Published web & Windows products
GMD product ecosystem
A live market dashboard and two published Windows apps.
Explore the GMD ecosystem
AI-Driven Software Engineer · Product Builder
Shlomo Shema.
I build software with AI, from business idea to customer delivery.
I connect Windows, web, and mobile applications with the services, data, and ERP systems behind them.
Engineering since 1993Three decades of delivery
Four startupsAccounting, gaming & real estate
Two Store appsPublished Windows products
Across industriesBusiness, enterprise & consumer software
Finance · Accounting · ERP · Hospitality · Defence · Gaming · Real estate · Manufacturing
PRODUCTS & CUSTOMER WORK
Featured projects.
Working ERP assistant
Priority ERP MCP integration
Connects an AI assistant to business data and operations through MCP and APIs.
Explore the Priority integrationReal order processing reached
QuickSales modernization
Preserves sales workflows during a .NET 10 migration, with real order processing reached.
Explore the modernization01 / SELECTED WORK
Built for real use.
Windows and mobile applications built around practical business workflows.
WINDOWS / MICROSOFT STORE
Live World Stock Exchange Clocks
A Windows application showing global stock exchange trading sessions, open and closed status, and countdowns to the next open or close. Published as GMD Markets - Live World Stock Exchange Clocks.
View on Microsoft StoreHYBRID MOBILE / ERP
Mobile sales for Priority
A hybrid mobile sales application for Priority ERP, connecting a mobile user interface with the business system behind it.
MOBILE / BUSINESS SOFTWARE
VDoc mobile application
A mobile application developed as part of my VDoc work, alongside the paperless invoicing solution, administrative interface, and REST APIs.
MOBILE / INVENTORY
Product stock counting
A mobile application for counting product stock, extending my business application work into inventory workflows.
Much of my work was built for customers or my own businesses. Public repositories tell only part of the story.
AI-ERA ENGINEERING
From using AI
to engineering with it.
Product delivery, enterprise modernization, agent tooling, and evaluation. Each project has a different scope and a clearly stated stage.
Explore all six projects. Detailed engineering case studies are available below.
The GMD product ecosystem
Web and Windows published; Android in development
A shared service and session architecture connects the market-data products across web, Windows, and mobile.
QuickSales modernization
Ongoing; real order processing reached
Preserving business contracts and sales workflows while replacing legacy application components.
Priority ERP MCP server & API integration
Working end to end; network deployment gated
An MCP server and .NET APIs connect the assistant to business knowledge and ERP operations.
Claude–Codex AgentBridge
In development
Operator-controlled agent communication with explicit contracts, a live viewer, and tamper-evident logs.
Evaluating AI Mood quality
Audited; quality monitoring continues
Snapshot replay exposes output failures and informs prompt corrections and model selection.
Interfaces across three frameworks
Demos published
Published React, Vue, and Angular ports demonstrate the same interface across different component architectures.
BEHIND THE WORK
Engineering case studies.
The problem, my responsibility, how AI fits, and what the implementation has demonstrated.
The GMD product ecosystem
The problem
Bring global market information into a coherent product that people can use on the web and Windows.
My responsibility
I own interface design, backend services, live data integration, and deployment across the product family.
How AI fits
I use AI-assisted development to build and evolve the applications, alongside market commentary and automation features within the platform.
Evidence and validation
The website is live and two Windows applications are published on Microsoft Store. A shared session library covers 24 exchanges, with explicit API contracts for clients.
Outcome and current scope
Published web and Windows products, with the Flutter Clocks Wall app still in development.
Architecture overview
- Web + Windows clients
- Trader Clocks API
- Shared session library
- Market data providers
QuickSales modernization
The problem
Modernize a legacy sales application while preserving the contracts and business workflows that customers already depend on.
My responsibility
I own the modernization work across application structure, service contracts, UI, data interfaces, and ERP integration. The ServiceStack contract layer is completed; replacing Enterprise Library components remains part of the ongoing migration.
