Allocate model spend to teams for internal reporting

Turn **Jobs** token rows ([View your audit logs](/tutorials/view-your-audit-logs)) into a **finance-friendly allocation** — map work to **teams** using agents, keys, and naming discipline — so departmental visibility spreadsheets defend budgets without pretending Auxot replaces formal accounting.

Plus: three Admin-Agent passes — draft an allocation rubric from your agent directory, propose spreadsheet columns for an API export habit ([Create a shared Team API Key](/tutorials/create-a-team-api-key)), and narrate double-count risks when everyone shares one login.

Audience	Admins · Executives · Developers
Time	~12 min
Prerequisites	Monthly usage instinct ([Stay ahead of agent usage and cost](/tutorials/stay-ahead-of-agent-usage-and-cost)). Comfortable filtering Audit Logs → Jobs ([View your audit logs](/tutorials/view-your-audit-logs)). Helpful: Business multi-team walls ([Set up multi-team isolation](/tutorials/set-up-multi-team-isolation)), `team.` automation keys ([Create a shared Team API Key](/tutorials/create-a-team-api-key), [Pick personal or team API keys for real work](/tutorials/pick-personal-or-team-api-keys-for-real-work)).
You'll end up with	A one-page allocation policy — which signals mean Team A vs Team B, how shared agents split, and where humans adjust monthly — plus a stub spreadsheet schema keyed off Jobs exports.

When a tutorial shows italic text in quotation marks, it usually mirrors a label or helper string inside Auxot. Product copy changes between releases — if something reads differently in your workspace, trust what you see on screen.

Callouts with a Worth knowing gold accent are meant as must-read context before you move on. Blockquotes that open with Tip are lighter, optional depth.

Why this matters

Finance asks a blunt question: which department burned the tokens? Audit Logs → Jobs already shows per-job input/output counts (View your audit logs): the gap is attribution, not vanity charts.

Auxot does not ship a magical cost center dropdown on every completion today. What it does give you are strong proxies:

Agents scoped to teams (Set up multi-team isolation): ownership maps fairly cleanly when agent ownership stays disciplined (Build your agent directory).
Automation authenticated with team. keys (Create a shared Team API Key): service traffic can align to the team that owns the key (Pick personal or team API keys for real work).
Intentional labels in workflows or intake JSON (Run a workflow, Trigger a workflow with an intake webhook): boring metadata beats heroic archaeology later (Batch spreadsheet rows through a workflow).

Internal reporting is education: cross-charging between departments may need tighter contracts with finance; this lesson stays at defensible allocation, not accounting doctrine.

Nothing lands in the right general ledger code on its own — you write the mapping rules, you reconcile edge cases.

Quick start

Pick allocation spine — simplest honest path wins: team-owned agents first, shared agents split by rule (even Headcount, revenue proxy, or manual percent: document which).
Freeze agent→team table — export or screenshot Settings → Agents alignment; publish beside your agent directory (Build your agent directory).
Pull Jobs slice — UI filters by time range + agent names, or Management API with team. key where automation demands spreadsheets (Create a shared Team API Key, Back up and export your Auxot data).
Join tokens — spreadsheet job_id, agent, tokens in/out, and derived_team column via VLOOKUP / join against your table: sanity-check totals vs Pick personal or team API keys for real work intuition.
Publish monthly — same day each month; deviations note model migrations (Migrate agents when models or providers change) or new shared agents (Chain steps so agents hand off cleanly).

Done? Allocation memo with method paragraph finance can quote, and explicit shared-resource caveats.

The agent can do that?

1. Rubric from directory prose

Chat → Admin Agent:

Paste our agent directory markdown ([Link your chat apps to your Auxot account](/tutorials/link-your-chat-apps-to-your-auxot-account)). Propose allocation rule: when agent row lists Team X, assign 100% to X; list shared — flag for manual split — table only.

Why it’s non-obvious: Directories go stale: conversational pass surfaces orphans after you paste; you still reconcile Settings.

2. Spreadsheet schema

We're exporting Jobs rows monthly — propose columns: identifiers, token math, team derivation, and confidence flag — markdown — assume finance reads CSV.

Why it’s non-obvious: Without confidence column, every estimate looks falsely precise until you label guesses.

Explain in five bullets why one shared admin login breaks departmental allocation — audience engineering leads — tie to personal vs team keys ([Pick personal or team API keys for real work](/tutorials/pick-personal-or-team-api-keys-for-real-work)).

Why it’s non-obvious: Tooling debates hide identity discipline: paste once before procurement argues dashboards lie.

Go deeper

Dashboards as narration

Pin a usage-shaped widget for executives (Build your first Dashboard): still not legal allocation proof without the spreadsheet spine.

Quarterly roster reviews

Quarterly ownership review belongs in Run a quarterly review of your agents: unsettled agents corrupt six months of history silently.

Leadership slides

Exec summaries pair with Brief leadership on your agent program: finance appendix carries methodology fine print.

Provider bills

Cloud invoices remain vendor truth: reconcile Auxot Jobs totals as directional unless metering pipelines merge (Connect a cloud AI model, Stay ahead of agent usage and cost).

Walkthrough

Step 1: Convene finance + platform

Agree goal: visibility for planning vs invoicing another department: wrong expectation breeds mistrust.

Step 2: Build agent→team lookup v1

CSV from manual audit acceptable: perfection waits.

Step 3: Export Jobs window

Month closed → export → freeze file jobs-YYYY-MM.csv.

Step 4: Allocate

Apply spine rules: mark LOW_CONFIDENCE rows for shared automation.

Step 5: Review meeting

15 minutes: anomalies list → retire stale agents (Audit and clean up your agents) if they distort totals.

What’s next

→ Stay ahead of agent usage and cost. Monthly token literacy before allocation debates matter.
→ Route intent to the right specialist agent. Clean routing labels beat “everyone uses the smart one” when you split spend by team.
→ Set up multi-team isolation. Walls that make team a real column in your org.
→ Build your agent directory. Human-readable ownership map feeding lookups.
→ View your audit logs. Where Jobs rows and token counts live.
→ Brief leadership on your agent program. Stitch totals into the story executives repeat.

Reference

Manual: API overview (export patterns), Teams
Pages in Auxot: Audit Logs → Jobs, Settings → Agents, Settings → Teams, Dashboards
See also: Answer vendor security questionnaires from your own evidence, Create a shared Team API Key, Pick personal or team API keys for real work, Migrate agents when models or providers change