If you have started researching automation for your manufacturing operation, you have probably encountered the term RPA (Robotic Process Automation). You may have seen pricing that surprised you, timelines that seemed long, or vendors who want to run a six-week discovery engagement before telling you what anything costs.
That experience is a signal. You may be looking at tools designed for a different problem than the one you actually have.
This post covers the automation spectrum from low to high complexity, explains where RPA fits, and describes what most 10 to 50 person manufacturers actually need.
The Automation Spectrum
Think of business process automation as a ladder, not a binary choice:
Level 1: Spreadsheet macros and manual Zaps Someone at your company has built a spreadsheet formula that does a calculation automatically, or set up a Zapier connection that copies data from one app to another. This is automation, and it works well for simple, high-frequency tasks where the input is always consistent.
Level 2: Integration platforms Tools like Zapier, Make (formerly Integromat), and n8n connect software applications through their APIs. When X happens in system A, do Y in system B. This handles the majority of data transfer, notification, and synchronization needs for SMBs without custom code.
Level 3: API-based custom integration When off-the-shelf connectors are not enough, custom code calls an API directly. This is appropriate when the logic is complex, when data transformation is required, or when your systems are older and require a specific integration approach. Setup takes longer, but the result is more durable than a no-code connector.
Level 4: RPA (Robotic Process Automation) RPA software controls a computer like a human would: it opens applications, reads screens, clicks buttons, fills fields. It is designed for processes that involve legacy software with no API, where the only interface is the graphical user interface a human would use.
Level 5: AI agents and intelligent automation At the top of the stack, AI agents handle tasks that require judgment: reading unstructured documents, interpreting ambiguous input, making decisions based on context. These are appropriate when the process involves content that varies significantly from instance to instance.
What RPA Is Actually For
RPA exists to automate processes in systems that have no API. If your ERP is from 1998, has no integration layer, and the only way to get data in or out is to log in and click through screens, RPA can simulate that human interaction.
This is a real and valuable capability in large enterprises that run legacy core systems they cannot replace. For those organizations, RPA is sometimes the only viable option.
For most small manufacturers, RPA is overkill. If your ERP, CRM, quoting tool, and email platform all have APIs, you do not need a tool that simulates mouse clicks. You need a tool that calls the API directly. That is faster to build, cheaper to maintain, and more reliable over time.
Why Buyers Land on RPA When They Should Not
The RPA category is well-funded and well-marketed. If you search “automate my manufacturing workflow,” you will encounter RPA vendor content. It uses the right language. It addresses real pain points. The demos look impressive.
The problem is that the solution is calibrated for enterprises with 500-person IT departments and multi-year implementation budgets. The same demo that looks like your problem usually comes with a price tag and timeline that does not fit a 25-person job shop.
Additionally, the term “workflow automation” is sometimes used by RPA vendors to describe what they do, even though the underlying technology is screen-scraping rather than API integration. A direct API integration for the same task would be faster to build, less fragile, and less expensive to maintain.
What Most Small Manufacturers Actually Need
For the common automation problems in a 10 to 50 person manufacturing company, levels 2 and 3 handle most of the work:
- Inbound inquiry parsing and CRM logging: Level 2 or 3 integration
- Quote-to-CRM sync: Level 2 integration
- ERP job creation from won deals: Level 3 if the ERP has an API; potentially Level 4 if it does not
- Follow-up sequences: Level 2 automation
- Automated reporting: Level 2 or 3
The situations that move up the stack toward RPA or AI:
- Legacy ERP with no integration layer: Level 4
- Processing handwritten or highly variable documents: Level 5
- Making pricing or scope decisions from unstructured input: Level 5
Cost Comparison
This is approximate but directional:
| Level | Typical build cost | Typical monthly cost |
|---|---|---|
| Integration platform (Level 2) | $2,000 to $8,000 | $50 to $300 platform fees |
| Custom API integration (Level 3) | $5,000 to $20,000 | Minimal |
| RPA implementation (Level 4) | $30,000 to $150,000+ | $10,000+/yr licensing |
| AI agent (Level 5) | $10,000 to $50,000 | Usage-based, varies |
For a shop spending 10 hours per week on manual data entry, a Level 2 or 3 integration that costs $8,000 upfront typically pays back in two to four months. An RPA solution for the same problem at $50,000 might take two years to break even, assuming it works as expected.
The Right Question to Ask
Before you engage any automation vendor, ask one question: does your solution require screen-scraping, or does it use direct API integration?
If the answer is screen-scraping and your systems have APIs, you are in the wrong conversation. Screen-scraping is fragile. Every time the software vendor updates their UI, your automation breaks. API integrations are stable because they call a published interface that the vendor maintains and documents.
If your systems genuinely have no APIs, then RPA may be the correct answer. But confirm that first, because the exception is less common than the vendors who specialize in it would suggest.
Getting the Right Fit
A 30-minute assessment will tell you which level of automation fits your actual workflow and which of your systems have integration paths that make the project tractable. No vendor pitch, no RPA demo. Just an honest mapping of your problem to the right solution tier.
