32dots

32DOTS / PEOPLE · SYSTEMS · IMPACT

The future of AI
is distributed.

A network of people building distributed AI.
Startup intelligence. Edge platforms. Impact analysis.
Shared memory—from the individual to humanity.

World Impact Model

32DOTS / A NETWORK OF PEOPLE

We connect expertise,
build AI systems,
and examine their impact.

HEIDELBERG, GERMANYLOCAL ROOTS. DISTRIBUTED FUTURE.
FILM: MARROS / PEXELS ↗
LOCAL CAPABILITY → CONNECTED INTELLIGENCEAI systems advisory · Platform development

THE 32DOTS NETWORK

Four connected fields of work.

Find relevant companies. Run AI locally. Examine who benefits and who pays. Keep control of knowledge.

01

STARTUP ECOSYSTEM / GALAXY

Find startups, funders
and their connections.

Galaxy brings startup profiles, funders, people and news into a searchable view of Europe and Israel. Use the map and connected profiles to research potential investments, partners and competitors.

Open live crawler ↗
FROM SOURCES TO CONNECTIONS
01 / PROBLEM

Research is scattered.

Company websites, funding information and news sit in separate places. Building a useful picture requires repeated searches.

02 / SYSTEM

Collect and connect sources.

Galaxy crawls websites and connects company profiles with people, funders and news.

03 / WHY IT MATTERS

Investigate with context.

Find relevant organisations, follow their relationships and check the source material before making a decision.

GALAXY CRAWLER RECORDED FROM THE APPLICATION
Explore Galaxy ↗
02

DISTRIBUTED EDGE AI

Run AI where
the data is generated.

Sending every sensor reading or internal task to a cloud service adds data transfers, latency and dependency. We are building local processing with 1hive, agent coordination with 1agent, and private device connections with 1net.

THREE COMPLEMENTARY PROJECTS
SENSE & ACT

1hive.

CONCEPT WALKTHROUGH
01 / PROBLEM

Sensitive data leaves the site.

Cloud processing can move operational and personal data outside your control and make local decisions depend on a network connection.

02 / SYSTEM

Process and act locally.

1hive connects sensor and machine data with local intelligence and actions. The design keeps processing close to the equipment and people using it.

03 / WHY IT MATTERS

Privacy, control, efficiency.

Local processing limits external data exposure and supports data minimisation for GDPR. Right-sized models and fewer transfers aim to reduce energy use; savings depend on the workload.

Platform & interactive demoExplore 1hive ↗
COORDINATE & REMEMBER

1agent.

CONCEPT WALKTHROUGH
01 / PROBLEM

Each task starts from scratch.

When tools lose project context, people repeat instructions, reconstruct decisions and manually pass work between systems.

02 / SYSTEM

Retain context across tasks.

1agent is our planned agent layer for coordinating tasks with persistent project context and memory.

03 / WHY IT MATTERS

Less repeated setup.

The goal is to carry decisions and relevant knowledge into the next task, reducing repeated explanations and manual handovers.

Coming soonAsk about 1agent ↗
CONNECT

1net.

CONCEPT WALKTHROUGH
01 / PROBLEM

Connecting devices is manual.

Devices across homes, teams and sites need approved connections without configuring each network link by hand.

02 / SYSTEM

Approve devices. Connect them.

1net provides device enrollment and encrypted private networking, with approval controlled by the network administrator.

03 / WHY IT MATTERS

Control network membership.

Connect approved devices without giving every member administrator access. Network membership and remote management are separate permissions.

Windows private preview

The ecosystem direction. Product availability and integration maturity vary by project.

03

WIM / WORLD IMPACT MODEL

Who benefits?
Who bears the cost?

Financial returns alone do not show who gains, who loses or which costs move elsewhere. WIM compares effects on money, energy, talent, material and information across eight levels—from an individual to humanity.

Open WIM in Galaxy ↗

THE GALAXY WIM MATRIX

Compare the change. Inspect the evidence.

Read across for the five flows; read down for the eight levels. Each affected cell compares the baseline with the scenario if the company succeeds. Higher means better; the gap shows the change. Negative effects and missing evidence remain visible.

Original Galaxy example: Ackerlicht, a fictional farm-sensor startup. The cells and evidence are invented; the scoring weights are not yet calibrated. This matrix shows impact if it succeeds, not a probability of success or a failure scenario.

Select a cell to read its explanation in Galaxy. On a narrow screen, scroll the matrix sideways. View the full worked example ↗

04

WORLD MEMORY MODEL

Keep the knowledge.
Change the model.

When knowledge is locked inside one AI service, changing supplier can mean rebuilding context and losing access to past work. Our World Memory Model separates memory from replaceable models. Its eight-level vision extends from personal knowledge to shared organisational and public resources.

Open full experience ↗
EIGHT LEVELS / WORLD MEMORY INTERACTIVE VISION

The original eight-level vision, embedded here. The animation illustrates the model; it is not a live deployment or measured forecast.

01 / Own the memory02 / Choose the model03 / Reuse shared knowledge

WHO WE ARE

32dots is a network.
AI advisory and platform development.

We bring together people who advise leadership teams, assess technology for investors and develop AI platforms. Our projects address local processing, private networking, startup research and control over institutional knowledge.

01 / Expertise

Assess the problem.
Build the system.

01 — ADVISE

C-level AI advisory.

Identify where AI is useful, what data it needs and what it will cost to operate. Decide what to build, buy or integrate, and which capabilities your team needs to retain.

  • AI strategy & opportunity assessment
  • Build, buy & partner decisions
  • CTO-level guidance & AI operating models
02 — ASSESS

Technical due diligence.

For PE and VC: examine architecture, data access, supplier dependencies, operating costs and team capability. Identify what is demonstrated, what is assumed and what still needs testing.

  • Technical & business due diligence
  • Architecture, data & dependency review
  • Technical risks & development priorities
03 — BUILD

AI platform development.

Develop agent systems, persistent memory, knowledge graphs and edge AI. Integrate models with data and existing software, define human approvals, and test the workflow against the task it must perform.

  • AI products & agent systems
  • Knowledge & intelligence platforms
  • Prototyping, integration & delivery

THE CONNECTION LAYER

1net.

Windows private preview

A Windows connector for approved device enrollment and encrypted private networking. The private preview includes a device dashboard and an administrator-controlled approval flow.

Windows installation and private network connections have been tested on selected devices. Access is a private preview; macOS and Linux packages are not available.

Network membership does not grant remote desktop, SSH or agent administration.

Discuss 1net ↗