2026-05-14 – Claude for Small Business Launches With 15 Agentic Workflows; Agent SDK Gets Metered Credits

🧭 Claude for Small Business: 15 Workflows, 15 Skills, and a Nationwide AI Fluency Tour

Anthropic today launched Claude for Small Business — a toggle-install package of connectors, pre-built agentic workflows, and skills designed to let owners of small and medium companies put Claude to work inside the tools they already pay for, without hiring a developer. The launch is accompanied by an AI Fluency for Small Business tour kicking off in Chicago today, offering free half-day workshops to 100 local business leaders per stop.

What ships in the package

The package includes 15 ready-to-run agentic workflows and 15 skills covering the recurring tasks that small business owners told Anthropic slow them down most. Workflows span six functional areas:

Finance — payroll planning, month-end close, cash-flow forecasting, invoice chasing
Operations — contract review, vendor reconciliation
Sales — lead triage, customer pulse, campaign attribution (via HubSpot)
Marketing — content strategy, social copy generation (with Canva)
HR — onboarding task generation, policy Q&A
Customer service — dispute routing, refund handling (via PayPal)

Integrations

Claude for Small Business ships with pre-built MCP connectors for:

Intuit QuickBooks — payroll prep, monthly close, reconciliation, tax-season summaries
PayPal — settlements, invoicing, disputes, refunds
HubSpot — lead triage, pipeline health, campaign performance
Canva — brand asset creation from a marketing brief
DocuSign — contract review and signature tracking
Google Workspace & Microsoft 365 — document drafting, email triage, meeting summaries
Slack — daily digest, action-item extraction

A cross-connector workflow example

The flagship demo Claude shows: match QuickBooks cash balances with incoming PayPal settlements, build a 30-day rolling cash-flow forecast, identify overdue invoices, and draft a chaser email — all in one conversation. Previously that sequence required exporting a QuickBooks report, opening it in a spreadsheet, and manually drafting the email. With Claude for Small Business, it becomes a single prompt.

AI Fluency for Small Business Tour

Starting today in Chicago, Anthropic is partnering with PayPal on a free, half-day live workshop for 100 local business leaders per stop. Spring 2026 tour cities: Chicago, Tulsa, Dallas, Hamilton Township (NJ), Baton Rouge, Birmingham, Salt Lake City, Baltimore, San Jose, Indianapolis. The companion AI Fluency for Small Business online course is available at no cost on anthropic.com.

Partnership programmes

Anthropic is also joining forces with Workday and LISC on the Workday Foundation Solopreneurship Accelerator, which in 2026 will equip an initial cohort of 15 aspiring solopreneurs with seed funding, Claude credits, and an AI-first curriculum developed by LISC.

⭐⭐⭐ anthropic.com

🧭 Agent SDK Gets Its Own Credit Meter From June 15 — What Changes for Subscribers

Starting June 15, 2026, Anthropic is splitting subscription usage into two separate pools: your familiar interactive limit (browser chat, Claude Code in a terminal) stays unchanged, but programmatic Agent SDK usage — third-party harnesses, automated scripts, and CLI-driven pipelines — will draw from a new, non-rollover monthly Agent SDK credit. Axios reported the change today, framing it as a response to OpenAI actively recruiting heavy agent users frustrated with Anthropic's prior restrictions.

Credit amounts by plan

Pro ($20/mo) — $20 of Agent SDK credits/month, billed at standard API rates
Max 5× ($100/mo) — $100 of Agent SDK credits/month
Max 20× ($200/mo) — higher limit; exact amount in billing dashboard
Team / Enterprise — negotiated per seat; existing API agreements unaffected

Credits expire at month-end. Once exhausted, programmatic usage stops unless you enable extra usage billing (charged at pay-as-you-go API rates). Interactive usage — including Claude Code in terminal — is never drawn from the Agent SDK credit pool.

What counts as "programmatic" usage

Agent SDK scripts and harnesses (e.g. OpenClaw, custom Python clients)
claude -p non-interactive CLI mode
Any API call made with a subscription OAuth token rather than an API key

Action required before June 15 if you use automated pipelines

If you run Claude via subscription auth in CI, cron jobs, or third-party agent frameworks, audit your monthly token volume now. Compare it against your plan's credit allocation. If you consistently exceed the credit limit, either: (a) switch those pipelines to an API key (then usage is billed at API rates from the start, with no interruption); or (b) enable extra usage billing in your account settings before the cutover date. The credit system is separate from Claude Code's interactive terminal session — your daily coding workflow is unaffected.

# Check if you're using subscription OAuth or an API key:
# If your agent scripts set ANTHROPIC_API_KEY → API billing, unaffected
# If your scripts use `claude auth login` tokens → will consume Agent SDK credits

# Switch an automated pipeline to API billing:
export ANTHROPIC_API_KEY="sk-ant-..."   # set your API key
# Remove any ANTHROPIC_OAUTH_TOKEN env vars from the pipeline env

⭐⭐ axios.com