TABLE OF CONTENTS
- What is Copilot?
- What problems does Copilot solve, and how?
- How about GDPR, EU AI Act, and security?
- Any Questions?
What is Copilot?
Talentech's Copilot combines the power of AI with Talentech's unique data to help you increase recruitment efficiency and quality.
Copilot is designed to save you time, allowing you to focus on what truly matters and make better decisions. It benefits from the power of AI in recruitment and HR processes while remaining secure and GDPR compliant.
Copilot’s purpose is to assist and inspire, not to make decisions for you—you remain the pilot in charge.
What problems does Copilot solve, and how?
Writing compelling job ads
The purpose of our AI-powered assistant is to help you get started writing, instead of starting with a blank page, and to optimize the performance of your ads. We know that you value speed and execution, thats why we want to streamline the process of turning a role idea into a publish-ready advertisement within a couple of minutes. You can get inspired and iterate with different settings to suit your needs.
The position name and any specifics added to this recruitment, like competencies or requirements, are used to create the draft and it's inspired by your organization's last 10 ads to make it more unique. But to help you create an ad optimized for relevant applicants, we've also given our AI instructions to make it gender-neutral, unbiased, and of adequate length and phrasing.
What affects what?
Position name/title and job requirements (if anything added in the ATS) affects the role specific part of the suggested drafts.
Previous ads affect the non-role specific parts of the suggested drafts, such as organization description and perks.
Note: If you encounter incorrect content when generating ads—such as details about a specific part of the organization that isn't related to the part receiving the draft—please contact Talentech. They can assist in overriding the organization summary with a more generic version.
Creating high-performing Social media campaigns
Copilot suggests suitable text optimized for social media to attract more relevant candidates. Suggestions are based on the job title and category, respecting platform constraints and tone choices without drifting into generic marketing language.
Using this feature requires that Social Media Ads is activated in your recruitment system. Read more about Social Media Ads here.
Providing better referral suggestions through LinkedIn
By offering Boolean search string suggestions in our Referral module, Copilot helps employees get better first-hand contact recommendations on LinkedIn for specific jobs – which drastically can increase the number of relevant referrals. Suggestions are based on the job title and ad text.
Using this feature requires that Talentech’s Referrals module is activated in your platform. Read more about our Referral module here.
Conducting competency-based job interviews
Copilot provides competency-based questions suitable for specific positions and industry, helping you prepare for interviews. The questions are based on the job ad and any specifics added to the specific recruitment, like competencies or requirements. This helps in adopting a competency-based approach, as situational questions that evaluate past behavior are good indicators of future performance. For an optimal and fair process, use the same questions for all candidates.
Choosing the best-performing publishing channels
Copilot helps you decide where to focus your publishing efforts by providing top suggestions for your market and job category, and offering insights into the data behind these suggestions. Data is based on Talentech’s publishing data, updated quarterly with aggregated data from the last 12 months.
The channels displayed are the ones where Talentech has enough data for the chosen country and job category to provide reliable numbers.
We recommend to focus on passive candidates scores and hire rates for hard-to-fill positions, or check out ad views and applications in recruitments with more active job seekers.
To ensure the best possible comparability, we account for data gaps resulting from candidates not consenting to be tracked. Adjustments are made for variations across different ATS platforms and cookie consent tools, providing data as close to reality as possible within the given constraints. Also, user actions and candidate behaviors can skew data. For example, recruiters not setting the candidate status to “hired” will affect the hire rate, and candidates applying from a different device will affect conversion numbers.
When analyzing this data, also keep in mind how different channels structure their user experience, as it affects metrics. Some require a click to view the full job ad, while others display it upfront, leading to different engagement counts (for Ad views).
Categories
To help you get relevant data you can filter on job categories. The standard used in this feature is The International Classification of Occupations (ISCO-08). Click here to read more and find out which category your job belongs to.
Copilot chat in People Core (WIP)
Our AI-powered assistant helps employees and managers to quickly find answers to questions about policies and other topics from your shared documents, all while ensuring People Core’s built-in access control. By providing automated responses in the user’s natural language, we aim to reduce the workload for HR and legal teams. We’ll keep adding chat capabilities to support HR and recruiters even more.
Application screening
Copilot screen candidate's application data and match them with the job ad and requirements, providing a score and an analysis as to if and why the specific candidate could be a good fit.
This is the result of a text-to-text match (1–10) based on the context between the job's ad, and requirements and application documents and answers to screening questions (but not tests). The instructions also included ignoring irrelevant personal details to reduce bias.
Think of the scale as a quick signal of application data alignment based on the job context:
- 9–10: Full alignment with stated requirements and responsibilities.
- 7–8: Strong fit with minor gaps.
- 5–6: Partial fit; some gaps or limited evidence.
- 3–4: Limited alignment; key requirements missing.
- 1–2: Weak alignment; little or no relevant evidence.
Evidence of must-have requirements is heavily weighted, along with demonstrated experience, certifications, skills, achievements, and relevant education. Language proficiency matters only if the job explicitly requires it. Administrative compliance such as background checks are ignored in the matching of candidates, links to external sites such as articles and Linkedin are not read. Pure keyword lists without supporting evidence are down-weighted to reduce "keyword-stuffing".
If the applicant makes updates or changes are made to the job (job ad or job requirements), the match scores and analysis will not update instantly but will be refreshed during nightly processing.
Remember that an AI text-to-text match won't always be right and that randomness is a significant element in generative AI models. Make sure to evaluate the complete application before making decisions.
What it takes for applications to be analyzed
For Copilot to perform a fair evaluation, information about experience and skills is essential. Any unreadable file formats (like image-based CVs) or information from links (like LinkedIn profiles) won’t be available to Copilot. Such applications won’t be analyzed unless the screening questions contain enough information about experience and skills. Applications older than 3 months or containing unethical or harmful content will not be analyzed.
