You can't audit what you don't understand.
Before you start hunting for AI opportunities, you need a clear picture of your business as it actually operates today—not how you think it operates, or how it's supposed to operate.
This step is about gathering the facts. The real ones.
Here's what you need to know—and why it matters.
1. Find Your "North Star"
Start with direction.
Write down your top 3 business goals for the next 12 months. These could be revenue targets, customer growth, new product launches—whatever matters most.
Then write down your top 3 constraints. Budget limits. Compliance requirements. Headcount freezes. Tight timelines.
Why this matters: AI should help you hit your goals faster or remove constraints that slow you down. If you don't know what those are, you'll build the wrong thing.
2. Map Your Operations
Get the org chart out of your head and onto paper.
List every department in your company and what each one is responsible for. Sales. Marketing. Operations. Support. Finance. HR.
Then write down the KPIs each department uses—even if they're messy or incomplete.
Why this matters: AI opportunities live inside departments. If you don't know who owns what, you won't know who to talk to or where the biggest pain points are.
3. Inventory Your Tech Stack
Every system. Every tool. Everything.
Make a list of every piece of software your business uses:
- Sales/CRM (Salesforce, HubSpot, Pipedrive)
- Marketing (email tools, ad platforms, analytics)
- Operations/ERP (inventory, scheduling, project management)
- Support (Zendesk, Intercom, helpdesk software)
- Finance/Accounting (QuickBooks, Xero, NetSuite)
- HR (payroll, benefits, time tracking)
- Communication (email, Slack, Teams)
- File storage (Google Drive, Dropbox, SharePoint)
For each system, capture:
- Who owns it
- What it's used for
- What key data lives inside it
- Where that data comes from and where it goes
Why this matters: Most AI opportunities involve connecting systems or pulling data from multiple places. If you don't know what you're working with, you'll miss the biggest wins.
4. Document Your AI History
What have you already tried?
List any AI tools your team is already using—official or unofficial ("shadow AI" that people adopted on their own).
Then document any AI projects that failed:
- What was it supposed to do?
- Why did it fail? (Bad data? No adoption? Poor quality? Security issues?)
- What will your team absolutely not tolerate again?
Why this matters: Past failures teach you what to avoid. If your team tried AI once and it burned them, you need to know why—so you don't repeat it.
5. Plan Your Access
Figure out what you need and who can give it to you.
Decide what access you'll need right away (Week 1) versus later (Weeks 2–4).
Write down:
- Who will grant access
- When you need it by
- What specific data or systems you need access to
Why this matters: You can't audit what you can't see. If you wait until week 3 to ask for access, you've already wasted two weeks.
6. Write a One-Page Summary
Pull it all together.
Create a single-page document that includes:
- Your top 3 goals and constraints
- Department breakdown
- Tech stack highlights
- Data sensitivity rules (what's confidential, what can't be shared)
- AI history (tools tried, what worked, what didn't)
- Known bottlenecks (if you already know where things break)
Save it with a clear name: YourCompany_IntakeSummary_v1
Why this matters: This summary becomes your reference point for the entire audit. You'll come back to it again and again.
What You Should Have Now
By the end of this step, you should have:
✅ A clear folder structure to keep everything organized
✅ A complete tech stack inventory with owners and data flows
✅ An AI history log showing what's been tried and what failed
✅ An access plan with names, deadlines, and specific scope
✅ A one-page intake summary that anyone can read and understand
Quality Check: Did You Do This Right?
Before you move on, make sure:
- Your intake summary is one page and readable by someone who isn't technical
- Your tech stack list is complete—no major system should show up later as a surprise
- Every major system has a clear owner and purpose
- Your data sensitivity is clearly labeled: what's sensitive, what can't be shared, what must stay internal
- Your access plan has a named approver, a deadline, and specific scope (not just "give us access")
Need Help Getting This Done Fast? (Free Tool)
We built a tool that generates your company information summary in minutes.
Use it here: Generate Your Summary
Rather have experts handle this? Learn how we work → OpsSystem.ai