 5 minutes. No code changes.

Paste read-only keys

See spend by provider, model, team and feature

Connect Stripe. See which customers are profitable.

Every AI dollar, accounted for. From the provider bill to the customer who caused it.

Ask what the invoice can't tell you.

Cost per customer, even on a shared key. No SDK to install.

Create an ingest key

Point your sender at Lamdy

See cost per customer

Simple pricing. Scales with your AI spend.

Frequently asked questions

Why I'm building Lamdy

Stop guessing. See your AI margins.

Know what every AI dollar earns.

Set up in 5 minutes. No code changes.

Paste read-only keys

See spend by provider, model, team and feature

Connect Stripe. See which customers are profitable.

Every AI dollar, accounted for. From the provider bill to the customer who caused it.

Ask what the invoice can't tell you.

Cost per customer, even on a shared key. No SDK to install.

Create an ingest key

Point your sender at Lamdy

See cost per customer

Simple pricing. Scales with your AI spend.

Frequently asked questions

Why I'm building Lamdy

Stop guessing. See your AI margins.