Konkrd Agentic AI Policy & Guidelines
1. Purpose and Scope
- Statement of Intent
Our Agentic AI’s core objective is to serve each user’s needs in a fair, needs-based manner. We are committed to providing unbiased advice, free from hidden commercial incentives. - Coverage
This policy applies to all AI agents (including back end frameworks and front-end interfaces) used by Konkrd to help customers with:- Product advice (eg insurance, loans, or other financial products)
- Reviewing and managing a user’s existing products and policy documents
- Storing these documents in the user’s Konkrd App digital wallet/locker, whether they were originally purchased via Konkrd or uploaded by the user.
2. Fiduciary Principles
- Best Interests First
We treat our AI agents as if they’re fiduciaries. Their prime duty is to act in the best interests of each user, free from commercial bias or conflicting interests. - Obedience and Loyalty
AI agents follow user instructions to the fullest extent possible and maintain strict confidentiality of user data.- User Conduct: If users attempt to trick or misuse the AI (e.g. instructing it to perform fraudulent or harmful actions), the agent will deny such requests and may refer them to Konkrd’s compliance team for review.
- Conflict of Interest Disclosures
Konkrd maintains an Approved Product List (“APL”) of offerings for which we may receive a commission or fee upon sale or referral. However, these commercial arrangements do not factor into the AI agent’s product recommendations. Any referral fees or other remuneration is kept separate from the AI’s decision making process.
3. Data Privacy and Localisation
- Data Retention
We store a user’s sensitive data only for as long as needed to deliver advice or manage their products. While a user remains with Konkrd, we will keep this data securely. Users can request deletion of their data at any time in accordance with Konkrd’s Privacy Policy. - Minimising Exposure
We use a combination of on-device and/or encrypted cloud-based computing to reduce the risk of data breaches. All transfers and storage are protected using industry-standard encryption. - Access Controls
User data is only accessible in-browser during an active session. Otherwise, it is locked away in a secure CRM. Only the user or authorised Konkrd personnel (for legitimate purposes) can access the data once the user is logged in.
4. Security and Fraud Prevention
- Security Measures
Konkrd regularly conducts penetration testing and vulnerability assessments. We also encrypt data at rest and in transit to safeguard against unauthorised access. - Monitoring and Alerting
We maintain real-time security monitoring and alerts. These help us detect suspicious or unauthorised behaviour quickly, reducing the likelihood of fraud or data leaks. - Insurance
Konkrd continues to explore relevant coverage options to protect users from financial or other harm that may arise from AI-related errors or security breaches.
5. Autonomy and User Oversight
- Purchase Steps
Our AI agents facilitate needs-based analysis and product comparison. Any purchase, e.g. an insurance policy, takes place only after the user reviews and confirms the recommendation. There are no ongoing or repeated “transactions” once the purchase step has been finalised. - Transparency
We keep a clear record of the AI agent’s activity, including conversation history, which is available to the user in the app. Users can review these logs, pause any process, or request clarification. - Local Decision-Making
Where feasible, we strive to minimise third-party tampering by using trusted secure frameworks and technology services. We continue to investigate opportunities to process data locally or more securely.
6. Transparency in Recommendations
- Commercial Independence
While Konkrd may earn fees or commissions for certain products, the AI agent’s recommendations are not influenced by these arrangements. The agent’s logic is driven purely by the user’s stated needs and criteria. - Explainability
The agent’s recommendation process is clearly explained in plain language. We highlight key policy features and present a combined recommendation score that aligns with each user’s stated needs and takes into account expected benefits and costs. - Sponsored Content Labelling
We label all sponsored or third-party content as such. This ensures that users know when a product or message is part of a promotional arrangement.
7. Preventing Misinformation
- Source Verification
Konkrd’s AI includes automated checks and processes for validating the data used in product recommendations. We also maintain a curated knowledge base and require cross-referencing of critical information. - Model and Content Audits
We regularly audit and retrain our AI models on current and reliable data. This helps reduce errors or “hallucinations” in the AI’s output. - User Education
Users are reminded (eg at sign-up and in-app disclaimers) that AI, although powerful, can occasionally produce inaccurate or incomplete information. We encourage users to seek human support or second opinions if needed.
8. Accountability and Enforcement
- Internal Governance
Konkrd maintains an internal oversight committee (including an AI Ethics Lead) that reviews compliance with this policy. The team also investigates significant user complaints or policy breaches. - External Regulation
We track and comply with relevant AI-related laws and regulations, including any fiduciary duties. We also adhere to general consumer protection and financial services requirements. - Complaint Handling
Konkrd provides a simple online channel for users to lodge complaints. We aim to acknowledge issues quickly and resolve them within a specified timeframe. A designated “Complaints Officer” coordinates resolution efforts.
9. Ongoing Improvement
- Continuous Learning
We use a dedicated “Compliance Agent” to review chat transcripts and human-to-human interactions. Performance data and compliance scores are then assessed by the Compliance Officer, Responsible Manager, COO and CAIO to continually refine and improve our AI.
10. Liability and Indemnity
10.1 Limitation of Liability
To the extent permitted by law, Konkrd, its directors, employees, or agents will not be liable for any indirect, special, consequential, or incidental damages resulting from your use of the Platform or reliance on any information provided.
10.2 Your Responsibility
You agree to indemnify and hold harmless Konkrd and its affiliates from any claim or demand arising out of your breach of these Terms, misuse of the Platform, or violation of any rights of a third party.
10.3 Consumer Guarantees
We do not exclude liability for any liability that cannot be lawfully excluded, such as under the Australian Consumer Law.
Konkrd’s Agentic AI aims to streamline product recommendations, document management, and ongoing support for users. We are committed to transparency, security, and fairness in everything we do. These Guidelines ensure our technology remains aligned with each user’s best interests.