Your busiest work, done by an AI that isn't allowed to make costly mistakes.

What we automate

If your team does it the same way every day, an agent can do it.

We connect to the tools you already use. No API? We build the connection ourselves, so even an old CRM or web system can be automated.

Proof

LiveOps: an operations agent that ran in production for months.

Our founder, Q, built LiveOps while working live operations for an international airport-transfer company (under NDA, so names and data are blurred). It worked inside the company's CRM, which had no API for this, and tagged and priced bookings across dozens of cities.

Results

• 50–300 bookings a day, depending on the season.
• 52% of the team's tags in high season. The other 6–7 people together did 48%.
• Zero wrong actions over its last 4 months in production, after the initial build-and-training phase. Not because the AI never errs, but because the software won't let an error through.
• About 10% escalated (5–30 cases a day), each for a real reason.
• It learns from every escalation. One plain instruction ("from now on, close any assigned transfer") is written by the AI into its own instructions and examples.
• Supervised from a phone through a paired Android app: see every case, pause the agent, or step in.

Next

Ops Agent Core: the agent and the CRM, built together

LiveOps had to work around a CRM it didn't control. Ops Agent Core is the next step: a WhatsApp customer-service agent with its own CRM, built for a transfer business. The AI talks; the CRM decides prices, payment links, bookings and driver dispatch. A customer who tricks the AI still can't get a discount, a fake payment link or someone else's booking.

It includes a flight-delay flow where the customer's "yes" is checked by the CRM, not the AI, and a live console for staff. Every action goes through an API, so it plugs into your other systems.

Status: working demo, all automated tests passing; first live run this month.

Why it's safe

The AI talks. The system decides.

An AI on its own will eventually make a mistake. So the limits don't live in the AI's instructions. They live in the software around it, where it can't talk its way past them.

1. Hard limits outside the AI.It never lowers a value below the current one, never raises it above a ceiling, never overrides a person.
2. Unsure means ask, not guess.A broken or unclear action is rejected, nothing runs, and the case goes to a person with the reason.
3. Facts come from your system.Prices, links and booking details are filled in by your CRM, never typed by the AI.
4. Everything is logged.Every action has a one-line reason you can read, so you can audit it any time.
5. You stay in control.Pause it, override it, or change a rule from your phone.

How we work

Start small, prove it, then grow.

Who we are

A small team of software engineers who have done the work our AI agents now automate.

Qmakes is led by Yasser Gad (Q), an AI and automation engineer: computer engineering, years in IT, sales and customer support, then live operations, where he built LiveOps to do his own job better. We've been on both sides: the people doing the repetitive work, and the people automating it. We build with our own team of AI agents, and a person on our side is accountable for everything that ships. English and Arabic.