Flexibles AI provides a critical piece of infrastructure for the agentic ecosystem: persistent, structured memory. For AI agents to move beyond simple chat and into complex, long-running task execution, they require a way to maintain context that is both human-readable and machine-portable. Flexibles solves this through its personal knowledge graph, which serves as a long-term memory store that can be accessed by various agents regardless of the underlying LLM.

The platform is particularly relevant to the "agentic workflow" trend. By integrating multi-modal inputs like WhatsApp and screenshots into a structured graph, Flexibles enables ambient agent interactions where the AI can "learn" from a user's daily life without manual data entry. Its move into specialized domains like IoT and fleet management suggests a path forward for verticalized agents that possess deep domain knowledge coupled with a persistent memory of the user's specific business environment.

The portable context layer

The primary friction in the current AI market is the memory silo. When a user spends months refining a project in ChatGPT, that context is largely inaccessible to Claude or Gemini. Flexibles AI is built to address this fragmentation by decoupling the memory of an AI interaction from the model that generates the response. Based in London and developed by the product studio IIkaya Solutions, the company provides a model-agnostic platform that treats LLMs as interchangeable commodities while treating user context as the core asset.

At the center of their offering is the Personal Knowledge Graph. Unlike standard chat history, which is a chronological log, the Flexibles system extracts entities, facts, and relationships from conversations in real-time. This graph grows as the user interacts with the system, creating a structured map of preferences, projects, and past decisions. Because the system is model-agnostic, users can switch between different providers without losing the thread of their work. The platform supports exports in JSON and Markdown, and specifically highlights support for the llms.txt standard, which allows users to feed their personal context back into other AI tools via a simple URL.

Beyond the chat interface

Flexibles is not just a wrapper for existing models; it is a workflow orchestrator. The product integrates utility features that standard chatbots often leave to third-party plugins. This includes direct WhatsApp capture, allowing users to log expenses or set reminders through a messaging interface they already use. It also features a financial tracking system and task management tool, both of which are populated through natural language conversation. For example, telling the AI that you spent fifty pounds on a client lunch automatically logs the expense and updates the relevant project node in the knowledge graph.

This workflow-first approach extends into a marketplace the company calls the Flexistore. This is a collection of business-grade applications that use the Flexibles AI core as their intelligence layer. Current products include FleetPulse for vehicle tracking and DevConsole for IoT monitoring. By building specialized tools for fleet management and hardware monitoring, Flexibles is moving away from the "general assistant" trap and toward a role as a specialized business operating system.

The neutrality strategy

Founded in 2024 by Harshvardhan Mahida, an engineer with a background in building AI solutions for brands like Sephora, Flexibles is clear about its place in the stack. They do not claim to build foundation models. Instead, they position themselves as a neutral party—a "Switzerland" for the AI era. This neutrality is a hedge against the ongoing model wars; as long as different LLMs excel at different tasks, a layer that manages context across all of them remains valuable.

The company operates on a freemium model with a focus on privacy and data sovereignty. Data is hosted in Europe (Supabase Europe), and the company emphasizes GDPR compliance as a core differentiator. For enterprises, they offer white-label solutions, allowing organizations to deploy the memory and workflow layer under their own branding while maintaining control over the underlying data.