How-to Guides

How do I export AI spend to my FinOps platform?

Download a FOCUS-aligned CSV from the Cognocient AI dashboard or pull it via API. Load it into Apptio, CloudZero, Spot, or any FinOps tool that supports FOCUS.

In the app: Reports → FOCUS Export · Growth plan and aboveOpen in app →No account? Start free

Goal: AI spend data in your existing FinOps platform — Apptio, CloudZero, Spot, or a data warehouse — so your team sees AI costs alongside cloud costs in one view.

Time: 10 minutes for a manual export. 30 minutes to automate via API.


Step 1 — Download a FOCUS export from the dashboard

On the Reports page or the Chargeback page, find the FOCUS Export card:

  1. Select the date range (e.g., last month)
  2. Click Download FOCUS CSV

The file is ready immediately. The export is CSV only, with one row per UTC day for each provider, model and tag combination.

Step 2 — Load into your FinOps platform

DestinationHow to load it
FinOps platform (Apptio Cloudability, CloudZero, Spot and others)Add the CSV as a custom or FOCUS data source, following your platform's own import guide
Snowflake / BigQuery / RedshiftLoad the CSV directly — join on BillingAccountId + BillingPeriodStart

Column names follow FOCUS conventions, but we haven't certified the import with any specific platform, so check the column mapping on your first load. Department and feature come through in Tags, and the model in SkuPriceDetails.

Step 3 — Automate with the API (for monthly pipelines)

On the Business plan you can skip this step: a FinOps connector pushes the export to Vantage, CloudZero or S3 daily. The API below is for any other destination.

Pull the export programmatically so your pipeline doesn't require a manual step:

# `from` and `to` are inclusive calendar days (UTC)
curl "https://api.cognocient.com/api/chargeback/exports/focus?from=2026-05-01&to=2026-05-31" \
  -H "Authorization: Bearer sk-cog-YOUR-PROXY-KEY" \
  -o cognocient-may-2026.csv

Schedule this as a monthly cron job (run on the 2nd of each month, pulling the prior month's data):

import httpx
from datetime import date, timedelta
from dateutil.relativedelta import relativedelta
 
def export_last_month():
    today = date.today()
    first_of_this_month = today.replace(day=1)
    first_of_last_month = first_of_this_month - relativedelta(months=1)
    last_of_last_month  = first_of_this_month - timedelta(days=1)
 
    resp = httpx.get(
        "https://api.cognocient.com/api/chargeback/exports/focus",
        headers={"Authorization": f"Bearer {COGNOCIENT_KEY}"},
        params={
            "from":   first_of_last_month.isoformat(),
            "to":     last_of_last_month.isoformat(),
        },
    )
    resp.raise_for_status()
 
    filename = f"cognocient-{first_of_last_month.strftime('%Y-%m')}.csv"
    with open(filename, "wb") as f:
        f.write(resp.content)
 
    return filename

Step 4 — Key FOCUS columns for FinOps reconciliation

FOCUS columnWhat it containsFinOps use
ProviderNameThe company billing you: OpenAI, Anthropic, AWS...Filter by AI provider
ServiceNameThe provider's API product: OpenAI API, Claude API, Amazon Bedrock...Cost by service
SkuPriceDetailsJSON with the model (x_ModelId) and its developerCost by model
SkuMeterInput Tokens, Cache Read Input Tokens, Output Tokens...Cache savings, input vs. output spend
BilledCostActual API cost in USDJoin against cloud billing
ConsumedQuantityTokens (or images, seconds...) for that meterToken-level unit economics
TagsJSON map of your attribution headers, e.g. {"department":"engineering","feature":"search"}Per-feature and per-department cost drilling

Step 5 — Reconcile AI spend against cloud billing

To get a unified view of all infrastructure + AI spend:

  1. Download Cognocient AI FOCUS export for the billing period
  2. Download your cloud billing export (AWS CUR / Azure export / GCP BigQuery billing)
  3. Load both into your data warehouse
  4. Join on BillingAccountId + BillingPeriodStart
  5. Filter on the department key in Tags to generate per-department chargebacks that include both cloud and AI spend on one line

On this page