Client Context
An AI life assistant built on AI Brain.
A leading global technology holding company partnered with 60x to build a production-grade AI personal assistant for busy urban professionals. The product needed to feel genuinely proactive, not like another reactive chat interface wrapped in a polished UI.
The client's internal AI team operates as an innovation function across a global portfolio spanning food delivery, classifieds, fintech, and edtech in more than 90 markets. Their brief was to prove what an AI-native product could look like when deep user context and real-world utility were designed in from the start.
The problem was not generating responses. It was giving the assistant the kind of structured memory required to truly know a user, act on their behalf, and improve over time.
The Challenge
A personal assistant is only useful if it has real memory.
The client wanted to build an assistant that could anticipate needs across food, travel, shopping, and planning. That meant surfacing the right information at the right time and increasingly taking action on the user's behalf.
To do that, the product had to ingest personal data from email and calendar, synthesise it into durable context, and make that context instantly accessible to agents operating in real time. Without that memory layer, the assistant would remain shallow, forgetful, and reactive.
Context had to persist
The assistant needed to retain and update meaningful user knowledge over time rather than rely on short-lived chat history.
Signals came from multiple domains
Calendar events, email content, user preferences, and interaction history all needed to be combined into one coherent view of the user.
Agents needed low-latency access
A consumer product cannot afford slow, noisy retrieval across many separate data sources every time it needs to answer or act.
The system had to be production-ready fast
The engagement was scoped to move from kickoff to full handover in 3 months, with a live internal rollout immediately after.
The Platform
AI Brain became the assistant's intelligence and memory layer.
60x deployed AI Brain, its proprietary knowledge graph platform, as the core memory and retrieval system behind the product. Each user received a personal AI Brain instance: a continuously updated knowledge graph built from their email, calendar, and interaction history.
Ingest
New personal signals flow into the graph as they arrive, giving the assistant a live view of the user's plans, relationships, preferences, and priorities.
Structure
Information is processed through entity extraction, relationship mapping, and semantic enrichment so the system stores context, not just raw records.
Retrieve
Agents access the user's graph through a small set of retrieval tools, avoiding the latency and complexity of making dozens or hundreds of tool calls across siloed systems.
That architecture let the assistant do more than recall facts. It could connect signals across life domains, infer relevance, and decide what to surface at the moment it mattered.
What We Built
A proactive assistant powered by personal knowledge graphs.
60x designed and delivered the full assistant application, using AI Brain as the context layer that made proactive behaviour possible. The result was not simply a conversational interface, but a system that could use live personal context to suggest, brief, and act with relevance.
What AI Brain enabled
The assistant could suggest a restaurant based on a calendar event and known food preferences, brief a user on a contact before a meeting, or generate a personalised daily audio update from the live state of their graph.
Why it mattered
A siloed retrieval model would have produced slower, shallower behaviour. AI Brain collapsed the complexity into a single structured graph per user, enabling deep retrieval with a small and reliable tool surface.
The breakthrough was not that the assistant could answer questions. It was that it could build and use a durable model of the user's life strongly enough to become genuinely useful.
Novel Application
Enterprise knowledge graph architecture applied to a single person.
AI Brain was originally built as an enterprise platform for unifying organisational data across systems such as CRMs, document stores, communication tools, and internal databases. This engagement proved that the same architecture can be translated into a consumer setting, where the "organisation" is one person and the relevant data is their personal digital life.
The same techniques that help a firm retrieve context across thousands of documents and expert relationships also work for recommending a restaurant, surfacing a buried travel preference, or structuring a user's day around what lies ahead.
The client was not buying a novelty chatbot. They were using enterprise-grade context infrastructure to build a consumer AI product that could actually know its users.
The Results
From kickoff to full handover in three months.
Kickoff to full handover
Employees in initial rollout
User scale built into delivered platform
- Delivery: 60x built and handed over a production-grade assistant application within a single quarter.
- Initial rollout: more than 50 employees were onboarded, each with their own live AI Brain instance.
- Ownership: the client received the full codebase, infrastructure, and associated documentation.
- Delivered scope: the product included deeper commerce and food-ordering integrations, with the platform built to support 5,000 users.
The engagement showed that AI Brain is not only an enterprise platform. It can also serve as the intelligence layer that makes a consumer AI product capable of deep user understanding, reliable action, and ongoing improvement.
Why It Matters
The memory layer is what separates AI products from AI demos.
Most assistants fail because they never build a durable model of the user. They can generate fluent responses, but they do not accumulate the kind of structured context required to become more useful over time.
By using AI Brain as the assistant's memory architecture, the client created a product with a much stronger foundation for proactive behaviour, personalisation, and long-term extensibility across multiple life domains.
That is the broader implication of the work: every serious AI product will need an intelligence layer capable of organising knowledge, not just producing output.