The honest opener: for fifteen years we’ve built mobile apps for Israeli enterprise — Toyota and Lexus, Phoenix, Assuta, Cynet, Theranica, Personetics, Matrix, Partner. This year we’re widening the lane. Starting today, MSApps runs a second division: AI & Automation — autonomous agents, process automation, and language-model integration inside the products you already run.
This isn’t a pivot. It’s a measured extension of what we’re already good at: turning complex business problems into code that runs in production without breaking.
Why now
Three reasons, in order of weight.
One — the models have hit engineering grade. Over the last two years, the gap between “impressive demo” and “system running for a paying customer” has closed. Open-weights models run locally. Agents operate external tools without hallucinating their way into trouble. Action chains are now genuinely testable. We’re not building on magic — we’re building on architecture you can deploy, monitor, and roll back.
Two — our clients are asking. Recurring questions like “can you add a conversational layer to our existing system?”, “can you automate the internal form our team fills in manually?”, or “we have 40,000 support tickets — can you classify them automatically?” showed up so often it’s clearly a market, not a fad.
Three — Israel’s labor reality. Israeli companies don’t want to fly in an American AI consultant who doesn’t understand the local regulatory environment, charges $500/hour, and disappears after three months. What’s missing is a partner who builds both the product and the organizational wiring around it.
What this looks like in practice
The new division offers four service lines.
1. Custom autonomous agents
Building agents that run entire processes — not chatbots answering questions, but systems that take actions. Example: an agent that receives a CV, extracts the data, classifies for the role, emails the candidate back, and opens a ticket in the ATS. Everything runs autonomously, with human-in-the-loop checkpoints in the right places.
We build our agents on OpsAgent — the internal platform we developed over the past year to run MSApps’s own business operations. Today it drives our sales pipeline, marketing, lead tracking, and continuous health checks — with no developer in the loop. What works on us works for you.
Above all of it sits SOSA — the methodology we developed for Supervised, Orchestrated, Secured Agents. It’s the framework that lets a large organization adopt autonomous agents without losing visibility into what they’re actually doing in production.
2. LLM integration into existing apps
Clients with mobile apps in production — iOS, Android, or React Native — can now embed language capabilities inside the screens they already have. Semantic search, auto-summarization, suggested-reply generation, voice assistance — each becomes a component one of our engineers drops into the app we already built for you.
A healthcare client of ours recently embedded a language-model-based medical-information assistant inside their existing app, with live answers for clinical staff. Build time: seven weeks. Model: open-weights, running on dedicated infrastructure — zero medical-data exposure to third-party clouds.
3. Internal-process automation
The automations we build target real pain points: the forms your team fills in by hand every time, the reports rebuilt repeatedly in spreadsheets, the manual classifications happening in your CRM. We identify the pain point, build a focused solution, wire it into your existing tools, and train the team.
In the last year we’ve open-sourced more than twenty plugins for Anthropic’s Claude platform — plugins we and our clients use to accelerate business automation. These are real tools running in production, not demos.
4. AI strategy consulting
Before a single line of code gets written, we sit down. What’s the business problem? What’s the expected ROI? Where are the regulatory risks? We have deep experience in complex regulated environments — health, finance, insurance — and we bring it to the first meeting.
What doesn’t change
Mobile development stays the core of the business. The team that built your apps isn’t rotating out. QA processes stay strict. Our approach to data privacy — stricter than ever, especially once language models enter the picture.
What we’ve added is another layer. Not a replacement.
How to get started
If you have a specific challenge in mind — a project stuck on manual classification, an existing app that could pick up language capabilities, or a process that feels like wasted time — let’s talk. A thirty-minute scoping call, no commitment, is enough to see if there’s a fit.
Contact us • Learn about AI & Automation • Read about OpsAgent
Michal Shatz is founder and CEO of MSApps. Fifteen years building enterprise systems, the last three years building autonomous AI infrastructure and leading SOSA, the methodology for safe enterprise agents.
WhatsApp
Call me