Spendline Guides

Field notes on AI cost, attribution, and FinOps.

Practical writing for engineering and finance teams that need to understand what AI costs per customer and how to control it. Written by the Spendline team.

Looking for current rates rather than an essay? See what each model costs per million tokens, with the source and checked date for each rate.

Comparing tools instead? See Spendline against specific alternatives or how it handles a specific problem.

Customer economics

What AI costs per customer, per run and per workflow, and how that compares with revenue.

The One-Week AI Cost-to-Serve Audit

A free, step-by-step method to measure what AI costs you per customer in one week: the fields to log, the queries, a worked example and the errors to avoid.
Spendline team · October 12, 2026 · Spendline facts last reviewed October 23, 2026

The Customer Who Costs You 50x More Than the Rest

AI usage varies by customer while the price they pay does not. Why the average hides your most expensive account, and the query that finds it.
Spendline team · October 4, 2026 · Spendline facts last reviewed October 23, 2026

What AI Did to Software Gross Margins

AI products averaged a 45% gross margin in 2025 against about 77% for SaaS. What moves the number, and why your own average hides the accounts that matter.
Spendline team · September 29, 2026 · Spendline facts last reviewed October 23, 2026

What an AI Agent Run Actually Costs

Agent runs vary up to 30x in tokens on the same task. Why per-request averages mislead, what drives the cost of a run, and how to cap it per run.
Spendline team · September 29, 2026 · Spendline facts last reviewed October 23, 2026

How to Calculate AI Gross Margin by Customer

The formula, the revenue mapping that trips teams up, a fully worked monthly calculation, cohort analysis, and what to do about negative-margin accounts.
Spendline team · August 25, 2026 · Spendline facts last reviewed October 23, 2026

How to Track LLM Costs Per Customer

How to tag, meter, price and reconcile LLM requests so you can calculate AI cost by customer, and what a proxy does and does not cover.
Spendline team · August 4, 2026 · Spendline facts last reviewed October 23, 2026

AI Has Broken SaaS Unit Economics

When inference becomes COGS, ARR is no longer enough. You need contribution margin by customer, workflow, and request.
Spendline team · May 1, 2026 · Spendline facts last reviewed October 23, 2026

Controls and integration

Budgets, attribution headers and how the tools in this space differ.

Best AI Spend Management Tools for SaaS Teams in 2026

Ten tools compared for SaaS companies that sell AI features: which record cost per customer, which can refuse a call before it reaches the provider, and where each one fits.
Spendline team · October 3, 2026 · Spendline facts last reviewed October 23, 2026

Langfuse Alternatives: Tracing, Evals, and Who Can Actually Stop Spend

A fair comparison of Langfuse alternatives for tracing and evals, and which tools can check an AI budget before the provider call as well as alert on cost.
Spendline team · September 22, 2026 · Spendline facts last reviewed October 23, 2026

Helicone Alternatives for Teams That Need Financial Governance

What Helicone does, what its status looked like on 23 October 2026, and a fair look at the alternatives, including which ones can check a budget before the provider call.
Spendline team · September 8, 2026 · Spendline facts last reviewed October 23, 2026

How to Stop Runaway AI Agent Spend

Why agents burn budget in minutes: loops, retry storms, context bloat and fan-out. The per-run caps, step limits and kill switches that actually stop it.
Spendline team · August 18, 2026 · Spendline facts last reviewed October 23, 2026

LLM Budget Enforcement: Alerts, Approvals, and Hard Caps

How to enforce LLM budgets before the provider call: the org, team, agent and customer hierarchy, block vs reroute vs approval, and the check-then-act race.
Spendline team · August 11, 2026 · Spendline facts last reviewed October 23, 2026

Anthropic API Pricing: What Claude Actually Costs and How to Control It

Anthropic API pricing per model, what really drives your Claude bill, a worked monthly example, and how to attribute and cap spend before the call.
Spendline team · August 7, 2026 · Spendline facts last reviewed October 23, 2026

AI Gateway, LLM Observability, or AI Spend Governance: What Each Layer Actually Does

What AI gateways, LLM observability, and spend governance each do, where they overlap, and how the three layers stack together.
Spendline team · May 2, 2026 · Spendline facts last reviewed October 23, 2026

Finance operations

Month-end reconciliation, cost classification and reporting standards.

How to Reconcile OpenAI and Anthropic Invoices

How to reconcile OpenAI and Anthropic invoices against your own cost ledger: the two bridges, the provider usage APIs, and a worked variance walk.
Spendline team · September 1, 2026 · Spendline facts last reviewed October 23, 2026

FOCUS 1.4 Made Token Spend Official. That's Not the Same as Making It Legible.

FOCUS 1.4 makes tokens a billing object. What the spec changes for vendors, and the attribution gap it leaves for finance teams.
Spendline team · June 14, 2026 · Spendline facts last reviewed October 23, 2026

AI COGS vs R&D: A CFO's Decision Tree for Classifying AI Spend on the P&L

When is AI inference COGS and when is fine-tuning R&D? A practical decision tree for classifying AI spend on the income statement.
Spendline team · May 1, 2026 · Spendline facts last reviewed October 23, 2026

The AI Month Close: How Finance Teams Reconcile LLM Spend Like Any Other COGS Line

How finance teams close the books on AI spend: the four-step reconciliation process and the controls auditors will eventually expect.
Spendline team · May 1, 2026 · Spendline facts last reviewed October 23, 2026