Anthropic's June 15 Billing Change: What Accounting, Law, and Consulting Firms Need to Know

Anthropic's June 15 Billing Change: What Accounting, Law, and Consulting Firms Need to Know

Here is the short version: if your firm uses Claude Pro and also runs Claude through third-party apps or automations, your usage is about to get divided into two separate buckets. Starting June 15, 2026, programmatic Claude usage — anything that goes through the API, the Agent SDK, or the claude -p command-line tool — draws from a dedicated credit pool that is separate from your standard chat credits.

For most casual Claude users, nothing changes. If you open Claude.ai in a browser and have a conversation, that is unaffected.

If your firm has started building workflows, using tools like Cowork, Firm360, Claude Finance, or Karbon Kai — or if someone on your team has set up automations that run Claude in the background — you need to understand what this means before June 15.

---

What Is Actually Changing

Claude Pro costs $20 per month. That covers your regular use of Claude's chat interface. Starting June 15, programmatic usage comes with its own credit allocation — also $20 per month for Pro subscribers.

Think of it as two separate fuel tanks. One for chat. One for automation.

If your firm's automations are running heavy — processing documents, pulling together reports, handling client intake drafts — the automation tank can run empty independently of your chat credits. When that happens, the automations stop working until your billing cycle resets.

This is not a price increase. It is a structural change to how usage is tracked and limited. But for firms that have quietly built Claude into their workflows over the last year, it is a meaningful shift.

---

Who This Affects

You are probably affected if:

• Your firm uses Cowork, Firm360, Claude Finance, Karbon Kai, or any other professional services tool that describes itself as "Claude-powered" or "built on Claude"
• Someone at your firm — or a consultant you hired — has set up automations using Claude's API or the Agent SDK
• You use the claude -p command in a terminal for batch processing
• You have connected Claude to other tools via integration platforms like Zapier, Make, or n8n

You are probably not affected if:

• You only use Claude at Claude.ai in a browser
• You have a Claude Team or Enterprise plan (those have higher limits and different structures)
• You use other AI tools entirely — ChatGPT, Gemini, Copilot — that are not Claude-based

The important word in the second list is "only." A lot of professional services firm owners have added Claude-powered apps to their stack without fully realizing they are running Claude behind the scenes. The next section covers how to find out.

---

What to Do Before June 15

This is a four-step process. You can do it in under an hour.

Step 1: Audit your Claude plan.

Log in to your Anthropic account at claude.ai/account. Confirm whether you are on Pro, Team, or another plan. Note your monthly renewal date. If you are on Pro and renewal happens before June 15, you may want to upgrade before then to avoid a mid-cycle gap.

Step 2: List the apps in your stack that use Claude.

Go through the tools your firm uses. For each one, ask: does this tool use Claude as its underlying AI? Common indicators include mentions of "Claude" in the tool's documentation, "powered by Anthropic," or integrations that ask you to connect your Claude or Anthropic account.

The tools most likely to draw from your agent credits: Cowork, Firm360, Claude Finance (Anthropic's template-based finance agents), Karbon Kai, and any custom automation your team built using the claude command or API keys associated with your personal Pro account.

Step 3: Confirm with each vendor.

Send a short message to each tool's support team. The question to ask: "Does your app draw from the user's Anthropic API credit pool, or does it run on your own infrastructure?"

The answer matters because tools that run on their own infrastructure — using Anthropic API keys they have licensed separately — do not count against your personal credit pool. Tools that use your credentials do. Most professional services apps use their own API access, but not all of them do, and the ones that do not will count against your programmatic credits starting June 15.

Step 4: Decide whether to upgrade.

If you find that your workflows are light (a few automations per week, mostly read-and-summarize tasks), Claude Pro's $20/month agent credit pool may be sufficient.

If your firm runs daily automations — document review, client report drafts, email triage, intake processing — Claude Team at $30 per user per month is worth the difference. The upgrade also gives each person on your team their own credit allocation rather than sharing a single pool.

If you have built custom workflows using the API directly, consider whether those should be migrated to Team or Enterprise-level API access with defined usage budgets.

---

What This Signals About AI Infrastructure

Anthropic splitting programmatic and chat usage is not arbitrary. It reflects something real: firms that have deployed Claude in workflows are using it very differently from firms that have a chat tab open occasionally.

A firm running Claude to process 200 client onboarding documents per month is generating a fundamentally different compute load than a firm that asks Claude to help draft an email three times a week. Separating those billing streams is Anthropic acknowledging that these are distinct use cases with distinct cost structures.

The broader signal: AI infrastructure for professional services firms is maturing. The era of a single flat subscription covering everything — chat, API, agents, integrations — is ending. As your firm's AI usage grows, you will increasingly need to track costs by workflow, not just by tool. Which automation is generating the most usage? Which client matters are driving the most AI spend? These are questions you will need to answer.

Firms that build that tracking discipline now will have a significant advantage when clients start asking what they are paying for in AI-assisted work. And they will start asking.

---

The One Thing to Do This Week

Before June 15, confirm whether any app in your firm's current stack uses your personal Anthropic credentials to run Claude programmatically.

Start with Cowork, Firm360, or Claude Finance if you use any of them. Check the settings or integrations section for API key configuration. If you find your own Anthropic key in there, contact the vendor this week to understand the usage implications.

If the sum of your automation use looks like it will exceed $20 per month in agent credits, upgrade to Claude Team before June 15. The cost difference — $10 per user per month — is smaller than the disruption of an automation stack going silent mid-month when you need it.

Ten days. Act now.

---

The Crossing Report covers AI developments affecting professional services firm owners every week. Subscribe here to get next week's issue.