.png)
At Siili, we’ve been working with Sitefinity for about 15 years. Together with Progress, we recently upgraded a fairly large and complex website of an educational institution. What made this one different was not the upgrade itself, but how AI agents were used throughout the process.
Here’s what we did, what worked, and where things still required manual effort.
The Challenge: A Real-World, Complex Upgrade
This was not a basic website. The setup included:
- Sitefinity MVC-based architecture with multiple controllers and views
- Training data module with scheduling logic
- Multi-system integrations (AWS services, Elasticsearch, Training backend, and more)
- Custom workflows, caching strategies, and event-driven logic
The upgrade path from Sitefinity 13.3 to 15.4 LTS was quite a jump considering that one major version was skipped.
In a traditional setup, this kind of upgrade would typically take about two or three weeks of effort for engineering when factoring in regression testing, debugging, and validation.
The Approach: AI-Assisted Upgrade and Testing
Instead of doing everything manually, the project team used Progress’ Sitefinity AI agents to support both the upgrade and testing.
Agents used:
- sf-upgrade-source-code-executor
- sf-post-upgrade-build-repairer
- sf-post-upgrade-runtime-repairer
- sf-post-upgrade-analyzer
- sf-test-generator
- sf-test-healer
- sf-test-dir-builder
The idea was simple: let the agents handle repetitive and structured tasks, and keep engineers focused on decisions and edge cases.
Build Tests First
One of the main risks in upgrades is not knowing what will break.
In most public website projects, comprehensive automated test coverage may be incomplete or outdated, often due to historical scope or maintenance constraints. That was the case here as well, so agents generated them:
- 76 tests in total (frontend + backend)
- Coverage across main navigation and custom admin areas
- ~2–3 hours to generate and stabilize
- Built with Playwright (TypeScript)
This gave us a baseline for both functional and visual validation (via visual regression testing).
The Upgrade
Once the tests were in place, the upgrade was straightforward:
- Execution time: under 1 hour
- Breaking changes: a few deprecated APIs
- Fixes: handled manually where needed
The agents took care of most of the mechanical work like code updates, fixes, analysis where usually goes a lot of the effort/time.
Validation: Fast Feedback Instead of Guessing
After the upgrade:
- Test results: 74 passed, 2 flaky
- Validation runtime: ~20 minutes
Only minor issues were detected manually:
- Inline editing not functioning
- UI quirk in template selection view
Both were fixed quickly.
The Outcome
End-to-end:
- About 8 hours total (tests + upgrade + validation) by AI agents.
Compared to the usual approach:
- About 50% less effort
- Days of manual work avoided
- Much lower risk due to automated testing
An important part is that we didn’t just finish the upgrade. We now have a working automated test suite that can be reused going forward.
What Actually Made the Difference
AI helped where the work is repetitive
Code updates, test generation, and fixes are all structured tasks this was very good fit for automation.
Testing stopped being a bottleneck
Instead of slowing things down, tests gave quick feedback and confidence.
Engineers focused on real problems
Less time on mechanical fixes, more time on edge cases and validation.
Lessons Learned
- Testing still takes time even with AI, stabilizing tests is a non-trivial step
- Integrations need manual review; there is no way around that
- Automation works best on predictable tasks
Why This Matters for Future Upgrades
This project demonstrates a shift in how digital platform upgrades can be approached:
- Faster execution
- Lower risk
- Better long-term maintainability
For organizations running Sitefinity (or similar platforms), the takeaway is clear:
Upgrades no longer have to be slow, risky, and resource heavy.
With the right combination of AI tooling and engineering expertise, they can become predictable, efficient, and even strategic.
Final Thoughts
At Siili, this collaboration with Progress showed what’s possible when AI is applied pragmatically not as hype, but as a tool to solve real engineering challenges.
The upgrade project wasn’t just successful it was a glimpse into how future platform maintenance will be done.
And if cutting upgrade effort in half becomes the norm, that’s a future worth building toward.
Related Case: HMS Networks Upgrade and Cloud Migration
If you are looking at a similar situation but on a larger scale, we’ve also worked on a full upgrade and lift-and-shift to the cloud for HMS Networks. That project covered both modernizing the Sitefinity setup and moving it to a cloud-based infrastructure, with focus on stability, performance, and maintainability across multiple sites.
See the full case
About Siili’s Sitefinity Expertise
That includes:
- Designing and developing complex digital solutions
- Complex upgrades across multiple major versions
- Cloud migrations and infrastructure changes
- Integrations with external systems and services
- Performance, scalability, and long-term maintenance
The focus is usually the same — keep things practical, avoid unnecessary complexity, and make sure the platform is something teams can actually work with long term.
About
