AI workflow optimization for agencies

We sell Workflow Optimization Cycles. One cycle covers one workflow and one bottleneck, in a fixed timebox with a fixed number of sessions, at a fixed price. Half is paid to book the slot and half at the second session.

After a completed cycle you can add the AI Workflow Support retainer: a flat monthly fee, month to month, paid in advance. It is a continuation, not a starting point, so it is only offered to teams that have finished a cycle.

Each review ends with the next recommended bottleneck. That is how a second cycle gets scoped, and it is your call whether to book it.

For: small and mid-sized creative agencies, studios, marketing and content teams in Lebanon and the Gulf that already produce with ChatGPT, custom GPTs, Claude or Higgsfield and want one process to get measurably faster and more consistent. Typical pain: too many generations to reach an acceptable asset, re-prompting the same brief, credits spent on output nobody keeps, brand fit that depends on who is at the keyboard.

Not for: teams that have not started with AI yet, which usually begin with the AI Program; or companies that need an assistant or automation built into their product, which is our AI and automation line.

1. Booking. The down payment confirms the slot. You receive a short intake form the same day and start logging time and iterations on your next three jobs.
2. Discovery, one to two hours, in person. We map the workflow, pick the bottleneck and agree two or three KPIs in the last twenty minutes.
3. Cycle brief, next day. One page: bottleneck, baseline numbers, KPIs, deliverables and dates. You confirm by reply. This is the contract for the cycle.
4. Design week, off-call. We redesign the workflow and build the prompt library, the Claude skill or custom GPT, and the Higgsfield presets and references it needs.
5. Session one, two hours, in person. We walk through the new workflow and run one real task end to end with your team.
6. Session two, two hours, in person or online. Your team runs tasks themselves. We handle edge cases and hand over the templates. The balance is due here.
7. Two weeks of use. The measurement window. The team logs each job the same way it did before the cycle.
8. Review, one hour. Before and after on the agreed KPIs, the next recommended bottleneck, and the retainer offer if it makes sense.

Your team spends about six to seven hours with us. MoonDev spends ten to fifteen, most of it in the design week.

Everything produced in the cycle is yours to keep, retainer or not.

KPIs are agreed at the end of discovery and confirmed in writing in the cycle brief. We choose from this list so there are no surprises:

• Time per asset. Minutes from brief to approved output, median over three to five jobs of the same type.
• Iterations to acceptable. Generations run until the one you kept, from the tool's history.
• Credits per approved asset. Credits spent divided by assets delivered. Objective, and it is money.
• Time on repetitive tasks. Minutes per day on prep, resizing and re-prompting, logged for a week.
• Output consistency. First-batch acceptance rate, plus a brand-fit rating by the same reviewer.
• Labour cost saved. Hours saved multiplied by your own loaded hourly rate.

Rules: same asset type before and after, medians not averages, logged at the time rather than from memory, and the result is called indicative because five jobs is a small sample.

• Bounded. One workflow and one bottleneck per cycle. A new bottleneck is a new cycle, not scope creep.
• Written down. The cycle brief is the contract. If it is not in the brief, it is not in the cycle.
• Your tools. We work inside the subscriptions you already have and do not resell software.
• Real jobs. Sessions use your live briefs, not demo prompts, so the team leaves having done the work once already.
• Honest numbers. We guarantee deliverables and a fair measurement, never a percentage.
• In person where it counts. Discovery and the first session happen at your office in Lebanon. Teams elsewhere run the same cycle over video.

Included: a WhatsApp support line with a stated response time, a monthly cap on guidance hours, one monthly review call, updates on relevant tools and models, and new prompts, skills and resources added to a shared library.

Not included: custom development, major redesigns, or a new bottleneck. Those are a cycle, or an AI and automation project such as the engine behind Questa.

• Team size. How many people attend the sessions and need accounts, presets or templates set up.
• Tool setup. A prompt library alone versus a Claude skill, custom GPT and Higgsfield presets together.
• Format. In person in Lebanon versus online for the Gulf and elsewhere.
• Continuation. A single cycle, a run of cycles, or the monthly retainer afterwards.

The cycle price is fixed and quoted on the discovery call. Start from contact.

Training is stage three of the cycle, not the whole product. A workshop on its own shows the team what the tools can do and changes little the following week. In a cycle we first map one workflow, redesign it, then train the team on that redesigned workflow using their own jobs, and measure whether it got faster.

Small and mid-sized creative agencies, studios and marketing or content teams that already produce with AI tools such as ChatGPT, custom GPTs, Claude or Higgsfield and feel the friction: too many iterations, inconsistent output, credits burned, hours lost on prep. In-house teams with the same pattern fit too. It is not for teams that have not started using AI at all; those usually begin with our AI Program.

Four to five weeks end to end. Your team spends about six to seven hours with us: a discovery meeting, two working sessions and a one-hour review. MoonDev spends ten to fifteen hours in total, most of it off-call designing the new workflow, the prompt library and the tool setup.

A cycle is a fixed price, quoted in USD, with half paid to book the slot and half at the second session. The optional support retainer is a flat monthly fee, month to month, paid in advance, and only offered after a completed cycle. The price depends on team size, how many people attend the sessions, which tools need setting up and whether sessions are in person or online. We confirm the number on the discovery call.

We guarantee the deliverables and an honest measurement against your own baseline. We do not guarantee a percentage. Success for a cycle means measurable improvement on the two or three KPIs we agree at discovery, logged the same way before and after, on the same asset type. Five jobs is a small sample, so we call the result indicative and say so in the report.

The ones your team already pays for. In practice that is ChatGPT and custom GPTs, Claude and Claude skills, and Higgsfield for image and video generation. We do not resell tools or require new subscriptions. If a tool change would remove the bottleneck, we say so in the cycle brief and you decide.

The review names the next bottleneck worth fixing. From there you can book another cycle, move to the support retainer for questions and monthly check-ins, or stop. The redesigned workflow, prompt library, templates and recordings are yours either way.