Ex10siv. alive Talk to Ava

Thousands of AI agents. Your rules. Your people in control.

Ex10siv builds expert AI agents that analyze, recommend and act, while people approve the important actions. Retail is the first module, and it runs today as a demo on a simulated 63-branch supermarket chain.

  • Replenishment: 756
  • Fresh markdown: 756
  • Pricing watch: 756
  • Supplier: 300
  • Incident sentinels: 126
  • Workforce: 63
  • B2B & credit: 63
  • Scouts & strategy: 32
Agents awake–
Proposals made–
Awaiting a person–
Executed within limits–
Blocked by policy–

2,852 agents in eight families. Each one sleeps until an event wakes it. Simulation, not live customer data.

A massive market. Relentless competition.

Retail remains a scale business, but growth and profitability are harder to capture.

$28.7T
Projected global retail sales, 2028 forecast Forrester, May 2024
+2%
Global retail real growth, 2025 actual Euromonitor, June 2026
76%
Of retail sales still in stores, 2028 forecast Forrester, May 2024

Retail doesn't need another dashboard.

The hard part is turning thousands of daily signals into timely, coordinated decisions.

Margin pressure
95% of surveyed retail leaders expect higher trade-policy-driven costs.
Stock vs. waste
Avoid empty shelves without overbuying or throwing out fresh goods.
Local complexity
Decisions vary by SKU, store, region, demand and timing.
Slow execution
Insights wait for cross-team review before anything reaches the store.
Traditional
  1. Report
  2. Investigate
  3. Coordinate
Agentic
  1. Sense
  2. Decide
  3. Approve
  4. Act

Business logic and policy checks remain essential. AI recommendations are not automatic approvals. Survey: Deloitte 2026 Global Retail Outlook, 330 retail executives.

One agent factory. Expert teams for any business.

Every agent learns in three steps, and each step adds its own rules.

  1. Core agent

    One shared engine: AI models, tools, approvals and audit trail.

    Behavior rules: own tools only, people approve key actions.

  2. Industry expert

    Learns the know-how of one industry, starting with retail.

    Industry regulation, such as consumer pricing law.

  3. Your expert team

    Connects to your data and learns the tasks you need done.

    Company policy, such as margin floors, discount caps and approval limits.

Modules: pick any industry and any department
IndustryStrategySales & pricingSupply chainOperationsWorkforce
Retail 8 agent familiesWorking todayWorking todayWorking todayWorking todayWorking today
Financial servicesBuilt as neededBuilt as neededBuilt as neededBuilt as neededBuilt as needed
HealthcareBuilt as neededBuilt as neededBuilt as neededBuilt as neededBuilt as needed
Any industryBuilt as neededBuilt as neededBuilt as neededBuilt as neededBuilt as needed

Working today Can be built as needed, for any department

From retail data to intelligent action.

A practical AI layer across the six core areas of retail operations, built for executives, category managers, branch teams and analysts.

Pricing & competition

Compare shopping baskets, expose price gaps and support pricing decisions.

Inventory & fresh

Recommend replenishment and markdowns before stockouts and expiry.

Promotions

Simulate campaigns and route proposed promotions for approval.

Branch intelligence

Spot underperformance and drill from region to department.

Suppliers & B2B

Manage order pipelines, invoice matching and supplier scorecards.

Workforce

Find staffing gaps, watch labor cost and manage shifts.

A 3-minute walkthrough of the retail module. Store data in the demo is simulated; market data is real.

The value: better decisions, executed earlier, at scale

Protect margin
Gross margin, price index
Reduce waste
Fresh shrink, days of cover
Lift availability
On-shelf rate, out-of-stocks
Speed decisions
Approval cycle, manager hours

Suggested pilot KPIs, not demonstrated financial results.

People stay in control.

Every agent has an autonomy level, from L0 to L3, and a kill switch. Move the slider to see which steps the agent takes and which a person approves.

  • People approve by role and amount, from branch manager to executive.
  • Policy checks test every proposal against your rules.
  • Every agent run, AI call and action is logged.
  • Personal data is masked before any AI model sees it.

Recommend. The agent proposes a prioritized action with evidence. People approve and carry it out.

  1. Detect
  2. Evaluate
  3. Propose
  4. Authorize
  5. Execute

Agent Person

On our roadmap

Ex10SIVE: a large model that runs like a small one.

Our own AI model, built on open weights and adapted for retail work and Hebrew. A router picks the 8 best of 256 experts for each word, so only 23B of its 501B parameters do the work.

21× less computing per word than a standard model of the same size. Fits on one 8-GPU server.

Lower cost at high volume

Routine agent work at a fixed server cost, not a bill for every word.

Your data stays inside

Runs on your own servers. Open weights mean no lock-in.

Leading models when needed

The hardest questions go to Claude or GPT, with personal data masked.

Figures: Beam, Reflection AI's open model (Apache 2.0, Oct 2026), which Ex10SIVE would build on. 21× compares computing per word with a standard model of the same size.

Where this goes

We specialize in agentic AI today, with superintelligence as our long-term direction.

  1. Today

    Governed teams of expert agents running a full retail chain, with people approving what matters.

  2. Next

    More industries and departments on the same agent factory, and our own efficient model.

  3. Beyond

    Toward more general, more capable intelligence, built on the same principle: it works inside your rules, with people in control.

The team

Dr. Adi Hod

Chief Executive Officer

Strategic execution. Head of Ex10siv's projects and solutions.

adi@ex10siv.ai

Yossi

AI expert, solution architecture

Applied mathematics and cybersecurity. Inventor on about 150 patent applications and CTO of high-tech companies for 30 years.

Scott

Market expansion

Senior software engineer in AI and cybersecurity, with four years focused on oversight, control and alignment of autonomous agents.

The 30-second introduction

The film that opens ex10siv.ai, with recorded English narration.

Watch the introduction

  1. Welcome to Ex10siv.
  2. Expertise in agentic AI. A fresh perspective on what’s next.
  3. Here, discovering our company begins with a conversation.
  4. Your questions. Your perspective. A conversation shaped around what matters to you.
  5. Meet Ex10siv ALIVE. Let’s discover what’s possible, together.

See it run on a full retail chain.

Ask Ava anything, by voice or text, or leave your details for a live demo with the team. No account needed.