🧩
Augment, don’t replace
Your teams keep working in Excel, WhatsApp, Outlook and the ERP they already know. We build the intelligence layer on top. No migrations, no “new system” your floor will reject.
AI for Manufacturing
AI Systems for Factories
Your factory runs on Excel, WhatsApp groups, Outlook threads, and thirty years of know-how in people’s heads. We build the AI layer on top of exactly that — without changing how your teams work.
From a one-day mapping to full implementation: quoting, tenders, project management, field service, design, and one organizational brain that connects it all.
Book a meetingWHY FACTORIES
The biggest AI upside is on the factory floor
Tech companies fight over the last 5% of productivity. Manufacturers are sitting on the first 50%. A typical mid-size factory we meet runs a large and growing order book on spreadsheets, notebooks, and email — 60–70 parallel projects in one Excel, field updates buried in dozens of WhatsApp groups, quotes that take days while competitors answer in hours, and decades of engineering knowledge that lives only in veterans’ heads.
None of that is a weakness. It means the processes already work — they’re just starved of leverage. AI is the first technology that can be layered on top of this reality without forcing a painful ERP migration or a new tool nobody on the floor will open. That’s exactly what we build.
None of that is a weakness. It means the processes already work — they’re just starved of leverage. AI is the first technology that can be layered on top of this reality without forcing a painful ERP migration or a new tool nobody on the floor will open. That’s exactly what we build.
HOW WE THINK
Four rules for AI that survives contact with a factory
⚡
A first win you feel
Every engagement is built around a first working tool your people use within weeks, not a slide deck. Trust is earned with a system that works, then we expand area by area.
📉
Measured by the bottom line
Every solution is scored by hours saved, orders unblocked, and mistakes prevented. Not by tokens, demos, or hype.
🔒
Human in the loop
AI drafts, sorts, and surfaces. Your people approve. Nothing goes to a customer, a supplier, or the production floor without a human decision where it matters.
THE PROCESS
From mapping to full implementation
Every engagement starts small, proves itself, and grows in stages. No big-bang projects, no year-one invoices for year-three value.
01
Mapping Day
One focused day on site. A deep session with management, then structured interviews across departments: engineering, project management, sales, procurement, service. We map the real workflows — not the official ones — including the tools, the spreadsheets, and the security constraints.
Output: a mapping document with your full process map, a ranked list of AI opportunities by value and speed, the low-hanging fruit, a roadmap, and a project cost estimate.
02
First win
We pick the highest-value, fastest opportunity from the map and build it first. A working tool in the hands of one team, on real data, within weeks. Good enough to cover 80–90% of the need on day one.
Output: a live system your people actually use — and the proof that this works in your factory, not just in theory.
03
Staged build
We expand department by department, following the roadmap. Each stage ships a complete, working system before the next begins. Budgets are staged the same way — you never pay for a big bang.
Output: an organizational brain that grows one proven piece at a time.
04
Training & embedding
Frontal, on-site training for every team that touches a system — shoulder to shoulder, on their real work. Plus long-term support so the systems keep improving after we leave.
Output: a factory that runs the systems itself, with people who know why they work.
THE CATALOG
What we build for factories
Every factory gets its own map — but these are the systems that come up again and again, department by department. Each one is a working solution we can build over the tools you already use.
📐
ENGINEERING & DESIGN
AI in the design room
Where products are born is where AI compounds fastest — and where almost no factory has it.
CAD & 3D modeling copilots
We connect your designers and engineers directly to AI through their existing tools — Rhino, 3D modeling environments, pattern and technical design software. The AI helps specify, model, and iterate on products at the prototype level: from a customer requirement to a first 3D concept in a fraction of the time.
Product specification assistant
An agent that takes a customer request — “we need part X” — and drafts the technical specification: materials from your own database, tolerances, precedents from past projects, open questions for the customer. Engineers review and refine instead of starting from a blank page.
Knowledge base of past work
Decades of drawings, specs, and solved problems usually live in veteran employees’ heads and in folders nobody searches. We turn that into a searchable engineering brain: “have we ever built something like this?” gets answered in seconds, with the files attached.
🎯
SALES, TENDERS & QUOTING
From inquiry to quote, faster than your competitors
In manufacturing, the first credible quote usually wins. AI collapses days of quoting work into hours.
Tender & RFQ response automation
An agent that scans incoming tenders and RFQs, extracts the technical requirements, matches them against your materials and solutions database, and drafts the technical response. Your team reviews a 90% draft instead of assembling one from scratch.
Quote generation with real production logic
For machining and metalwork: an assistant that helps run the quoting chain — time studies, machine selection, tooling, materials — and produces a structured quote draft grounded in how your floor actually works.
Customer memory
Every email thread, meeting, and past order for each customer, summarized and ready before a call. Your salespeople walk in knowing the full history — even for accounts they inherited.
⚙️
OPERATIONS & PROJECT MANAGEMENT
One view of every project
Factories we meet run 40–70 parallel projects out of one Excel and one very tired project manager. The fix is not a new PM tool — it’s intelligence over the Excel you have.
Project dashboard over your existing files
A live dashboard that reads your existing project Excel, your inbox, and your folders — and shows every project’s status, the tasks that are blocking, and a near-term work plan. One evening summary tells management what moved today and what is stuck.
ERP & supplier paperwork automation
Purchase orders, supplier confirmations, delivery notes, invoices — the paperwork that eats your back office. AI reads the documents, cross-checks them against the ERP, drafts the responses, and flags mismatches for a human decision.
