AI that makes your recruiters betterpeoplepeople

Like the rest of the JobAdder platform, Adder Intelligence is built on the foundations of GDPR. You are the data controller, you hold the direct relationship with your clients and candidates, and you own the client and candidate records you store in JobAdder. JobAdder is the data processor, which means we act on your instructions when processing the candidate and client information in your account.

For more information, you can download JobAdder's security FAQs or visit our Trust Centre.

We use a selection of LLMs hosted by AWS Bedrock. We select an appropriate mix of models to deliver the best experience and accuracy. The chosen model might change over time depending on performance, but they are all reviewed by our security, and engineering teams for safety, security and compliance before usage. The LLM providers we use do not retain your inputs and outputs, or use them to improve their services and your data does not leave the AWS environment.

No, your data is not used to train the models.

Data stays within the AWS environment. For customers in AU, it may travel to AWS Singapore for processing before returning to the AU region to be stored at rest. For UK and EU customers, data travels within the EU region (London) for processing before returning to our production environment in Ireland for storage at rest.

In developing Adder Intelligence, JobAdder has applied the principles set out in the EU's Artificial Intelligence Act (Regulation (EU) 2024/1689) and the Australian Government's 2024 proposals paper, "Mandatory Guardrails for AI in High Risk Settings". We mitigate both explicit bias, by preventing the algorithm from using legally protected attributes, and implicit bias, by testing whether the algorithm introduces bias in its outputs despite our instructions. Once released, AI features are monitored and reviewed on an ongoing basis. Because AI can't be fully predictable, no system can guarantee a perfect result in every case, so it's important to keep humans in the loop. You remain responsible for reviewing outputs and ensuring a diverse candidate pool for each role, and a person must always make the final decision in shortlisting and selection. Adder Intelligence is designed to support the people making those decisions.

Our AI services are built on Amazon Bedrock, which provides strong security and governance, and none of your client or candidate data is used to train the underlying foundation models. Amazon Bedrock Guardrails automatically detect and mask sensitive personally identifiable information (PII) in candidate profiles, keeping your data private and helping you meet compliance obligations such as GDPR. Bedrock was the first platform from a major cloud provider to achieve ISO/IEC 42001 certification, an independent audit of responsible AI management and security. We also use its customisable Guardrails to filter harmful or biased language and block topics outside recruitment, such as legal or financial advice. This reduces AI "hallucinations" by keeping answers grounded in the source data, like a candidate's resume or job description, and helps our services align with new regulations, including the EU AI Act.

Yes. JobAdder has an internal risk management framework for developing AI features, which defines high, medium and low risk use cases with examples. Every AI feature, whatever its risk level, goes through legal and security review and approval, and must meet foundational guardrails such as documented prompts, a data flow diagram, clear labelling of AI functionality in the product and human review. High-risk use cases are also tested for bias and discriminatory outcomes before release, and audited and reviewed once live. Keeping a human in the loop for high-risk functionality is a core part of our AI strategy. JobAdder's AI features are designed to assist recruiters and make them more efficient, while their expertise and judgement remain central.

Yes. JobAdder maintains a third-party risk assessment process that sets the minimum security controls and requirements for AI vendors who handle our systems or information. It evaluates each vendor's security controls, data handling practices and privacy safeguards, drawing on third-party assessments such as penetration tests and SOC 2 reports. For AI infrastructure and model providers, our engineering and security teams also assess data isolation and privacy controls, compliance certifications, model security and performance, contractual data protection guarantees, and alignment with a shared security responsibility model. This evaluation led us to select Amazon Bedrock, a fully managed, serverless platform for building and scaling generative AI applications. Amazon Bedrock is the only vendor that powers JobAdder's AI features, so your data is not shared with model providers.

Amazon Bedrock does not store inputs or outputs, and applies a zero-day data retention policy to the data it processes. Once an input is processed, the output returns to your JobAdder account, where it's stored within AWS under the same retention policies we apply to the rest of your account data.