How AI fits
I use Claude-assisted development for the migration, bringing the knowledge of the existing application and responsibility for its target architecture. AI assistance supports implementation; contract compatibility and customer workflows determine whether the work is useful.
Evidence and validation
The work has reached real order processing. The integration work also established three Priority text-storage representations, making encoding behavior an explicit part of the migration rather than assuming every stored string behaves the same way.
Outcome and current scope
A working order flow and completed contract layer within an ongoing .NET 10 modernization. Further legacy component replacement remains in progress.
Architecture overview
- Sales interface
- Service contracts
- Application services
- Priority ERP
Priority ERP MCP server & API integration
The problem
Make Priority ERP knowledge and system operations accessible to an AI assistant through a defined integration layer.
My responsibility
I built the Priority MCP server and its API integration as part of the ERP Assistant. The implementation connects the chat interface, .NET services, structured ERP knowledge, and the business system interfaces.
How AI fits
The assistant uses MCP to reach the ERP integration capabilities and business context exposed by the server. This combines AI-assisted development with an AI-enabled application: the model is a participant in a system with explicit integration boundaries.
Evidence and validation
The assistant works end to end. Network deployment remains gated on a security review; a functioning integration is not presented as proof that it is ready for unrestricted network access.
Outcome and current scope
A working Priority ERP assistant and MCP/API integration, with the deployment boundary stated explicitly.
Architecture overview
- Chat interface
- AI assistant
- Priority MCP server
- API integration
- Priority ERP
Claude–Codex AgentBridge
The problem
Enable two coding agents to exchange work while keeping an operator informed and maintaining a traceable record of the exchanges.
My responsibility
I direct the design of the local Claude–Codex relay: message contracts, agent adapters, operator controls, the viewer, and the durable log. The architecture separates communication and observation from future containment work.
How AI fits
Claude and Codex contribute to the engineering work through separately scoped implementation and review records. Their communication is itself the subject of the product, making the boundaries between agents and operator decisions explicit.
Evidence and validation
The project has recorded design and milestone reviews. The current scope is cooperative local messaging. The log supports traceability; action containment requires additional work.
Outcome and current scope
An ongoing engineering tool with a local relay, operator viewer, explicit contracts, and tamper-evident logging.
Architecture overview
- Claude adapter
- Local relay + contracts
- Codex adapter
- Operator viewer + audit log
Evaluating AI Mood quality
The problem
Assess whether AI-generated market commentary meets the task's quality needs, including cases where an API request succeeds but the output still falls short.
My responsibility
I own the evaluation decisions: investigating observed failures, revising the prompt, comparing model suitability, and deciding which changes to retain.
How AI fits
AI produces the commentary being evaluated. Stored snapshots and real-API replay provide concrete cases for examining behavior; model selection is based on the task rather than accepting a substitution on cost or novelty alone.
Evidence and validation
The audit used stored snapshots and real-API replay. It led to a prompt correction and the rejection of an unsuitable model choice. Monitoring continues because an improved prompt does not establish that every future output will be correct.
Outcome and current scope
An audited AI feature with a corrected prompt, an unsuitable model choice rejected, and continuing quality monitoring.
Architecture overview
- Stored market snapshots
- Prompt + model
- API replay
- Output review
02 / ENGINEERING RANGE
The interface. The engine.
The connections between them.
I work across disciplines because a product succeeds when its parts work together.
Windows, web & mobile
Hands-on interface design and application development across desktop, browser, and mobile. Business workflows, rich interfaces, administrative tools, and hybrid applications.
Services, interfaces & APIs
Windows services, background processing, REST APIs, data interfaces, and integrations between business systems. Database experience spans SQL Server, Oracle, Couchbase, MongoDB, and Redis, with Kafka and RabbitMQ for messaging. ERP integration and production troubleshooting connect these systems to working applications.
