What Flower does

Flower Computer Company is a Brooklyn-based, privately held software company developing infrastructure for the AI era, with founding research dating to 2018. The company's current product portfolio centers on two core offerings: Hivemind, a meta skill for autonomous agent harness systems integrated with Codex and Claude Code that enables agents to share skills and discrete experiences through a shared memory pool; and Ese, a static embedding model built in Rust that materializes model weights and tokenizers as static globals at compile-time using build.rs, achieving two orders of magnitude faster embedding inference than the source SentenceTransformers static-retrieval-mrl-en-v1 model and roughly 400x faster than transformer-based models on CPU, with no accuracy degradation. The company's research program also yielded Bog (a queryless database runtime) and Fold (an incremental programming kernel), both of which underpin a novel networked-memory architecture that treats memory as fundamentally relational between agents rather than contained within them, and reportedly delivers query response times orders of magnitude faster than PostgreSQL for relevant workloads.

In April 2026, Flower shut down Yuma, its consumer-facing iOS animist social network launched in summer 2025, after determining that the database architecture developed for Yuma represented a more valuable asset than the consumer app itself. The company is now repositioning as an infrastructure platform targeting AI developers and agent builders. Flower is pre-revenue, having raised a single $1.5M pre-seed round in October 2024 co-led by Village Global and Worldbuild. Its go-to-market is primarily product-led and community-led, distributing software through GitHub, Rust's crates.io registry, and an active Discord server. The only documented third-party commercial validation is an Enzyme integration of Ese that achieved 6x improvement in corpus ingestion times.

The company operates with 1-10 employees, including co-founder and CEO Édouard Urcades (formerly of Gnosis and Tlon) and CTO Sam Panter, with William P. Abney as a contributing researcher and Owen James as a NYU intern. The team is actively hiring founding engineers despite limited capitalization, signaling both ambition and runway pressure. Distribution is global via open-source channels but geographically restricted on the consumer side via iOS exclusivity, and no enterprise pricing model has been disclosed.

Name

Flower

Legal name

Flower Computer Company

Website

https://flowercomputer.com

Company type

Private

Founded year

2025

Operating status

Operating

Headcount range

1–10 employees

Short description

Flower Computer Company is a Brooklyn-based, pre-revenue software startup building AI-era infrastructure, including Ese, the world's fastest static embedding model, and Hivemind, a shared memory system for autonomous AI agents, serving developers building AI applications.

Ownership category

akta.pro rank

Product category

AI Infrastructure Software

NAICS

Software Publishers (5132), Software Publishers (51321)

SIC

Services-Prepackaged Software (7372)

akta.pro primary industry

End-to-End Enterprise AI Platforms (MLOps & Model Lifecycle Management) (HDAEANAA)

akta.pro secondary industries

Model Deployment, Serving & Inference Platforms (HDAAABAF), Model Development & Training Platforms (AutoML, Notebooks, Feature Stores) (HDAEANAB), AI/ML Solution Integration & MLOps Enablement (BPAEAAAJ)

Keywords

• Static embedding models
• Queryless database runtime
• AI agent infrastructure
• Incremental programming kernel
• Networked memory systems

Headquarters

HQ city

Brooklyn

HQ country

United States

HQ region

North America

Markets served

GTM type

B2B

Offering type

Software

Cost components

Personnel, Technology or R&D, Infrastructure, Marketing or Sales

Revenue model

1. Pre-revenue / Research Stage: The company raised $1.5M pre-seed funding and is in research and development phase. Revenue model not yet established - focusing on building infrastructure platform based on database architecture discoveries from Yuma development.

Go-to-market motion3 records

Distribution channels2 records

Marketing channels7 records

Core offering

Flower Computer Company builds novel software infrastructure for AI systems, including a queryless database runtime (Bog), an incremental programming kernel (Fold), the world's fastest static embedding model (Ese), and a meta skill for autonomous agent knowledge sharing (Hivemind). The company turns research in networked memory and compile-time model materialization into developer-facing software distributed through open-source channels. Current revenue is pre-revenue; the intended monetization model is subscription-based infrastructure.

Product overview

Flower Computer Company is building foundational software infrastructure for the AI era. Its current product portfolio centers on Hivemind (a meta skill for autonomous agent skill-sharing) and Ese (the world's fastest static embedding model). The company's prior work includes Bog (a queryless database runtime), Fold (an incremental programming kernel), and Yuma (an iOS animist social network that was shut down in April 2026). Flower is currently pivoting toward developing the database architecture discovered while building Yuma — a novel system that combines vector, relational, and graph database characteristics. The company has also published open research on database design, memory systems, and static embedding models.

