Practical frameworks for deciding when automation makes sense.

Not a philosophy essay — a practical resource for self-assessment. Use these frameworks to figure out if you should build, buy, or skip automation entirely.

Is Your Process Actually Automatable?

Not everything should be automated. Here's a quick decision matrix.

Automate

High volume: 50+ documents/week
Consistent format: Same vendors, same layouts
Clear rules: If X, then Y — no judgment calls

Example: Processing invoices from your top 10 vendors. Same format every time, 200+ per month, extract the same 8 fields.

Partial

Variable formats: Different vendors, inconsistent layouts
Some judgment needed: Edge cases require human review
Exceptions common: 20-30% need manual handling

Example: Contracts from multiple parties. Extract key clauses, flag unusual terms, but lawyer reviews all flagged items.

Skip

Low volume: <10 items/week
Highly variable: Every document is different
Requires expertise: Domain knowledge needed for each decision

Example: Complex legal analysis of unique M&A documents. Each one is different, requires lawyer expertise, not worth automating.

What LLM API Costs Actually Look Like

Real numbers for common document processing tasks. These are operational costs — development is separate.

Process 500 invoices/month

~$200/month in Claude API costs

Standard B2B invoices, ~$0.40 per invoice average

Extract data from 1,000 contracts/month

~$400/month in GPT-4 costs

Multi-page contracts, key clause extraction

Email categorization (300/week)

~$150/month

Route to categories, extract key info, create tickets

For context:

VA at $15/hr for 20 hrs/week = $1,200/month
Full-time data entry clerk = $3,000+/month

Note: These are operational costs only. Development costs are separate — see Services & Pricing for typical project ranges.

Build vs Buy Decision Framework

Four questions to determine if custom development makes sense.

Volume: How many items per month?

<100: Probably not worth custom development
100-1,000: Sweet spot for automation
>1,000: Definitely automate — cost savings compound

Consistency: How variable is the input?

High consistency: Same format, same fields → simple extraction
Medium: Some variation → handle with prompting + validation
Low consistency: Every doc different → harder, may need human review

Integration: How many systems need to talk?

1-2 systems: Simple webhooks or API calls
3-4 systems: Moderate complexity, still manageable
5+ systems: Consider Zapier/Make or a real integration platform

Maintenance: Do you have technical capacity?

Yes: You can maintain custom code after handoff
No: Buy a solution with vendor support, or budget for ongoing help

Quick guide:

Build: High volume + consistent + 1-3 systems + tech capacity
Buy: Low volume + variable + 5+ systems + no tech capacity
Hybrid: Build the core logic, use existing tools for integration

When to Just Buy a Tool

Sometimes the best automation is admitting you don't need custom work. This is where I tell you not to hire me.

CRM automation: Use HubSpot, Salesforce, or Pipedrive built-in workflows. Don't build custom.
Multi-system workflows (5+ tools): Use Zapier or Make. They've already solved the integration headaches.
Email marketing automation: Mailchimp, ConvertKit, or Klaviyo. Not worth building.
Standard accounting/invoicing: QuickBooks, Xero, FreshBooks. Just use them.
Basic form processing: Typeform + Zapier. Don't overcomplicate it.

Why am I telling you this? Because my job is to give you the right answer, not to maximize billable hours. If buying is smarter than building, I'll tell you — usually in the first call.

Still not sure?

That's exactly what the first call is for. Tell me what you're trying to automate, and I'll give you an honest assessment — including whether you should just buy something off the shelf.

Get in touch See case studies

↑ Back to top