AI-driven development
AI is now my primary development workflow. I bring the architecture, product decisions, interface design, and engineering judgment, using AI throughout implementation.
Product thinking. Delivery ownership.
Product management and project management courses complement my practical experience leading projects and owning design and development. I connect what a product needs to do with how it gets built.
03 / ENTREPRENEURSHIP
Four startups.
Hands-on engineering ownership.
I owned and participated in ventures across accounting, gaming, and real estate. I took full design and development ownership in the accounting and gaming ventures, and responsibility for the API backend at rbns.co.il.
ACCOUNTING
Sybo Systems
Ownership and participation, with full responsibility for design and development.
ACCOUNTING
VDoc / Vesper Systems
Design and development ownership across business software, including invoicing, mobile, and APIs.
GAMING
Texas Hold’em
Ownership and participation, taking full responsibility for the game’s design and development.
REAL ESTATE
rbns.co.il
Responsible for the API backend of the real estate venture.
04 / THE JOURNEY
Experience that keeps evolving.
From systems and networks in the 1990s to enterprise products, multidisciplinary systems, and AI.
- 2025 to today
Independent software engineering
Product development, enterprise modernization, and AI integration across my own ventures and customer work.
- 2020 to 2025
Multidisciplinary defence systems
Elbit Systems, through ABRA
Services, interfaces, production debugging, and mentoring across multidisciplinary hardware and software teams.
- 2010 to 2019
Business products & integrations
Silverbyte · Vesper Systems
Hotel ERP integrations and business applications spanning sales, inventory, invoicing, Windows services, and mobile.
- 2004 to 2009
Enterprise delivery & interactive applications
Malam Systems · One Technologies · Sybo Systems
Led and developed CRM, web, telecom, BPM, semiconductor workflows, and gaming interfaces.
- 1993 to 2004
The foundations: systems, code & customers
Iscar Group · MediaTek Group
Built Windows and database systems, managed networks, supported international rollouts, and taught Linux and C.
EDUCATION
Computer science. Software engineering.
B.Sc. Statistics & Computer Science, Haifa University, 1988 to 1992.
Software Engineering studies, Technion, 1999 to 2000.
CONTINUING DEVELOPMENT
Product, projects & data.
Product management and project management courses. Additional database training and certifications include MongoDB, Couchbase, and SQL Server DBA.
05 / A LITTLE MORE CONTEXT
Questions, answered.
Which projects should I look at first?
Start with the three featured projects below the hero, then follow their links to the engineering studies for implementation details and current scope.
What does your Priority ERP MCP integration do?
It connects an AI assistant to business knowledge and ERP interfaces. The implementation works; security review remains a prerequisite for network deployment.
How can I explore projects by engineering layer?
Select a layer in the From Idea to Delivery diagram. Interface takes you to the frontend demos, Logic to AgentBridge, Data to AI Mood evaluation, and Integration to the Priority ERP MCP/API case study. Related project cards are highlighted and all projects remain visible. Select the same layer again or use Clear highlight to reset.
Which of your products are available on Microsoft Store?
Global Markets Dashboard and GMD Markets - Live World Stock Exchange Clocks. Direct Store links are beside my name at the top of the page.
What kinds of software do you develop?
Windows, web, and mobile applications, plus the services, databases, integrations, and APIs behind them, across enterprise and consumer industries.
How do you work with AI?
I use Claude and Codex for implementation and review while retaining responsibility for product decisions, architecture, integration, and customer delivery.
What was your role in the four startups?
I owned and participated in the four ventures listed above: full design and development responsibility in accounting and gaming, and API backend responsibility at rbns.co.il.
Does GitHub show all of your work?
No. Much of my work belongs to customer projects or my own businesses and is not represented in public repositories.
How does product and project management fit your experience?
Courses in both complement my experience defining products, leading projects, and taking responsibility for delivery.
LET’S BUILD SOMETHING USEFUL
Have a product
in mind?
Talk to me about a new application, a system integration, or putting AI to work in your business.