Should you build or buy Robotic Process Automation (RPA)?

Robotic process automation software scripts repetitive, rules-based work across desktop and web applications: data entry, form filling, and moving information between systems that lack a clean API. Back-office, finance, and operations teams buy it to automate manual work without rebuilding the underlying applications.

Do not renew on autopilot. Where applications have stable APIs or selectors and you have engineers, move those automations to code and cut the bot licenses; keep a commercial platform for desktop-UI, attended and audit-heavy processes. Compare the full cost of each existing bot, including break-fix, against the cost of rebuilding and governing its replacement.

Build it, buy it, or bridge?

⚒ Build it
✓ Buy it
➔ Bridge
Cost shape
A ten-workflow Python and Playwright stack models at $840K over three years.
The same ten workflows on commercial RPA license and orchestration cost $855K.
Move automations with stable APIs or selectors to code; keep commercial bots for desktop UI.
Time to value
BI Group moved 50+ processes to Python RPA with two developers and no vendor licenses.
A commercial platform ships orchestration, RBAC, and audit out of the box immediately.
Migrate stable, high-volume bots first; keep the platform for anything still shifting.
Differentiation captured
Owning the automation code gives no edge; it's back-office efficiency, not a product feature.
The vendor's orchestration, governance, and attended-bot tooling is the actual product.
No real differentiation on either path; the decision is entirely about total cost.
AI feasibility today
AI coding tools plus mature frameworks like Playwright make rebuilding RPA bots in code routine now.
Vendors are adding agentic AI on top of RPA, but it isn't yet displacing the orchestration core.
Use AI coding tools to rebuild stable-target automations; keep the vendor's orchestration for unstable UI work.
Who it fits
Teams with engineers and stable application targets, where license renewal is the trigger.
Processes with unstable UIs, heavy audit requirements, or attended, human-in-the-loop steps.
Most mid-size RPA estates, split by which automations have a stable target and which don't.

When building makes sense

Named production migrations off commercial RPA are real and specific, not theoretical. BI Group moved more than 50 processes from UiPath to Python-based automation, saving roughly 1,000 hours with a team of two developers. A medical-device manufacturer migrated more than 40 automations from UiPath to a Python-based RPA platform and reported a 70% reduction in licensing cost, plus bots running roughly twice as fast on pure Python. Imperial Treasure replaced its UiPath point-of-sale-to-SAP automation with a competing automation platform in October 2025, claiming 80% savings. These are code-based replacements using mature open frameworks like Playwright and Robot Framework, not experimental LLM computer-use agents, and the pattern only works where the target application has a stable API or predictable selectors. Build when you have engineers who can own the migration and your automations hit applications that don't change their UI every release. The payoff is a smaller license bill, not a new capability.

When buying makes sense

The migrations above are real, but they haven't dented the incumbent's business the way the narrative sometimes suggests: UiPath's quarterly revenue grew 13% year over year in mid-2026 with annual recurring revenue at $1.938 billion, so most customers are staying put. A ten-workflow, three-year cost model puts commercial RPA at $855,000 against $840,000 for a Python and Playwright stack, a gap too small to justify a migration on cost alone. What the platform keeps that code-based automation doesn't: orchestration, role-based access control, audit trails, and attended-bot support for processes where a human supervises each run. Renewing without checking the alternative is the actual mistake here, not the platform itself. Don't renew on autopilot: price the full cost of each existing bot, including break-fix time, against rebuilding it in code, and move the automations that hit stable targets while keeping the commercial platform for desktop-UI and audit-heavy work.

The desk read

Build-versus-buy analysis for Robotic Process Automation (RPA) is being written. In the meantime, the framework that drives every B4 call is on the B4 Index page.

Representative vendors Automation Anywhere Automation Success PlatformBlue PrismUiPath PlatformMicrosoft Power Automate + 2 more, listed in the full index

Frequently asked

What is robotic process automation (RPA) software?

RPA scripts repetitive, rules-based work across applications that lack a clean API, without rebuilding the underlying systems. Named companies including BI Group and a medical-device manufacturer have already retired commercial RPA licenses in favor of code-based automation on mature open frameworks.

Have real companies actually replaced commercial RPA with code?

Yes, with named results. BI Group moved 50+ processes to Python RPA and saved about 1,000 hours with two developers, and a medical-device manufacturer cut licensing cost 70% migrating 40+ automations from UiPath to a Python-based platform.

Is UiPath's business actually shrinking because of this?

No. UiPath's quarterly revenue grew 13% year over year in mid-2026 and annual recurring revenue reached $1.938 billion. The migrations are real, but they're scoped to specific automations with stable targets rather than a wholesale exit from commercial RPA.

Why isn't this a clear buy-or-build call either way?

Because the cost gap is thin. A modeled ten-workflow comparison puts commercial RPA at $855K and a Python-based rebuild at $840K over three years, close enough that the right call depends on your specific mix of stable-target and unstable-UI automations rather than on a single rule of thumb.

What are the main RPA vendors?

Representative vendors include Automation Anywhere Automation Success Platform, Blue Prism, UiPath Platform, Microsoft Power Automate. B4 Pro includes the category score and the full vendor list.

The B4 Index scores every software category on two axes, strategic differentiation and AI feasibility, to classify it Build, Buy, Bridge, or Beware. See the full methodology.