How Doghouse uses AI
Doghouse Agency uses artificial intelligence in two distinct contexts. Within our Civio product suite, which is delivered to client organisations, and internally, to accelerate software development, design, and quality assurance workflows.
Our primary AI product is Civio Assist, a Retrieval-Augmented Generation (RAG) chatbot designed specifically for government and public sector organisations. Civio Assist is deployed on behalf of our clients to help citizens and staff access information quickly and accurately.
How Civio Assist works
Civio Assist uses a two-stage architecture designed to keep responses grounded in verified, client-approved content.
When a user asks a question, the system searches a dedicated knowledge base made up of approved information provided by the client organisation. For example, the content of a government agency's website or policy library.
The system then uses a foundational large language model (LLM) to generate a conversational response based only on the relevant information retrieved from the knowledge base. The LLM's general training data is not used to answer questions.
To generate these responses, Civio Assist uses foundational LLMs provided through secure, enterprise-grade platforms, primarily Amazon Web Services (AWS) Bedrock and Microsoft Azure AI Foundry. This gives us access to a range of models, including the Claude series (Anthropic) and the GPT series (OpenAI).
Classification of AI use
| Category | Detail |
|---|---|
| Usage Pattern | Generative AI. Natural Language Processing. |
| Domain | Service Delivery. Public Communication. Software Development. |
| Public Interaction | The public interacts directly with Civio Assist. Responses are generated without a human reviewing each individual response. Escalation pathways to human staff are always provided. |
| Internal Use | AI-assisted tools are used by Doghouse staff to accelerate code generation, design review, and documentation. Human review is applied to all AI-assisted output before delivery. |
Knowledge base governance
The accuracy and currency of information within a Civio Assist knowledge base is the responsibility of the client organisation. Doghouse provides the platform and the tooling to manage the knowledge base, but the client organisation owns and governs the content the chatbot uses to form its answers.
We work with each client organisation during onboarding to establish a content governance process, including review cycles, content ownership, and procedures for updating or removing information. We strongly recommend that client organisations designate a named content owner responsible for the knowledge base.
Data handling and privacy
Our data handling practices are designed to enable our client organisations to comply with their obligations under the Australian Privacy Act 1988 and the Australian Privacy Principles (APPs).
What we process
- User prompts and chatbot responses, the content of each conversation.
- Context retrieved from the client organisation's knowledge base.
- Anonymous interaction data (e.g., message count, session duration) used to monitor system performance.
How we handle the data
- Conversation data is not shared with or made available to other customers.
- Conversation data is not used to train or improve any foundational large language model.
- Conversation data is not stored by third-party LLM providers. It is passed to them only to generate a response, then discarded.
- All data processed and stored by Civio Assist is held securely within Australia.
- The platform includes automated data sanitisation processes designed to detect and remove personal information, such as names, phone numbers and email addresses, from conversation logs before they are stored.
Monitoring, oversight and accountability
Civio Assist is a tool for information retrieval, not a decision-making entity. It is designed to augment, not replace, human support. No AI output from Civio Assist should be used as the sole basis for a legal, financial, or official decision.
Doghouse continuously monitors the performance and effectiveness of the Civio Assist platform, including metrics related to response accuracy, user satisfaction, response latency, and system uptime. We publish a service status page and provide client organisations with access to interaction analytics.
Client organisations determine their own processes for human review of chatbot interactions and for providing users with pathways to contact a human staff member. Doghouse requires all Civio Assist deployments to include a clearly visible escalation path, a link or button that connects the user to a human contact, within the chatbot interface.
Fairness and bias
We are committed to providing a platform that enables our client organisations to deliver fair and inclusive services. We use a multi-layered approach to mitigate the risk of bias in AI-generated responses.
The RAG architecture is our primary method for ensuring fairness. By grounding responses in the client organisation's own approved content, we reduce the influence of the LLM's general training data, which may reflect biases present in its source material.
The platform includes guardrails designed to detect and block hostile or inappropriate language and to prevent the processing of sensitive personal information.
Before deployment, we conduct rigorous internal testing. We also work with client organisations to perform their own quality assurance testing to identify and correct issues specific to their knowledge base and user base.
Users can flag inaccurate or biased responses directly within the chatbot interface. This feedback is reviewed by the client organisation and by Doghouse as part of our continuous improvement process.
Bias can still exist in underlying data or language models. We encourage users to report any instances of biased or unfair responses. We take all such reports seriously.
Limitations of this technology
Civio Assist is designed to provide general information. It is not a substitute for professional advice. The AI models powering the platform are probabilistic. They may sometimes produce responses that are inaccurate, incomplete, or unreliable, even when grounded in a well-maintained knowledge base.
For this reason, information provided by Civio Assist must not be used as the sole basis for making legal, financial, medical, or any other official decisions. For any complex, sensitive, or personal enquiry, or for any matter of importance, users must contact the client organisation directly through their established channels to speak with a qualified person.
Doghouse does not warrant that Civio Assist will always be available, accurate, or free from error. We are committed to continuous improvement and will act promptly to address any reported failures.
Internal use of AI at Doghouse
In addition to our AI products, Doghouse staff use AI-assisted tools to support software development, design, documentation, and quality assurance. This includes tools such as GitHub Copilot, Claude, and GPT-4 for code generation, code review, and technical writing.
Our internal AI use policy requires that all AI-generated code and content is reviewed by a qualified human before delivery, no client data or confidential information is submitted to public AI services, and AI tools are used to augment, not replace, the professional judgement of our team.
We do not use AI to make employment decisions, performance assessments, or any other decisions that materially affect individuals.
Questions about AI?
If you have any questions about this statement or how Doghouse uses AI, contact us at hello@doghouse.agency.
If your question relates to the specific information provided by a Civio Assist chatbot on a client organisation's website, please contact that organisation directly.
Questions about our AI?
Contact our team for any questions about how we use AI or how Civio Assist works.