DataSapien’s three-tier approach to developing personal AI

Personal AI and Personal AI Agents are suddenly all the rage, and the promise of adding AI to existing personalisation technology is alluring. However, Generative AI is still very early in its development. Delivering trustable personal AI is mission-critical. Therefore, rolling it out to customer touch points at scale too quickly is already revealing risks for organisations, as well as headline-grabbing harms to customers.  

With current technology, brands planning to offer personal Generative AI agents to customers will be on the hook for unknown and potentially unlimited liabilities.  

Delivering trustable Personal AI

Our mission is to deliver what we call ‘Private Personalisation‘. It is deployed on the customer’s own Smartphone and enables them to enjoy hyper-personalised services with complete control over data that is shared. This will evolve into a private and confidential personal shopper, concierge, guide, mentor or agent.

Therefore, at DataSapien, we are taking a considered and measured approach to building personal intelligence capabilities.  The first step is (of course) getting the data right: empowering customers to see, control and participate with their own data.  
Next, we take a multi-track approach to building intelligence. Firstly, we deliver simple, safe personal intelligence using proven models and powered by this on-device data.  Secondly, we simultaneously develop and test more advanced personal generative AI agents, rolling them out as they mature and become customer-ready. 

Embedding In Existing Apps (The Personal AI SDK)

The essence of DataSapien’s innovation lies in its unique Software Development Toolkit (SDK). This is a bundle of software elements, ingeniously combined, to embed personal intelligence directly into existing customer apps. This intelligence is meticulously structured into three progressive levels: Standard, Advanced, and then the avant-garde Next GENeration generative Intelligence.

Let’s delve into the intricacies of these levels and the transformative potential they hold.   

The Foundation: ‘Pure’ Zero-Party Customer Data 

The essential starting element when creating any personal intelligence is the ingredients: clean, complete, contextual, consented data with integrity. The only way to ensure that personal data is ‘pure’ is to give individuals personal control and enable their participation. Our architecture utilises on-device, human-centric, personal data stores, vaults and wallets and we have over a decade of experience building these. Clean, transparent, structured, consented ‘Zero-Party Data’ is the foundation of the DataSapien personal AI pyramid. 

Standard Intelligence: Tier One  

At the base of DataSapien’s Personal in-app Intelligence pyramid is Standard level intelligence. This foundational layer is designed to enhance user experience through basic, yet essential, personalisation features. It applies straightforward deterministic rules and logic locally, to include capabilities such as personal insights, basic recommendation algorithms, and simple automation tasks.  

These features work together seamlessly to streamline user interactions, making digital interactions more intuitive and responsive to individual user needs. This relatively simple, safe and established ‘Standard’ Intelligence layer is where personalisation begins, setting the stage for more advanced functionalities.  

Advanced Intelligence: Elevating Experiences 

Building on the solid foundation laid by ‘pure’ data and the Standard Intelligence level, Advanced Intelligence introduces more sophisticated data processing and analysis capabilities. This level harnesses the power of machine learning algorithms to offer customers deeper insights into their preferences, behaviour patterns, and potential needs.  

Advanced Intelligence enables features such as personalised content curation, product and service recommendations and more nuanced automation of tasks. It’s designed to not just react to human inputs but to anticipate and predict needs, thereby crafting a more tailored and engaging user experience. This level of intelligence ensures that every interaction with the app feels personal and relevant, significantly enhancing user satisfaction and engagement. 

Next Generation Intelligence: The Future Unveiled 

At the pinnacle of DataSapien’s vision for in-app intelligence is the Next Generation Intelligence. This revolutionary layer is currently under development in the DataSapien R&D Lab with the integration of open source GPT-based Large Language Models (LLMs) directly into customer apps. The ambition here is to create a personal agent (shopper, guide, or concierge) that resides within the app, offering unprecedented levels of contextually personalised interactivity. Imagine having a virtual assistant that not only understands your needs and preferences but also anticipates them, offering suggestions, solutions, and content that feel remarkably personal and timely. 

DataSapien Next Generation Intelligence aims to overcome the limitations of current LLMs by developing models that are compact enough to fit on a device, yet powerful enough to deliver bias-free, hallucination-free, and highly personalised interactions.  

Critically, personal LLMs require trustable guardrails. We’re experimenting with tools such as RAG (Retrieval-Augmented Generation) to enhance the accuracy and reliability of generative AI models with facts cross checked against multiple internal and external sources. This has potential to dramatically reduce hallucinations and biases in LLMs, enabling them to operate safely. 

This level of intelligence could revolutionise customer-brand interactions, making every digital platform feel like it has a deep, intuitive understanding of its user – while operating in a completely private and confidential way. The potential applications are vast, from personalised shopping experiences and interactive guides to sophisticated personal concierges that manage every aspect of digital interaction with a brand. Eventually, we imagine future versions of these models will become suitable for mental and physical health applications – providing safe, personalised patient centric care pathways. 

The Road Ahead 

The journey to fully realise the potential of Next Generation Intelligence is paved with challenges, primarily around making these advanced models compact, efficient, ethical and safe. DataSapien’s commitment to research and development is a testament to our dedication to overcoming these obstacles and delivering a future where digital interactions are as personal, intuitive and trusted as human ones. 

As we stand on the brink of this new era, it’s clear that DataSapien’s three-tiered approach to in-app intelligence is not just an evolution but a revolution in how we think about, and interact with, technology. The Standard and Advanced levels provide a robust foundation, while the Next Generation Intelligence promises to unlock a future where technology understands and anticipates our needs in ways we’ve only begun to imagine. 

By embedding personal data and intelligence directly into customer apps, DataSapien is not just changing how we interact with technology; the technology will redefine the very nature of those interactions. The future of digital interaction is hyper-personal, and that future is being built by DataSapien. 

DataSapien provides a suite of Private Personalisation capabilities for customer applications – packaged as a productised and embeddable software bundle (SDK), combined with an enterprise server-side data orchestration environment for easy and comprehensive integration with Customer Data Environments (CDP, CRM, LLM, etc). For more information check out our developer documentation site.