Prompt design
Setting the goal, context, constraints, examples and output format, then iterating and evaluating. A clear request saves hours of rework.
A Guide to Writing a Clear PromptApplied AI
From writing a prompt to building a platform.
AI isn't one tool that does everything; it's a set of skills that runs through the whole job. These are the six areas I work and train in, and how they connect.
Image production
Visual briefs, composition, consistency with the brand, working from real product photos, and reviewing output before use.
Visual Brief Template for Image ProductionVideo production
Idea, then script and storyboard, then short shots with defined camera movement, then assembly and review.
AI Video Planning TemplateAI-assisted coding
Clear requirements, architecture, iteration, debugging, testing and security review — generated code needs someone reviewing it.
Checklist for Reviewing AI-Generated CodeWebsites and platforms
User journeys, interfaces, data, integrations, deployment and improvement after launch.
Website Requirements Workbook Before Building with AIBusiness workflows
Spotting repetitive tasks, choosing a suitable tool, defining human review, documenting the method and measuring real usefulness.
AI Use Map for Your CompanyFrom a written idea… to something that works
Four stages, with real examples from each.
Examples come from separate pieces of work, not stages of one project.
Objective: I will use the result to [the next decision or action]. Audience: the output is for [who, what they know, what they care about]. You are [role: specific expertise tied to the task]. Context: [two or three sentences: who we are, the project, where things stand, what has been tried]. Inputs: [paste the text, data or notes here after removing sensitive data] Rely only on the inputs above. If something is missing, write: not stated.
A clear request saves hours of rework
A template from the "Clear prompt guide": goal, context, constraints, format and a quality bar. It is how I start any AI work, and how I teach it.
Use the prompt builderAI-produced video · 4 seconds
A scene built on a precise description
A historical scene I produced with AI: an old cannon at sunset. A description of place, costume, light and camera movement, then a review before use.
Image and video programme
Real platforms, from idea to launch
From Matjar and Masarak: a merchant dashboard that points to the next setup step, and a student experience on mobile.
Product and platform buildingPassion OS · internal automation
- Every dayDaily content tasks for each monthly account are created automatically for the person responsible.
- Every monthMonthly billing is prepared for sales, with printable invoices.
- On every changeA notification reaches only the phone of the person concerned.
- Every nightA backup of the working data.
Repeating work? Let the system handle it
From Passion OS: automations running on the server that took over what used to be followed up by hand inside the agency.
Workflow & automation sessionHow the areas connect
A fictional example of a small business launching a new service: each stage hands off to the next, with human review at every step.
Illustrative example Daw photo studio — a new service: product shoots for small online shops.
- 1Brief
Goal: 20 bookings in the first month. Audience: small online shops in the north. Offer: a fixed-price session with 15 edited images.
- 2Copy
A five-part prompt produces three versions of the page headline and service description; one is chosen after review.
- 3Visual concept
One image brief with a fixed style sentence: light background, side window light, a real product from the studio as reference.
- 4Storyboard
A 20-second video in three shots: before, during the session, final images on a phone screen.
- 5Landing-page spec
Headline, three benefits, examples, a three-field booking form and a contact button. Acceptance: works on mobile, requests land in a sheet.
- 6Review checklist
Do the images match what the studio actually delivers? Is the Arabic text in the design correct? Did a test booking reach the sheet?
This is an illustrative example to explain the method, not real client work.
Review is part of quality
A model predicts the most likely text; it doesn't "know" the truth. So every number and reference gets checked, every image is compared with the real product, and all code goes through testing and a review of permissions and secrets. These aren't warnings — they're the difference between a nice demo and work you can rely on.