Article · Substack · October 7, 2026
Why AI keeps giving you generic results (and the 3 decisions that fix it)
Explaining the skill that makes AI worth using
David Kotler · AI & automation engineer
I build with AI. I explain how it actually works.
Projects, articles and solutions from the field on models, agents, automation and integrations - with technical depth, no noise and no magic promises.
Areas of work
- AI Agents
- Automation
- MCP & Integrations
- AI in Production
Working on now
- EssentialAppA daily digest that reads 131 sources across AI, business, management, marketing and engineering, and sends you the few pieces worth your time, in one short email.
- Eligibility Check for Computing, Digitization, and AI Grant (CEO Directive 4.56)SkillA Skill for Claude and ChatGPT that determines if a small or medium-sized business in Israel meets the threshold conditions for the Ministry of Economy's computing-digitization-AI grant (35% up to 350,000 NIS), verifies if a specific expense is recognized, and outlines submission requirements — including a live status check against gov.il, and an eligibility report and submission checklist in Hebrew.
- iCount MCPOpen sourceAn MCP server that allows AI agents to open, search, and close iCount documents — invoices, receipts, orders, and quotes — and manage customer records directly from a conversation. Runs locally on stdio, with no storage and no OAuth — just a single API token.
What's new
What shipped lately
Video · October 7, 2026
A look at this morning's Essential digest
Five pieces, about 33 minutes, picked for my role. Each pick gives the point in one sentence, the key facts, a quote and one-tap feedback.
Video · October 7, 2026
Essential's whole architecture, mapped from its code
I pointed Archify, an open-source skill, at the Essential repo. It drew the system overview, the daily digest data flow and the Paddle checkout sequence, and every box links back to the functions behind it.
The latest article
What's changed in AI - and what it means for whoever builds with it
Explained from the inside · September 12, 2026 · 3 min read
Most businesses don't know if their AI is working. Here is how to actually check
Why the question "Is the AI giving us a return on investment?" is almost never answered correctly, and exactly what needs to be measured to know the answer.
Read the articleExplained from the inside · September 10, 2026 · 3 min read
7 steps to guide your team on AI — and why "just start using it" isn't training
Most companies invest in AI without a real training process for the team. Here are seven concrete steps to turn casual AI use into actual organizational capability — and why it starts with management, not the employees.
Read the articleExplained from the inside · September 10, 2026 · 4 min read
When adding a second AI agent to the business, who coordinates between them?
Why adding more AI agents to a business doesn't necessarily save more time — and how to build a process where someone is responsible for the coordination between them from day one.
Read the article
Where to start
The same AI. Different questions.
For businesses
Turn a repetitive process into a system that works
Diagnosis, planning and implementation of AI solutions inside your existing work - including the part most projects miss: accountability, approvals and team adoption.
For developers
Build AI that stays reliable outside the demo
MCP, agents, APIs, observability, permissions and approval flows - documentation and lessons from real projects.
Built it myself
Essential
A daily digest that reads 131 sources across AI, business, management, marketing and engineering, and sends you the few pieces worth your time, in one short email.
- Read tools versus write actions
- Approval before a sensitive action
- Design that fits an existing Israeli system
Active
What I write about
Three angles, one subject
Built it myself
Projects, experiments, engineering decisions and what didn't work.
Read the articlesExplained from the inside
Accessible explanations of models, agents, MCP, security and production.
Read the articlesData + an engineer's angle
News and research, with the practical question: what actually changes for whoever builds or buys AI.
Read the articles
AI and automation engineer
I'm David Kotler. I build AI systems, then break them down into an explanation you can actually use.
I'm a software engineer working on AI, automation and integrations. Lea AI is where I share what I learn from the building itself — and also help businesses turn a technical capability into a process that works day to day.