Differentiator

Problem solved

Functional benefit

Brands

• Yuma: An iOS-exclusive animist social network and magic camera app that lets users photograph physical objects and chat with them as digital characters. Shutdown announced in April 2026.
• Hivemind
• Ese
• Bog
• Fold

Products and services

• Hivemind A meta skill for agent harness systems such as Codex and Claude Code that enables autonomous agents to share skills, strategies, and discrete experiences through a shared memory pool, allowing agents to commit, search, and vote on 'mindchunks' of knowledge without human orchestration.
• Ese (Static Embedding Model) The world's fastest static embedding model, built in Rust and compiled directly into application binaries at compile-time. Achieves zero active parameters, two orders of magnitude performance improvement over the original static model, and roughly 400x faster than transformer-based models on CPU with no degradation in retrieval accuracy on the NanoBEIR benchmark. Distributed as an early-access Rust crate for developers building semantic search and knowledge base pipelines.
• Bog A queryless database runtime developed by Flower, representing a novel database architecture designed to represent relationships and memory in fundamentally different ways than conventional relational or graph databases, built on selective processing and concealment principles.
• Fold An incremental programming kernel based on incremental computing principles, enabling efficient recomputation of derived values when inputs change without recomputing everything from scratch.

Quantifiable outcome

• 6x improvement in corpus ingestion times for partner Enzyme
• +3 more outcomes

Named customers1 record

Segments2 records

Ideal customer profiles2 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration3 records

AI capability9 records

Feature4 records

Partnerships

Two partnerships are on record, tiered notable and minor.

• EnzymenotableTechnology or IntegrationEnzyme integrated Flower's static embedding model into their knowledge base product, achieving significant performance improvements. Josh from Enzyme documented the integration results publicly.
• CyrilminorOthersLent Raspberry Pi hardware for benchmarking the static embedding model against Mac Mini.

Scale indicators3 records

Recent moves6 records

Expansion highlights6 records

Direct peers

• Weaviate: Open-source vector database combining vector, keyword, and graph-style retrieval. Comparable to Flower's Bog for hybrid retrieval and networked-memory use cases in AI applications.
• Pinecone: Managed vector database widely used for embedding-powered search and agent memory. Comparable to Flower's Bog as a high-performance vector/AI-native database serving production retrieval workloads.
• Chroma: Open-source embedding database widely adopted by AI developers and agent builders. Comparable to Flower's developer-focused database tooling for retrieval-augmented and agent-memory workloads.
• Qdrant: Rust-native open-source vector database with managed cloud offering. Highly comparable to Bog given shared Rust ecosystem and focus on performance-oriented vector retrieval.
• Model2Vec (MinishLab): Static embedding models (Potion family) optimized for fast CPU inference. Direct technical competitor to Flower's Ese static embedding model on the same axes of throughput, footprint, and accuracy trade-offs.
• Zilliz (Milvus): Commercial steward of the open-source Milvus vector database. Comparable to Bog as an enterprise-grade vector database with similar performance and scale targets.

Emerging players

• Mem0: Memory layer for LLM applications and agents, focused on personalized long-term memory. Comparable to Hivemind in providing a shared/persistent memory primitive for agent ecosystems.
• Letta: Open-source framework and cloud service for stateful AI agents with persistent memory. Closely comparable to Flower's Hivemind in targeting the agent-memory and agent-state layer.
• Turbopuffer: Serverless vector database with object-store-backed storage targeting large-scale retrieval workloads. Comparable to Bog as a performance- and cost-optimized vector retrieval platform for AI applications.
• LanceDB: Open-source columnar vector database built on the Lance format, emphasizing serverless and embedded deployments. Comparable to Bog as a developer-friendly AI-native database with an open-source distribution model.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat4 records

Key risks6 records

Key highlights6 records

Customer concentration

• GitHubgithub.com/flowercomputers
• Instagraminstagram.com/flowercomputers
• Xx.com/flowercomputers
• Facebookfacebook.com/profile.php
• LinkedInlinkedin.com/company/flowercomputers

Revenue estimate

Valuation estimate

Number of profiles

Profiles4 records

Funding overview

Funding rounds1 record

Investors2 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Flower