ARTIFICIAL INTELLIGENCE

Cortex

Cortex is a Swymble Labs AI platform built around persistent long-context memory and operator decision support for complex digital workflows.

IN DEVELOPMENTTEASERUPDATED MAR 2026

REQUEST PRIVATE DEMO READ BLOG

What Cortex is

Cortex is Swymble’s AI research project: a cognitive platform whose whole premise is that an assistant is only useful over a long engagement if it can remember one. It is aimed at operator decision support — the person driving a complex digital workflow — rather than at one-off question answering.

Architecturally it is built on persistent memory with a hybrid retrieval layer over pgvector and Redis, fronted by a multi-provider LLM gateway with failover so no single model provider is a single point of failure.

The interface side is treated as its own research problem. Cortex’s UX experiments are operator-first: what a person needs to see to trust and correct a system that is holding context on their behalf.

Cortex is under active R&D, with the detailed architecture private and controlled pilot evaluations in progress.

How it works

Persistent long-context memory

Context carries across sessions instead of resetting, which is what makes the platform useful for ongoing work rather than single questions.

Hybrid retrieval

Retrieval runs over pgvector and Redis together, trading off semantic recall against latency.

Multi-provider LLM gateway

Model calls route through a gateway with failover across providers, so one provider outage is not a platform outage.

Operator-first UX

The interface research focuses on what an operator needs in order to trust, inspect and correct the system’s memory.

Highlights

Private architecture under active R&D
Operator-first UX experiments
Controlled pilot evaluations in progress

At a glance

Category: Artificial intelligence, R&D
Retrieval: pgvector and Redis hybrid
Model access: Multi-provider gateway with failover
Status: In development — controlled pilots

Frequently asked

What is Cortex?

Cortex is an AI platform built by Swymble Labs. It focuses on persistent long-context memory and operator decision support, using a hybrid pgvector and Redis retrieval layer behind a multi-provider LLM gateway.

Is Cortex publicly available?

No. Cortex is under active R&D with a private architecture and controlled pilot evaluations. Private demos can be requested by email at [email protected].