Actiquest's Membria platform provides an edge‑first AI assistant that runs offline on field devices, using a hybrid memory layer and on‑device inference to deliver expert knowledge without continuous cloud connectivity. The system reduces GPU inference costs by up to 75 % and cuts technicians’ problem‑resolution time by roughly 37 % while maintaining enterprise‑grade security and compliance.
Funding
Funding not disclosed
Founders
Product
Problem
Enterprises with large, mobile field workforces often rely on cloud‑based AI services that require constant connectivity, incur high GPU inference costs, and introduce latency that slows decision‑making and reduces on‑site productivity.
Solution
Actiquest delivers hyper‑personalized, embodied AI through its Membria platform, an edge‑first assistant that runs offline on field devices and executes expert knowledge without a network connection. The system combines a hybrid memory layer (graph, vector, SQL/NoSQL) with on‑device inference to cut GPU costs by up to 75 % and reduce technicians’ search time by roughly 37 %. Membria’s AI Orchestrator decomposes complex tasks into executable plans, while SkillForge applies LoRA‑style model patches to specialize the assistant for any domain. A two‑tier edge‑to‑private‑cloud architecture ensures data stays under enterprise control, and built‑in security, encryption, and compliance meet privacy requirements. Enterprise licenses, pilot‑project fees, and optional support subscriptions fund the offering.
Target Audience
Primary customers are enterprises that operate large, mobile field workforces in sectors such as industrial robotics, automotive R&D, energy, utilities, and field‑service logistics.
Features
- Edge‑first cognitive architecture that performs inference locally on robots or handheld devices, eliminating reliance on continuous cloud connectivity.
- Hybrid Knowledge Layer that indexes local files in a multimodel database (graph, vector, relational and NoSQL) for fast, context‑aware retrieval without external queries.
- Distillation on Demand: when a knowledge gap is detected, the assistant securely queries powerful external models, distills answers into verified local knowledge, and updates its memory.
- SkillForge: dynamic LoRA‑style patches from a marketplace or generated on‑the‑fly to instantly specialize the base model for specific tasks.
- AI Orchestrator with visual workflow builder that decomposes high‑level goals into step‑by‑step plans, invoking tools and managing stateful execution.
- 75 % reduction in GPU inference costs and 37 % faster problem resolution for field technicians.
- Enterprise‑grade security: end‑to‑end encryption, role‑based access controls, and compliance with major data‑privacy standards.
- 24/7 support, implementation assistance, and a six‑month free trial for pilot deployments.