How much funding has Thalamus Intelligence raised?

Thalamus Intelligence has raised $5M in total funding. Their latest funding round was Seed. Key investors include Shenzhen-based top-tier fund, Industrial capital (specific names not publicly disclosed).

Where is Thalamus Intelligence headquartered?

Thalamus Intelligence is headquartered in Shenzhen, China. The company has a team of 11-50 people.

What does Thalamus Intelligence do?

Thalamus Intelligence operates in the AI Infrastructure sector. China's first native multimodal long-term memory AI company, building a memory infrastructure layer that enables AI to remember, understand, and proactively serve users across sessions, modalities, and time. A user-centric third-party memory layer that exists independently of foundation models, enabling AI to achieve personalized, continuous understanding of users across different models, sessions, and modalities — the critical missing layer between base model capability and truly useful AI applications.

When was Thalamus Intelligence founded?

Thalamus Intelligence was founded in 2025. It is based in Shenzhen, China.

Who is the CEO of Thalamus Intelligence?

Zhang Yuan (张源) is the CEO of Thalamus Intelligence. The company was founded in 2025.

Where is Thalamus Intelligence located?

Thalamus Intelligence is headquartered in Shenzhen, China.

What funding stage is Thalamus Intelligence at?

Thalamus Intelligence is currently at the Seed stage.

How many employees does Thalamus Intelligence have?

Thalamus Intelligence has approximately 11-50 employees.

Thalamus Intelligence

Category: AI Infrastructure

China's first native multimodal long-term memory AI company, building a memory infrastructure layer that enables AI to remember, understand, and proactively serve users across sessions, modalities, and time. Thalamus Intelligence was founded in 2025. The company is led by Zhang Yuan (张源). Based in Shenzhen, China. Team size: 11-50. Total funding raised: $5M. Latest round: Seed. Key investors include Shenzhen-based top-tier fund, Industrial capital (specific names not publicly disclosed).

Founded: 2025
Headquarters: Shenzhen, China
Team size: 11-50
Total funding: $5M

Value proposition

A user-centric third-party memory layer that exists independently of foundation models, enabling AI to achieve personalized, continuous understanding of users across different models, sessions, and modalities — the critical missing layer between base model capability and truly useful AI applications.

Products and solutions

MemAura (native multimodal memory foundation module, launched June 2026), OmniMemory (memory framework with ADK/API), STKG (Spatio-Temporal Knowledge Graph), MEMLENS benchmark (co-released with NVIDIA, HKUST, CUHK)

Unique value

Native multimodal memory architecture (not text-only) with bionic hippocampus-inspired design; Spatio-Temporal Knowledge Graph (STKG) that anchors memory in time and space; 40-49% token cost reduction with sub-400ms retrieval latency; the only company in China exclusively focused on multimodal long-term memory.

Target customer

AI hardware manufacturers (companion robots, AI toys); vertical Agent scenarios (AI customer service, digital employees); embodied AI/robotics companies; legal and financial sector Agent applications

Industries served

AI companion hardware, AI customer service, digital employees, embodied AI/robotics, legal tech, fintech

Technology advantage

First-mover in multimodal long-term memory in China; bionic memory architecture mimicking human hippocampus; STKG with time-space anchoring for accurate recall; cold-hot storage tiering for cost efficiency; MEMLENS benchmark (global first multimodal long memory benchmark); SOTA on LoCoMo and LongMemEval benchmarks; latent space memory (E2P) for proactive intelligence

How they differentiate

Native multimodal architecture (not text-only RAG); time-anchored Spatio-Temporal Knowledge Graph vs. pure vector search; bionic hippocampus-inspired design; independent third-party memory layer not tied to any single foundation model; focus on proactive/active intelligence rather than passive response; sub-400ms latency with 40-49% token reduction

Main competitors

Mem0 (open-source AI memory), Letta (formerly MemGPT), Zep (context engineering and agent memory platform)

Key partnerships

NVIDIA (co-released MEMLENS benchmark), Hong Kong University of Science and Technology (HKUST), Chinese University of Hong Kong (CUHK)

Notable customers

Companion hardware manufacturers (including IP-licensed products like Beijing Zoo panda Menglan), AI solution providers for companion products, software companion products

Major milestones

Founded November 2025, STKG achieved SOTA on LoCoMo and LongMemEval benchmarks (December 2025), Launched MemAura memory foundation module (June 2026), Co-released MEMLENS benchmark with NVIDIA, HKUST, CUHK (May 2026) — global first multimodal long memory benchmark, Hugging Face Daily Paper top 3, Completed seed round of tens of millions RMB (July 2026)

Growth metrics

MemAura: 40-49% input token reduction; <400ms memory retrieval latency; <1s end-to-end first response; >80% comprehensive accuracy; STKG achieved 82.5% accuracy on LoCoMo benchmark with P95 retrieval latency under 2 seconds

Market positioning

First-mover and only dedicated multimodal long-term memory company in China; positioning as the independent memory infrastructure layer (analogous to Supabase for memory); competing in a nascent trillion-yuan (RMB) market with few players; highest competitor valuation in space under ¥20B

Geographic focus

China (domestic market) with global benchmark contributions

About Zhang Yuan (张源)

Ex-COO at an autonomous driving company; venture capital industry experience; embodied intelligence sector experience; Peking University (Electronics & Economics dual degree)

Latest news about Thalamus Intelligence

Thalamus Intelligence raises tens of millions in funding for multimodal long-term memory base model, targeting proactive AI

More AI Infrastructure companies

Official website: https://www.omnimemory.ai/