Daily morning briefing
Every morning at 9:00, each manager gets a digest of their projects: what arrived in email overnight, which threads relate to which project, what needs an answer today. Nothing falls between the cracks because nobody had time to scroll.
📱
FIELD SERVICE & WHATSAPP
Intelligence over the WhatsApp chaos
Field teams live in WhatsApp groups — one per project, full of free text and photos. We don’t fight that. We harvest it.
WhatsApp group intelligence
Structured form links inside the same groups your technicians already use: they tap a link, upload photos, mark visit purpose and status. The data flows into the organizational system automatically — photos treated as first-class data, not lost scroll.
Daily group scanning & digest
An agent sweeps all project groups every morning, detects what’s new, and produces a service digest per project: open issues, waiting-on-parts, escalations. Management sees the field without reading 40 groups.
Service report automation
From the messages and photos a technician already sends, the system drafts the formal service report and files it against the right machine and customer. The technician confirms instead of typing.
📧
EMAIL & OFFICE
An inbox that sorts itself
Applications teams, sales, and procurement run entire product journeys inside Outlook. AI turns that thread-jungle into structured work.
Outlook triage by project
Every incoming email is read, classified, and attached to the right project or customer. “Which threads relate to this order?” becomes a click, not an archaeology dig.
Drafting with your house style
Replies to routine requests — status updates, document requests, standard confirmations — drafted automatically in your tone and language (including Hebrew, English, and mixed threads), waiting for one-click approval.
🧠
THE ORGANIZATIONAL BRAIN
Where it all converges
Each solution above is useful alone. Together, they feed one system — a brain that serves the whole organization, not one person.
Cross-department knowledge system
WhatsApp data, email threads, ERP records, project files, and engineering knowledge feeding one connected system. Ask it a question in plain language — “what’s blocking the delivery for customer X?” — and get an answer with sources.
Management visibility
A single management view: every project, every department, status, blockers, and a this-week plan. Plus one evening summary that replaces the “what happened today?” round of phone calls.
Prototype rescue
Someone in your company already built a helpful tool — in Base44, Lovable, Power Automate, or a personal dashboard? That’s a signal, not a problem. We take these self-built prototypes and industrialize them into secure, organization-grade systems. Starting the mapping yourselves is half the way to a product.
SECURITY & GOVERNANCE
Built for factories that can’t afford leaks
Your cloud or on-premises
Systems run in your environment — your Microsoft 365 tenant, your cloud account, or a local on-prem server when required. We also work with local open-source models for air-gapped or classified contexts.
No training on your data
We work with enterprise-grade AI configurations where your data is not used to train anyone’s models. Data residency and retention are defined in writing before anything is built.
Defense-supplier ready
We work with defense-sector manufacturers. Security constraints, human-in-the-loop checkpoints, and customer commitments are mapped on the Mapping Day and baked into every solution — third-party penetration testing included where needed.
SUCCESS STORIES
Built, shipped, and working in production
Defense Manufacturer Tactical Gear · Israel
WHAT WE BUILT
A full AI transformation at a veteran Israeli defense manufacturer (three decades of production, supplier to the IDF and NATO countries) that runs a large order book on Excel and paperwork. The engagement runs from ERP and supplier paperwork automation, through process mapping across engineering, project management, and tenders, up to connecting the design team to AI through a 3D-modeling MCP — AI that helps specify and build products at the prototype level.
THE OUTCOME
A staged transformation, starting with a one-day on-site mapping across all departments and a ranked roadmap — with designers, project managers, and the tender team each getting AI built into the tools they already use.
Ginger Insurance Insurtech
WHAT WE BUILT
An AI agent for end customers that doesn’t just answer questions but takes real actions: it sends approval requests by email to both the customer and the company, and books live calls with case managers — with an eval mechanism and an isolated, secure embed inside Ginger’s system.
THE OUTCOME
The agent is live and performs real actions in production, including opening requests and booking meetings.
“Wow. This is amazing. This is amazing.”
— Oshrat Ben Moshe, CEO and Founder, Ginger
JVP Venture Capital
WHAT WE BUILT
An agent for the fund’s IR team that automatically produces a full briefing document on an investor: it pulls from the CRM, the web, and LinkedIn, cross-references everything, attaches quotes and sources, and outputs a ready document in the fund’s own format.
THE OUTCOME
A report that took hours of manual work is now built in about five minutes, with factual accuracy of 95% and above. Presented to the fund’s partners and green-lit for expansion.
“this to me feels like you’ve figured it out. It looks great.”
— Lauren Kiel, Investor Relations, JVP
iPracticom Telecom
WHAT WE BUILT
A “marketing brain” built on four connected skills (creative → content → scheduling → monitoring) over a wiki architecture, designed to replace an external marketing agency. The system produces live posts and graphics.
THE OUTCOME
The system works in production, and the CEO asked to expand it into a company-wide organizational brain serving the support team as well.
COMPANIES I'VE WORKED AND WORK WITH
Diesenhaus 
Diviners IPRACTICOM Spark Founder Lab WHO'S BEHIND THIS
Yuval Keshtcher
AI Consultant & Implementer · with a senior engineering team
I lead company-wide AI transformations — business mapping, identifying the highest-value opportunities, and building the AI layer on top of how your teams already work. Behind me is a senior full-stack engineering team specialized in building and embedding enterprise-grade AI agent systems, with experience across manufacturing, defense, telecom, insurance, and venture capital. A decade ago I founded UX Writing Hub, an international school with 1,000+ graduates, built and run entirely on automations — today I build those systems for organizations.
You have a factory. We have the know-how.
One short call to figure out where AI would actually move your bottom line — and whether a one-day mapping makes sense for you. No commitment.
Book a meeting