Insights

Lessons from the lab.

The methods, economics, and lessons behind AI that lasts in production, written by the engineers who build it, for anyone who wants to learn. Read online, or take the full paper with you.

Free tools

Try it on your own figures

Two tools to use right in your browser. Your data stays on your device.

Running Cost CalculatorThe monthly AI model bill for one task, at three model sizes, before and after prompt caching and overnight batch processing.$200 → $60a monthExample: 20,000 supplier invoices a month on a mid-size model, before and after caching and an overnight batch.Open the calculator →Security Alert CheckSix questions on how your team handles security alerts, and where AI triage would change the work.72 of 500alerts reviewed a dayExample: alerts two analysts can review in a day, at 10 minutes each, out of 500 received.Take the check →
Our own model
AppsTech Labs · Security

AppsTech Aegis-1™

Security AI that runs inside your own network.

Aegis-1 is built for one domain: your security environment, with its vulnerabilities, threats, suspicious code, and alerts. It runs entirely on your own infrastructure, so sensitive data stays on your network, and it is efficient enough to run every day at a predictable cost.

a.
White Paper

Demystifying AI Agents

What separates an AI agent from a chatbot, and what it takes to run one safely. In plain language: how agents perceive, reason, and act on their own, where they are already earning their keep in businesses large and small, and the five questions to ask before you let one touch your operations.

PDF · 6 pages · English & French
a.
White Paper

The Running Cost of AI

Once an AI system leaves the pilot and goes into daily use, its monthly bill grows with every new use. This paper explains where the running cost comes from and sets out five engineering disciplines that keep it low and predictable, with a worked example on supplier invoices.

Read onlineDownload the PDF
Web and PDF · 7 pages · English & French
a.
Research Paper

Cybersecurity Intelligence, Built From the Ground Up

The case for cybersecurity models designed for the domain from the first token: specialized, efficient enough to run on your own infrastructure, and governed for dual-use from day one.

PDF · 5 pages · English & French
Notes from the lab

Short notes from our week

What we built, what worked, and what we would do differently.

October 2026
AI for small teams

AI is within reach of small companies. SaaStr, for one, generates more than $10 million in revenue with three people and more than twenty AI systems. Here is how we use it every day at AppsTech Labs.

Get each new note by email

See our privacy notice.
Interested in Aegis-1 for your organization? Start the conversation →