Additionally, jobs lacking sufficient information - such as ad text or job requirements, cannot be used to analyze applications.
Best practices to improve quality
For recruiters, clarity upfront is key. We recommend being specific in the job requirement section in the ATS. Write must-haves and nice-to-haves explicitly, and ask evidence-seeking screening questions such as specific years, tools, and certifications. Avoid generic for must-have requirements such as "doer" without context. To get out the most value of the application screening, try to avoid vague and ambiguous requirements.
Examples of vague requirements decreasing the quality of the screening:
- Several years of experience as a leader
- Must be a team-player and a doer
- Advanced computer skills with strong database knowledge
- Excellent problem-solving skills and ability to build relationships
- Any certifications is considered as a plus
- Bachelor's degree or a Master's degree in IT
Example of clear requirements enhancing the quality of the screening:
Must-haves:
- Leadership (Tech Consulting): 8+ years leading cross-functional delivery teams (engineering, data, UX), direct management of atleast a team of 6–15 consultants, owns portfolio P&L
- Team-player & doer: manage all project coordination, runs discovery workshops, drafts SOWs, does architecture/solution design, reviews data models. Leads stand-ups/retros; produces tangible artifacts such as roadmaps
- Bachelor's degree or a Master's degree in Computer Science, Data Analytics, or equivalent practical experience
Nice-to-haves
Certifications: Azure/AWS (SA-Assoc/Pro), Azure Data Engineer, PMP/PSM, TOGAF/ITIL; familiarity with SOC 2 / ISO 27001.
Commercial impact: Proven proposal ownership, average deal ≥ €250k, and expansion/renewal growth.
Questions about GDPR, EU AI Act or security - please read "How about GDPR, EU AI Act, and security?" or reach out!
Administrators can decide who can view the Match column and in which projects it appears. Per default from start, only administrators will have access.
Candidate finder (WIP)
Our AI helps you rediscover strong candidates who have previously applied to your company, and see who best matches your current job requirements. Candidate finder will match candidates in your talent pool with the job ad and requirements, providing you with the most relevant candidates. Maybe the runner-ups in your last recruitment is a perfect match for this new role you’re hiring. Candidate finder is designed to help you rediscover runner-ups and qualified candidates you already know, so you can move faster with people who are highly likely to fit your company needs.
How Candidate finder works
We match applicants from your organization's previous recruitments against the job's ad and requirements, the highest matching scores at the top. Check out their profiles, either add them to your recruitment or reach out to get them to apply themselves.
- You'll get up to 20 of the strongest matches from your existing candidate pool, shown in order from highest to lowest match
- For each match you'll get a clear overview and analysis of their strengths and why they potentially will fit the role
- When you add a new candidate to your candidate pipeline or update the job requirements, the list will refresh to reflect those changes
- The results will refresh automatically after a week, especially if new candidates have applied during that time
We've also instructed our AI to ignore irrelevant information like gender, age, and other personal details to help you reduce bias during recruitment. Remember that these candidates might have emphasized different skills and qualifications if they had applied to this specific job and that an AI text-to-text match won't always be right. Use the suggestions as potential candidates and not decisions, as candidates may have written their previous applications for different roles.
Questions about GDPR, EU AI Act or security - please read "How about GDPR, EU AI Act, and security?" or reach out!
Note: You can choose between two configurations: one with an add-button, allowing you to add candidates directly to your recruitment process, or one without.
How about GDPR, EU AI Act, and security?
Security and GDPR compliance are foundational principles of our Copilot. We take responsibility for offering secure and compliant solutions to our customers. This means ensuring that data is never shared externally and is protected within our infrastructure, maintaining GDPR compliance and security.
Built within our Microsoft Azure infrastructure, hosted within the EU, Copilot ensures data is encrypted and secured according to our ISO27001-based Information Security Standard. We utilize Microsoft Azure OpenAI LLM models for text generation, combined with data from Talentech’s infrastructure. None of your personal data will be directly used to train any models. However, the output generated based on your input may contribute to ongoing improvements in the system.
We comply with EU AI Act, the EU Commission Guidelines on prohibited artificial intelligence practice, and local legislation and guidelines for AI use in public and private sector, such as from DIGG and Datatilsynet.
Our Copilot is also built in compliance with the EU AI Act. It never replaces human decisions as we highly value human-in-the-loop; instead, it serves as a decision-support tool with full transparency for both recruiters and candidates. We have developed our Copilot entirely within our own infrastructure, without sharing data with external parties, ensuring neutrality and clearly defined boundaries for the AI.
To ensure transparency towards candidates, we recommend reviewing and updating your privacy policy to reflect the use of AI in recruitment. Here's our recommended wording, which you can find in Nordic languages here:
Use of artificial intelligence (AI) in the recruitment process
To ensure a more efficient and fair evaluation of applications, we use artificial intelligence (AI) as support in our recruitment process. This means that your application documents may be analyzed by an AI-based solution provided by our system supplier, Talentech.
The AI solution is used to identify and structure information in application documents (e.g. education, experience, and skills), and to assess how well the application matches the requirements in the job posting. AI is used solely as a support tool – all final evaluations and decisions are made by humans.
If you have any questions, or if you wish to opt out of AI-based analysis, you can contact us at [contact info].
Be aware that our Copilot can make mistakes and must be used in conjunction with something AI can never replace: human judgment.
If you have any questions about GDPR or security, please feel free to reach out!
What does the Copilot cost?
Prices vary by market and depend on your package, so fill out the form of the bottom of this page for more information.
Any Questions?
Any questions? Please reach out to us on through this page or to our support: https://talentech-support.freshdesk.com/en/support/tickets/new