An independent engineering lab · Europe and Dubai

Mannat means a wish. We engineer it into a working product.

AI-powered products, automation, security reviews and DevSecOps for startups and real-estate teams that do not have an engineering department yet. One founder, nearly twenty years of platforms, and a lab of AI agents that are always labelled as such.

Talk to the lab See what we have built

Two products live in production

Two open-source projects

Methodology published: Zenodo 2026, ICISSP 2024

Runs on its own GitOps, default-deny Kubernetes platform

Every automated action reviewed by a human before it ships

Who we work with

Two kinds of team, one thing in common: nobody whose job is the engineering.

Startups without a security team

You have a product and a deadline and nobody whose job is to worry about what happens when it gets popular. We build with you, and we review what you already have, so the first security conversation with an investor or a customer is not the one that costs you the deal.

Real-estate teams that run on spreadsheets and WhatsApp

We built a sales-execution system that a brokerage uses every day, and the automation that keeps a brokerage’s social channels alive without anyone opening a design tool. We know what a listing day looks like.

Services

Four things we do. All of them end with something you can run without us.

Turning ideas into products

You have something in mind and no engineering team to build it, or no idea how to get it in front of customers. That is the normal starting point. You bring the idea and the domain knowledge; we bring the architecture, the AI where it earns its place, and the discipline to ship a first version in weeks, not quarters. Everything is built to be handed over: your repo, your cloud, your keys.

How we build products

Automation

If a person can do it on a screen, it can almost always be automated: the report someone assembles every Monday, the leads copied between two systems, the posts nobody has time to write, the invoices that wait for a human to press send. We build the automation, put a person at the one step that needs judgement, and hand you the switch.

What we automate

Security consultation

A plain-language review of your product, your cloud and your AI features, with findings ranked by what would actually hurt. AI systems included: we red-team model-driven features, because a chatbot with a database behind it is an attack surface.

What a security review covers

DevSecOps consultation

Pipelines that block the fixable findings and let everything else through. GitOps so nobody runs commands on production by hand. Secrets that never sit in plain text. We run this platform ourselves; we set it up for you the same way.

What we set up

How the lab works

Three rules. Your project runs on the same ones.

A human approves everything that leaves.

Our own AI agents draft, review and build. Nothing they produce reaches a customer, a repository or a social feed without a person looking at it first.

Built in public.

We write up what we run, including the failures. If you want to know how we work, read the field notes before you write to us.

Nothing you cannot walk away with.

Your code in your repository, your infrastructure in your account, documented well enough that the next engineer does not need us.

Work

What the lab has built, with its status stated plainly.

All work

Command Center

Live

Sales execution for real-estate teams. In daily use at a Dubai brokerage.

Aqilra

Live

Multi-tenant publishing: authors write, admins review, readers get fast pages. Hosts our own blog.

KavachRT

In development

An autonomous AI red-team operator that ends every run with a reproducible score and a go/no-go.

Mizaan

In development

Explainable multi-agent scoring: a scorer, an auditor, a human. Methodology published on Zenodo.

Muneem Ji

Open source

Self-hosted GST billing for Indian small businesses. MIT-licensed; the invoices stay on your machine.

Social publishing pipeline

Case study

An AI drafts the post, paints the picture, and never gets to publish. Running for two organisations.

Field notes

From a security-first platform run by one person.

All field notes

2 Aug 2026 · Security · 11 min read

Part 0: Security Without a Security Team

2 Aug 2026 · Security · 13 min read

Part 1: What a WAF Monitor Taught Me About the Real Cost of a Public IP

2 Aug 2026 · Security · 14 min read

Part 2: The Morning I Realized the Door Had No Lock

About

Mannat AI Labs is an independent engineering lab based in Europe and Dubai, run by Mohammed Jawed, a platform, DevSecOps and automation engineer with nearly twenty years in IT, alongside a working set of AI agents. The agents are AI. The judgement is his.

IIT Kanpur alumnusTU Wien alumnusAMU alumnusPublished: ICISSP 2024 · Zenodo 2026

About the lab

Have an idea, a product, or a worry?

Tell us in three sentences. We reply with what we would do, what it would cost, and whether we are the right people for it.

Talk to the labNo sales team. You are writing to the engineer.