Business Policies & Rules
Your AI Agent Is Making Decisions for Your Team. Does It Know Your Business Rules?
What counts as 'enterprise'? When does an account become 'at-risk'? What triggers an escalation? Right now, that knowledge lives in three senior people's heads. The moment you deploy an AI agent without capturing that knowledge, it makes decisions based on its own assumptions — not yours.
Three Types of Business Rules That Need to Be Written Down
Not in a wiki nobody reads. In Fig — where every decision, every analysis, and every AI agent is held to the same standard.
Who counts as what
"Enterprise = ARR > $100K AND employees > 200."
When every team defines segments differently, analyses can't be compared. Store the definition once in Fig and every analysis, every agent, and every Flow uses the same one — so that you stop re-litigating what 'enterprise' means before every meeting.
When to act
"Alert when CAC exceeds $450 in any channel for 3+ consecutive days."
Thresholds that live in someone's head don't trigger automatically. Store them in Fig and monitoring fires the moment they're crossed — so that your team acts on the signal, not on whoever happened to notice first.
What requires approval
"No discount > 20% without VP Sales approval."
Decision guardrails that aren't enforced aren't guardrails. Fig checks every output against your constraints before it fires — so that your AI agents can't recommend actions that contradict your policies.
How Policies Work in Practice
Four steps from rule to enforcement — with accountability built in.
Write
Define the rule in plain English. Fig translates to enforceable logic.
Approve
Assign an owner. Policy is draft until approved by a named stakeholder.
Enforce
Fig checks every query, every agent call, every Flow output against active policies.
Review
Policies have review dates. Fig flags when a policy hasn't been confirmed as current.
Policies don't expire silently. Every policy has a review date and a named owner. When a policy hasn't been confirmed as current, Fig flags it — so your business logic stays accurate, not just documented.
Why Policies Matter for AI Agents
AI agents are powerful. But they only know what you tell them. Without your policies, they operate on generic assumptions — not your business rules.
“Your AI agent doesn't know your business rules. It knows the world's business rules.”
Give it yours.
“An agent that doesn't know your segment definitions is making up segments.”
Define once. Enforce always.
“When an AI makes a confident recommendation that contradicts your policies, nobody catches it.”
Fig blocks non-compliant outputs before they reach anyone.
Frequently Asked Questions
What is a business policy in Fig?+
A business policy is a rule that governs how Fig analyzes and acts on your data — how you define customer segments, what thresholds trigger an alert, which actions require approval. Policies are written once by your team, approved, and then enforced automatically on every query and every AI agent.
How do business policies affect AI agents?+
When an AI agent calls Fig, it operates within your active policies. An agent can't return a 'churned customer' list using a definition your business hasn't approved. It can't trigger an action that violates a constraint your team has set. Policies are the guardrails that make AI agents safe to deploy on real business decisions.
What types of policies can I create?+
Three types: segment definitions (how your business categorizes customers, accounts, or products), threshold rules (what values trigger alerts or actions), and decision constraints (what actions require human approval or are off-limits entirely).
Who can create and approve policies?+
Any team member can draft a policy. Approval is assigned to named owners — usually the team or person responsible for that domain. A policy in draft status has no effect until it's approved.
Explore the Platform
Metric Relationship Tree
Causal map of what drives what in your business
Learn moreMetric Definitions
Versioned, approved metrics for every agent and team
Learn moreBusiness Entities
Accounts, customers, products resolved across sources
Learn moreFlows
Signal → Decide → Act automated end-to-end
Learn moreMonitoring
24/7 anomaly detection with automatic root cause
Learn moreDecision Algorithms
12 built-in algorithms for governed analysis
Learn moreRoot Cause Analysis
Automated causal chain tracing
Learn moreIn-Memory Cache
Reduce warehouse costs, sub-second answers
Learn morePut Your Business Rules Where They Can Be Enforced.
Stop relying on senior people to remember the rules. Store your segment definitions, thresholds, and decision constraints in Fig — and let every human and every AI agent operate from the same playbook.