Our AI and LLM Solutions
At DataSing, we help organisations leverage artificial intelligence to enhance innovation, efficiency, and decision-making. Our flagship KeaAI platform is specifically designed with governance and responsibility in mind – allowing our clients to scale the use of GenAI throughout their organisations with confidence. We also offer expertise in evolving AI capabilities such as prompt engineering, Retrieval Augmented Generation (RAG) and AI advisory services to support diverse business requirements.
Introducing
KeaAI
KeaAI by DataSing is a powerful AI solution designed for organisations that want to use generative AI on a large scale. It's built on the LLaMA Large Language Model (LLM) and offers all the usual GenAI features, but with added controls, specifically for GenAI security and governance. This means our clients can trust the results it provides and feel confident about how the AI is being used, while also ensuring data security and privacy standards are upheld.
Many organisations are starting to test AI solutions in Proof of Concept mode, but at DataSing, we’re focused on the ‘last mile’—making sure AI is ready for real-world use with automated controls.
KeaAI is designed for production GenAI deployment, making the transition from Proof of Concept mode seamless and safe. It can be installed on-site, reducing concerns about data protection and export. It also includes tools to monitor and track how GenAI is used across the company, ensuring compliance with data access rules.
An Overview of KeaAI
01
Governance User Interface
KeaAI features an easy-to-use governance interface that allows clients to manage GenAI outputs effectively. It enables both business and technical teams to monitor, track, and set alerts for GenAI usage at individual query level. This ensures full visibility and oversight of how the AI is being used in the organisation.
The interface also allows users to recreate GenAI outputs exactly as they were, helping to eliminate concerns about AI ‘hallucinations’ or inaccuracies. Additionally, it visually breaks down each GenAI output step by step, clearly showing how the system arrived at its answer.
03
Secure Flexible On-Prem Option
KeaAI can be installed on-site, giving our clients full control over their sensitive data. This setup meets strict security and legal requirements, which are especially important for the public sector.
It provides peace of mind for business and technical owners by combining advanced AI technology with strong data protection, making it perfect for government agencies, NGOs and other organisations with strict data governance policies.
.
02
KeaAI User Interface
KeaAI provides RAG (Retrieve-Augment-Generate) functionality, enabling users to enhance GenAI outputs with accurate, real-time information. By retrieving data from trusted sources, KeaAI ensures that AI-generated content is enriched with up-to-date facts and automatically includes citations for transparency. Users can specify which sources are referenced during the retrieval phase, aligning results with internal guidelines or industry standards.
This feature guarantees that AI outputs are factually grounded, with clear citations to back up the information, making KeaAI ideal for industries such as Government, NGOs, finance, healthcare, and legal services where accuracy and accountability are critical.
04
Fine-tuned to the New Zealand Context
KeaAI is fine-tuned to your organisation and to the specific legal, cultural, and social contexts of New Zealand. It incorporates local legislation, government policies, and cultural nuances, ensuring that the AI’s outputs are highly relevant and accurate for New Zealand-based applications.
This delivers insights and analyses that are perfectly aligned with the unique needs and challenges of New Zealand organisations, making it an indispensable tool for local government agencies, businesses, and NGOs.
Customisable Deployment Approach
​
DataSing offers the flexibility to train and deploy KeaAI two distinct ways: on-premises within our client's infrastructure or within Azure Cloud infrastructure. This flexibility ensures our clients can select the option that best aligns with its security, operational, and scalability requirements. DataSing work closely with our clients to determine the most suitable deployment option based on their specific needs and constraints. This includes assessing the sensitivity of the data, the existing infrastructure, budget considerations, and the desired scalability.
On-Premises Deployment
For heightened security and to ensure that all sensitive data remains within our client's domain, KeaAI can be fully trained and deployed on-premises at our client. This deployment option ensures that future documents and data shared with the model are contained entirely within our clients secure infrastructure. DataSing offer this through our ”as-a-Service” provision of on-premises AI server (Supermicro GPU SuperServer 420GP-TNR - 4U GPU).
•Security and Compliance: All data processing and model inference occurs within our client's physical premises, minimising the risk of data breaches and ensuring compliance with internal and governmental security standards.
•Data Sovereignty: Our client retains full control over its data, with no exposure to external networks, ensuring data sovereignty and compliance with local regulations.
•Cost Control: By deploying infrastructure on-premises costs can be better managed and controlled. Once our infrastructure is in place, costs associated AI are fixed, avoiding the ongoing operational expenditure (OPEX) associated with cloud service fees and AI GPU usage/token usage.
Cloud Deployment
Alternatively, KeaAIcan be deployed in Microsoft’s Azure secure cloud environment, providing our clients with the benefits of cloud scalability and flexibility while still maintaining stringent security controls.
•Scalability: Cloud deployment allows our clients to easily scale the AI model’s capabilities, handling large volumes of data and accommodating increased demand without the need for additional on-premises hardware.
•Accessibility: A cloud-based model can be accessed by authorised personnel from various locations, supporting remote work and collaboration.
•Disaster Recovery and Redundancy: Cloud infrastructure offers built-in redundancy and disaster recovery options, ensuring the AI model remains operational even in the event of hardware failures or other disruptions.
•Hybrid Integration: DataSing can configure a hybrid setup where sensitive data remains on-premises, while the AI model is hosted in the cloud, striking a balance between security and scalability.