This page covers how we use AI at Float — what tools we use, what's allowed, how to keep data safe, and how to get integrations approved.

This is one of three connected docs that make up our AI operations infrastructure:

1. AI Strategy — the why: our mindset, principles, and direction
2. AI: Practical and ethical use — (you are here 😉) the rules: guardrails, data classification, and integration approvals
3. 📚 AI Playbooks — the how: team-specific use cases and practical examples

At the core of this is GOOSE 🦆 (Governance and Oversight Of Smart Engines) — our AI decision framework.

<aside>

If something isn't covered, raise it in :slack: #open-it-ops

</aside>

<aside>

Quick guide

• Use Claude for day-to-day work
• Don't paste sensitive data (PII, financials, customer data)
• Always review outputs before sharing
• Only use approved integrations
• When unsure → #open-it-ops </aside>

Our AI tools 🤖

Primary tool

• Claude (Anthropic) — day-to-day work: comms, ops, drafting, research, analysis

<aside>

🪑 Need a Claude seat? Go to claude.ai, sign up with your @float.com email, and you'll be prompted to request to join the Float workspace.

</aside>

Other tools

• Fin (Intercom) — customer support AI
• ChatGPT / OpenAI — engineering only (product POCs)
• Breeze (HubSpot) — CRM, used by Sales & CX