🚨 This is something I found in an ATS that might be silently hurting the job application experience without anyone noticing

I was exploring the CV upload process in a recruitment platform (ATS like Hiring Room) and found several interesting details from a UX and integration flow perspective.

The system offers multiple ways to complete a profile:

Upload CV with AI auto-parsing
Boomerang integration
ZonaJobs integration
Manual input

🧩 What I noticed in the flow

The Boomerang and ZonaJobs integrations can fail if they are used directly on the first attempt.

However, if the user first uploads a PDF CV using the AI parser and then goes back to select the integration, the flow works correctly and the auto-completion behaves much more consistently.

This suggests a possible dependency on the initial profile state or the CV parsing step before external integrations are properly enabled.

🤖 AI vs external integrations

Another interesting point is the difference in data quality between methods:

AI-based CV parsing still shows inaccuracies in data extraction and field assignment.
On the other hand, integrations such as Boomerang and ZonaJobs tend to be more structured, consistent, and accurate.

This reinforces an important idea: in ATS systems, structured data remains more reliable than direct PDF parsing through AI in many scenarios.

⚠️ Interesting flow optimization

One unexpected advantage of the AI-assisted flow is related to data pre-filling.

When the CV is initially parsed by AI, duplicate fields may be generated. However, in practice this can actually be beneficial because:

it is easier to review and remove duplicate information than to manually re-enter all the data from scratch

This makes the hybrid approach (AI + integrations) potentially more efficient when proper deduplication mechanisms are in place.

⚠️ Additional UX detail (captcha placement)

I also noticed a design issue related to the captcha placement.

The captcha is positioned in a low-visibility area within the flow. A more intuitive location would be between:

👉 the "Next" button 👉 and the CV upload method selection

This would improve:

visibility of the verification requirement
clarity of the overall flow
reduction of unexpected user errors

💡 Conclusion

Rather than isolated "bugs", this case illustrates how:

flow state can directly impact integrations
AI-based CV parsing still has limitations in accuracy
small UX decisions can significantly affect critical processes such as job applications

Article written by Pablo Marchetti Orayen – SolarisPKN Labs