CV Parsing with AI

Automatically populate a candidate’s record using AI

Firefish will process any CV added to a candidate record using artificial intelligence (AI) to structure and populate candidate information into the main record profile.

CVs are parsed automatically by the CV Processor whenever a readable document is added – this includes all registration methods, direct apply and when manually adding a document to a candidate’s record.

What Fields are Populated?

Firefish’s AI Parser will attempt to populate all core profile fields where they are currently blank. This currently includes:

• Personal Details
• Availability
• Social
• Contact
• Candidate Summary
• Recruiter Summary
• Education
• Certifications
• Employment History
• Skills

Where a field offers a dropdown selection, Firefish will attempt to match information on the CV to one of your dropdown values – if a match cannot be determined by the AI based on the information available, the dropdown will not be populated.

For example, if a candidate lists a Certification on their CV, and there isn’t an equivalent value available added in System Configuration -> Properties -> Certifications, Firefish will ignore it and not add it as a value.

How Accurate is the Parser?

The AI Parser relies on contextual information in a CV in order to structure and populate the data against the record. As such the accuracy and reliability of data capture relies on the quality of the CV – this includes the way the document is laid out, the information that can be extracted and the available relevant dropdown options in Firefish.

CV Parsing is a background process, so you should check back in on your record after adding a new CV document. The speed of the parser will vary from CV to CV based on size and complexity. Depending on the amount of data being processed, you may notice that the parser populates your candidate record in stages.

In the record header, you'll see document icons - the colour of these icons tell you what stage CV processing is at:

Orange = CV is awaiting processing
Blue = CV has processed but contains less than 200 words
Black = CV has processed successfully
Red = CV has failed processing

Managing Your Parsed Skills List in Properties

The Firefish AI CV Parser uses a curated Skills list in System Configuration > Properties > Candidates, with a maximum of 150 values. This list drives how skills are linked, ranked, and surfaced across candidate records.

Why it matters:
The AI will prioritise and match skills from your Properties list to ensure clean, consistent data for ranking, reporting, and searchability. However, the parser will also suggest additional skills it believes are worth highlighting - even if they’re not on your list. These suggested skills are searchable and rankable, but may vary in naming and won’t scale as reliably as linked ones from your controlled list.

What to Include

Focus your list on:

• Technical Skills – e.g. “Python”, “AutoCAD”, “Google Ads”
• Role-Specific Skills – e.g. “Pipeline Management”, “Care Planning”, “Bid Writing”
• Measurable Skills – Anything that can be objectively demonstrated or quantified in a CV

Avoid soft skills like “Communication” or “Self-starter” as these won’t add value to search or ranking.

Use the ‘Merge Into’ option in dropdowns to clean duplicates and standardise terminology:

Need inspiration? Use this ChatGPT prompt:

"Act as a recruitment consultant working in the [industry] sector. Your task is to produce a clean list of 100 technical, role-specific, and objectively measurable skills that typically appear on CVs across roles such as [Job Titles].

Only include skills that are:

• Clearly demonstrable on a CV (e.g. 'SQL Query Writing', not 'Good Communicator')
• Free from duplicates, overlaps, or soft skills
• Formatted as single-line dropdown entries (no descriptions)

Output the list as a clean bullet point list with no additional explanation or commentary."