AI Spend Report Generator
Turn your monthly numbers into a clean markdown spend report finance will accept.
Live report preview
# AI Spend Report — June 2026 ## Executive summary Total AI spend for June 2026 was **$9,740.00**, rising month over month (blended +10.2%). The biggest mover was **Anthropic (Claude API)** at +18% MoM ($6,200.00 this month). **Anthropic (Claude API)** is the largest line at $6,200.00 — 64% of vendor spend (high concentration — single-vendor risk worth flagging). ## Vendor breakdown | Vendor | Spend | MoM | Share | |---|---|---|---| | Anthropic (Claude API) | $6,200.00 | +18% | 64% | | OpenAI | $2,400.00 | -5% | 25% | | GitHub Copilot | $1,140.00 | +0% | 12% | | **Total** | **$9,740.00** | | | ## Top cost drivers - Platform team onboarded to Claude Code mid-month (+9 seats), driving the Anthropic increase. - OpenAI decline reflects migrating embeddings to batch pricing. ## Action items - Set per-team budget alerts at 80% before next cycle (owner: J. Park). - Review Copilot seats with zero activity for 30 days (owner: A. Chen). --- *Generated with the FORG AI Spend Report Generator (forg.pro/tools/spend-report-generator).*
How it works
This generator turns the numbers you already have — a monthly total, per-vendor amounts with month-over-month change, the cost drivers you know about, and your action items — into a clean markdown spend report that finance will actually accept. Fill the form on the left and the report builds live on the right; copy it or download it and paste it into your wiki, your email, or your budget review doc. Everything runs in your browser and none of your financial data is transmitted anywhere.
The executive summary is the part most engineers skip and most readers need, so the tool writes it for you — deterministically, from your numbers. It identifies the biggest mover (the vendor with the largest absolute month-over-month swing), computes each vendor's share of total spend to flag concentration, and characterizes the overall trend. Because every claim in the summary is derived from the table beneath it, a skeptical reviewer can verify the whole thing in thirty seconds, which is what separates a report that builds trust from one that invites questions.
Structure follows the inverted pyramid: answer first, evidence second, plumbing last. The vendor table carries dollar amounts, MoM deltas and share-of-total so trajectory and concentration are visible at a glance. Cost drivers turn alarming deltas into explained ones — "Claude spend up 40% because the platform team onboarded" reads very differently from an unexplained 40%. Action items with owners close the loop and give next month's report something to follow up on.
One honest note about where the numbers come from: this tool formats your data, it cannot verify it. If assembling the per-vendor figures means a monthly tour of five billing dashboards with screenshots, the report will eventually slip a month and then stop. Continuous per-vendor, per-team, per-developer tracking — the kind FORG provides — makes the data assembly automatic, so the monthly report becomes a five-minute paste instead of an afternoon of archaeology. The cadence is what controls cost; tooling is what sustains the cadence.
Frequently asked questions
What should a monthly AI spend report actually contain?
Four sections, in the order an executive reads them: a summary that states the total, the biggest mover and whether the trend is acceptable in three sentences; a per-vendor table with month-over-month deltas so concentration risk is visible; the named cost drivers behind any meaningful change; and action items with owners. Reports that open with methodology or raw data get skimmed; reports that open with the answer get read.
How is the executive summary generated from my numbers?
Deterministically, from arithmetic — no AI involved. The generator finds the vendor with the largest absolute month-over-month swing and names it as the biggest mover, computes each vendor's share of the total to flag concentration, and characterizes the overall trend from the weighted average of the deltas. The result is a summary a finance reviewer can verify against the table below it, which is exactly what makes it trustworthy.
Why report month-over-month percentage instead of just the dollar amounts?
Because the question finance asks is never the absolute number — it is whether the trajectory is intentional. A vendor at $4,000 flat for six months is boring in the best way; a vendor that doubled from $800 to $1,600 is a conversation even though the dollars are smaller. MoM deltas surface trajectory at a glance, and pairing them with named cost drivers turns a scary delta into an explained one, which is the difference between a budget review and a budget interrogation.
How do I get accurate per-vendor numbers in the first place?
Pull them from each provider's billing dashboard on the same day each month — consistency matters more than the exact date, because comparing a 28-day window against a 33-day one manufactures phantom deltas. Better still is continuous tracking that attributes spend per vendor, per team and per developer automatically; that is what FORG does, and it turns this report from a monthly scavenger hunt across dashboards into a copy-paste from one place.
Who should receive this report and how often?
Monthly to whoever owns the engineering budget, with the same structure every time so deltas are readable across months. Skip-level leadership generally only needs the executive summary; the full vendor table matters to the engineering managers who own the action items. The discipline of a fixed monthly cadence does more for cost control than any single optimization — anomalies that surface in a routine report get fixed in days, while anomalies discovered at invoice time get escalated.
FORG tracks this automatically across every agent session — live cost attribution, budgets, and alerts.
Start tracking with FORG