A team of AI agents that runs the operational half of social media for an agency — and stops at your approval.
Built for agencies in Egypt and the wider MENA region
A small demo in the real product's language, running one client's month. It walks itself through the screens — take it over whenever you like.
Konecta · March
Approval rate
94%
Active agent tasks
7
Total clients
15
Pending reviews
1
Orchestratorcoordinator
Reads the request, loads the client's context, plans the steps and hands each one to the right specialist.
Trends
Scrapes and scores social posts per client, then clusters them into themes with the evidence attached.
Creative
Writes the copy and produces the visuals for every slot, then checks its own output before submitting it.
Media buyer
Builds the media plan and reads back performance exports. It advises — a human still launches.
Moshoyour interface
The chat layer over the whole team. Ask for work, check what's pending, approve things — in plain language.
Overview: One account, every client. The five agents below run the repeatable half of the month and stop at your queue.
Your agency doesn't need more people. It needs its operational layer back.
The judgement calls, the pitch, the taste. Nothing here is automated.
The repeatable layer that grows with every client you sign.
Workarounds that only existed because that layer had no owner.
The same loop your team runs every month — research, write, design, approve — with the manual parts handed to the agents.
Sound is off until you press play. With the player focused: space plays, arrows seek and set volume, M mutes, F goes fullscreen.
The same loop runs for every client. You join it twice — at the brief, and at the approval. The four steps between them are the ones that used to take your week.
Swipe the rail, or tab into it and use the arrow keys.
Agencies have all been burned by generic output. These are the three reasons this work survives a client review — and one caption that shows it.
Konecta
Ramadan campaign · slot 3
Egyptian dialect, brand-safe vocabulary and calendar awareness are built in — the reason global copy tools keep failing this market.
Each theme carries the real posts it was clustered from. If a trend can't be evidenced, it doesn't ship.
Approval gates sit in front of every plan and every creative. What you reject shapes what comes next.
Profiles, campaigns and history are stored per client and loaded only for the job at hand.
Nothing on this page is a mockup. Agencies who join early get to reorder the second column.
Tell us which one blocks you and it moves up.
Every plan runs the same loop on the same product. What moves between them is how many clients you can carry, how much the agents can generate, and how many of your people are on the account.
5 clients
2,500 credits · 3 seats · Spark and Momentum · $8 a seat beyond three
$50a month
Book the walkthrough15 clients
7,500 credits · 10 seats · all three modes · $7 a seat beyond ten
$100a month
Book the walkthrough50 clients
25,000 credits · 25 seats · all three modes · $6 a seat beyond 25
$200a month
Book the walkthroughBeyond 50 clients
Credits, seats and the client ceiling set against the roster you actually run.
Talk to us
Ask for a custom planCredits are what the agents spend as they work — a caption costs little, a set of visuals costs more. If a month runs long you top up without changing plan: 500 credits for $19, up to 50,000 for $899.
Spark, Momentum and Apex are the model tiers your team switches between. Spark is the quick one, Momentum is the default, Apex renders visuals at 2048px. Starter carries the first two; Growth and Scale carry all three.
Monthly, cancel at the end of a cycle, prices before tax. One live session per person on every plan — seats are people, not devices.
We don't send output and ask you to trust it. Book one 20-minute walkthrough, bring the people who'd use it, and watch the whole system run on your kind of work. You already know what it costs before it starts.
One call, end to end. We open the real product and run your kind of work through it — your industries, your languages, your kind of client — screen by screen: how a brief is written, how a trend is picked, how a post is generated and corrected. Bring whoever would actually use it. Not a deck, and not output we prepared the night before.
The prices are published above, so there is nothing to negotiate at this step — you pick the tier that fits the number of clients you run, or ask for a custom one. We say plainly which part of your month we'd take on first.
You are not asked to commit until after the walkthrough, and nothing reaches a client feed at any point in it.
From there, you run the month — brand profiles loaded, your team owns the loop, and we stay reachable for whatever is still new.
Prefer email? Write to info@mash.solutions
That's the first thing to judge, so the walkthrough shows it being written live rather than pasted from a deck. The Creative agent writes from the brand profile it is given — voice, tone, forbidden words, past approved posts — and the trends it builds on are clustered from real posts that ran on Instagram, TikTok and Facebook, not invented by a model. Nothing leaves the system without a human approving it first.
It writes Arabic natively, in Egyptian dialect, and English when the brand needs it — never translation. Local calendar moments such as Ramadan are built into the content guidelines. This is the main gap between MASH | Creative and global copy tools like Jasper or Copy.ai.
Not yet. Today you approve the finished posts and publish them, and automatic publishing to Meta is the next thing being built. Everything up to that point — trends, plans, copy, visuals, approvals, reports — is live and in use.
It advises on paid, it doesn't spend. The media buyer agent writes a launch-ready plan — budget split, ad-set structure, cold and retargeting audiences, creative angles, KPIs and UTM notes, plus compliant Google search ads when search is in the mix. A human buyer reviews it and launches in Ads Manager. Upload the performance export afterwards and it reads back what to scale, cut or reallocate.
Each client is isolated. Brand profiles, campaign history and approval logs are stored per client and loaded only for the job being worked on, so nothing bleeds between the brands on your account, and nothing is used to train public models. Your agency owns the account and the data in it.
Nothing before the walkthrough — that runs on our setup. Once you're on a plan, the first client you load needs a brand profile: voice, tone, audience, guidelines and a handful of posts you're happy with. That takes about an hour to fill in properly, and the quality of everything downstream depends on it. From there you get trends and a week of content back within a couple of days.
Three published plans, monthly: Starter at $50 (EGP 2,999) for up to 5 clients and 3 seats, Growth at $100 (EGP 4,999) for 15 clients and 10 seats, and Scale at $200 (EGP 9,999) for 50 clients and 25 seats. Beyond that it's a custom plan. Each plan includes a monthly pool of credits the agents spend as they work, and you can top up mid-cycle from $19 without changing tier.
Count the clients you'd actually put through it, not the ones on your books. Most agencies land on Growth: 15 clients, 10 seats, and all three model tiers including Apex, which Starter doesn't carry. If you're testing with two or three brands, Starter is the honest answer and moving up later is a change of plan, not a migration.