Edge AI Innovations Inc.
We are proud creators of
Build AI agents that remember, learn, and stay yours.
We are the proud creators of Memanto and Moorcheh — two open-source tools built to give your AI agents the memory and retrieval infrastructure they actually need.
Deploy intelligent AI agents and chatbots that improve with every interaction
Give your agents persistent, queryable memory with Memanto's companion memory agent
Power retrieval with Moorcheh's information-theoretic search engine — no vector database required
Maintain strict data privacy and comply with industry regulations with on-prem or serverless deployment
Switch to the newest foundation models as soon as they become available — Moorcheh is compatible with all LLMs, embedding models, and multimodal AI
Scale your knowledge base to millions of documents without paying for cloud clusters or vector databases
Use everything through a simple API on your website, application, or agent framework of choice
Why Moorcheh?
Moorcheh is a fundamentally new mathematical approach to semantic search — not an optimization of existing vector search, but a complete architectural rethink using information-theoretic principles.
40% higher semantic accuracy than standard vector search methods
32× less memory — runs on edge hardware and serverless cloud with zero always-on infrastructure cost
100× less computation — scale-to-zero means you only pay when your agents are working
No vector database needed — Moorcheh handles context retrieval end-to-end, from efficient vector representations to ranked results
Kiosk Mode — pre-emptive guardrails that restrict agent responses to your uploaded documents only, with customizable fallback responses for out-of-scope queries
Model-agnostic — works with any LLM, embedding model, or multimodal foundation model, with seamless switching between them
Why Memanto?
Memanto is the companion memory agent that turns stateless AI agents into systems that actually learn. Built on Moorcheh's retrieval engine, it gives agents the ability to remember what mattered, recall it in under 90ms, and reason over it with full RAG intelligence — across every session, every interaction, every workflow.
SOTA on LongMemEval (89.8%) and LoCoMo (87.1%)
13 typed memory categories with automatic conflict detection
Open-source, MIT licensed — your memory stays yours
Our Vision
A world where the most powerful AI capabilities belong to everyone — not just those with the largest infrastructure budgets or the deepest research teams. We believe the next era of AI will be defined not by who builds the biggest models, but by who builds the most thoughtful systems around them.
Our Mission
To advance human-centric AI by building open, efficient, and trustworthy tools that give developers, researchers, and organizations full ownership of their AI systems — their data, their memory, their intelligence.
Our Contribution
The AI ecosystem moves fast. Most of that motion concentrates value in a handful of platforms, leaving the broader community dependent on infrastructure they don't control and can't inspect. We're working against that.
Everything we build starts from first principles — mathematically grounded, architecturally honest, and designed to run where you need it: on the edge, in your cloud, or on a single machine. Moorcheh's information-theoretic search engine and Memanto's companion memory agent are both open-source because we believe the infrastructure of intelligence should be a public good, not a proprietary lock-in.
Open Source & Community
We are proud contributors to the open-source AI ecosystem. Our repositories are built to be forked, extended, and improved. We actively maintain integrations with LangChain, LlamaIndex, CrewAI, LangGraph, MCP, n8n, HuggingFace, and Cohere — not because it serves our roadmap, but because it serves the community.
We host a growing Discord of researchers, builders, and practitioners working on agentic memory. We publish our research openly on arXiv. We run community benchmarks so the field can hold us — and itself — accountable.
Human-Centric by Design
Every architectural decision we make asks the same question: does this give people more control, or less? Deterministic retrieval over probabilistic approximation. On-premise deployment over forced cloud dependency. Typed, inspectable memory over opaque context dumps. Open licensing over proprietary walls.
AI should augment human judgment, not obscure it. That principle is non-negotiable for us — and it shows up in every line of code we ship.
Our strategic partnerships with industry leaders ensure cutting-edge performance and reliability:
NVIDIA: Leveraging state-of-the-art GPU technologies to accelerate AI computational capabilities
Intel: Utilizing advanced CPU and AI acceleration technologies for enterprise-grade computing
Advantech: Providing robust, industrial-grade edge computing hardware designed for mission-critical AI deployments
Together, we combine best-in-class software and hardware to deliver unparalleled AI solutions that meet the most demanding enterprise requirements.
Our Partners
Events
Edge AI Community & Learning Initiatives
Edge AI Engineering Online Course
A comprehensive, self-paced online learning program that takes professionals from foundational concepts to advanced implementation of edge AI technologies. The course covers:
Secure AI infrastructure design
Edge compute architecture
AI model optimization
Privacy-preserving AI techniques
Hands-on project development with real-world case studies
Edge AI Innovations Hackathon
An annual coding challenge that empowers students, developers, researchers, and entrepreneurs to solve real-world problems using edge AI technologies. Participants compete for:
Potential seed funding
Mentorship from industry experts and university professors
Opportunity to showcase innovations to leading tech investors
Annual Edge AI Innovations Summit in Toronto
Our flagship conference uniting industry pioneers, academic researchers, and technology leaders to discuss the future of decentralized AI. The summit features keynote presentations, technical breakout sessions, and networking opportunities for professionals committed to advancing AI innovation. You can access the recordings of the past event by clicking on the link below:
Go to https://genaigenesis.ca/ for Registration and more information
Canada’s Largest AI Hackathon, In-Person: Toronto, ON
Edge AI Innovations Summit 2024 Toronto. Past Event.
Edge AI Innovations Summit 2024 Toronto. Past Event.
In collaboration with:
Our Team
Tara Khani
Co-Founder and CEO @ Edge AI Innovations Inc.
With 17 years of experience in engineering and leadership, Tara Khani is passionate about collaborating with customers to understand their real pain points and deliver tailored solutions that drive growth. She holds a Master of Applied Science in Aerospace Engineering and spent 8 years as a Senior Engineer at two leading Japanese automation, sensor, and machine vision companies, Keyence and Mitutoyo. Transitioning into business development and growth leadership, she served as a Territory Manager at Ansys partner (a Synopsys company), coaching and leading a team to become trusted advisors for simulation software, consulting, POCs, and project implementations. Later, as a Senior Enterprise Manager at SOTI, an MDM company, she led her team to provide solutions that addressed the specific engineering needs of Fortune 500 customers, maximizing their ROI. Tara's diverse experience and dedication to lead a passionate team to solve customer challenges make her a valuable asset in bridging the gap between engineering innovation and real-world business solutions.
Majid Fekri, PhD
Co-Founder and CTO @ Edge AI Innovations Inc.
For the past 12 years Majid has been working as scientific programmer, and lead data scientist in various tech startups and large companies. He is a dynamic and engaging scientist and engineer who is passionate about the transformative power of AI. With a PhD in Atmospheric Sciences from McGill University and a proven track record in scientific research and hands-on experience in data science and AI, Dr. Fekri brings a unique blend of scientific expertise and real-world experience to the stage.
