ManaOS is the marketing operating system your brand owns. It runs a campaign from cultural signal to live activation on your instance, and your team signs every gate. Every decision, result and lesson stays with you.
The problem
The answer
How it runs
They run together as one system, and nothing moves from one to the next until a person signs it off. Athena is also available on her own, and the rest are being built to stand alone.


Athena understands.
Cultural intelligence
Reads the culture and the customer before anyone decides anything.
Gate 01 · you sign the read before strategy begins
Meet Athena →


Apollo plans.
Strategy and planning
One clear plan from a thousand signals, argued from Athena’s evidence.
Gate 02 · you sign the plan; its budgets are then locked
Meet Apollo →


Hephaestus makes.
Creative production
Crafts the work to the signed plan, at the speed of the calendar.
Gate 03 · you approve the work, per asset, per market
Meet Hephaestus →


Artemis delivers.
Activation and optimisation
Hits the target the plan defined. Nothing fires without a human.
Gate 04 · launch and every optimisation are yours to confirm
Meet Artemis →
Fig. 01 · the four products and the gates between them. Fourteen configurable agents sit behind them.
START SMALLER
Not ready for the whole system? Many clients begin with Athena on her own: category intelligence, sourced reports and the data supply, no other product attached. Everything she learns comes with you if you adopt the rest later.
Hermes is the interface under every product. It carries each question, instruction and confirmation between you and the system, in plain language. You choose the door. The rules do not change.
The full working surface: every product, block and plan, seen and edited directly. Where the team lives day to day and where the gates are signed.
ShippedAsk, instruct and confirm in conversation, grounded in the same plan on the same page. Consequential changes come back as proposals to confirm, never silent writes.
ShippedYour brand’s agent queries campaign insights, strategy and performance from the chat tool your AI policy already approved. Same governed view, same guards, different doorway.
Available and expandingYour agencies work inside your ManaOS, with permissions you set. Their strategy, plans and results land on your instance and stay there when the agency changes.
On the roadmapFig. 02 · whichever door a request comes through, reads are open and grounded in the live plan; writes arrive as proposals and land in the same ledger.
Mnemosyne is the brand memory under every product. Each campaign, decision and result is kept there: strategy argues from it, creative is shaped by it, the next plan starts smarter because of it.
Memory lives on your instance, isolated per brand at the database layer. It is never pooled into a network data lake, never used to train shared models, and never visible to another brand. When an agency or a person moves on, the memory stays. Agency access to your instance is on the roadmap, so their work will land there too. That is the opposite of the holdco model, where your performance history sharpens someone else’s next pitch.
Most AI tools ask you to trust the answer. ManaOS is built so you never have to. These are design decisions, not policy promises.
Every campaign lives as a structured plan: brief, budget, audience, media, content, creators. The AI fills it in and suggests changes. It does not own it and cannot rewrite it behind your back.
The same signed-off brief produces the same budget and plan every time. Change one number by hand and the system respects it. Nothing is silently reinterpreted.
Move a market, cut a budget, swap an audience: the system flags what is now out of date downstream and re-derives it. No stale decks that look finished.
A ledger records who or what changed each field, when, and from what, with one-click revert. Human edits are remembered, not overwritten by the next regeneration.
Budgets flow from the signed brief through locked stages. The AI cannot invent or move a number. Optimisation arrives as a proposal; a person sets it live.
Insights carry linked, checkable sources. Outputs are verified against the plan before they ship. What goes upstairs can be audited line by line.
One system, three ways to run it. Your security team chooses the perimeter; your marketing team gets the same OS. Built to pass your AI policy, not to route around it.
In every deployment: SSO, role-based permissions, row-level isolation per brand and market, your choice of model provider, and no training on your data.
No. The campaign lives as a structured plan the system owns end to end; the AI proposes into it and people sign it. A chat window gives you fragments and forgets you on Monday. ManaOS carries strategy to activation, and remembers.
The four products inside ManaOS. Athena handles research, Apollo strategy and planning, Hephaestus creative production, and Artemis activation. They run as one system with a sign-off gate between each, and Athena is also available on her own. Hermes is the interface and Mnemosyne is the brand memory they all share. Your team keeps taste and judgment; the products supply the machinery underneath.
It was designed for it. Deploy in our cloud, your VPC or on your own infrastructure; choose the model provider your policy approves or bring your own endpoints; nothing trains on your data; and every AI write is gated, logged and reversible. Your security team gets architecture, not assurances.
Yes. ManaOS is also available over MCP, so the agent your company already uses can query campaign insights, strategy and performance directly. Reads are open and grounded in the live plan; writes come back as proposals for a person to confirm, exactly as they do inside ManaOS.
Neither. It replaces the machinery: the research grind, the planning spreadsheets, the setup work, the reporting scramble. Judgment, taste and the relationships stay with your people, which is why the pilot has your team driving from day one. Keep the agencies that earn their place. Agency access with permissions you set is on the roadmap, so their thinking and results will sit on your instance and stay when the agency changes.
On your instance, isolated per brand at the database layer, in the region or perimeter you choose. Brand memory compounds there and leaves with you. It is never pooled across customers, never sold, and never used to train shared models.
Run it through ManaOS beside your current process for two weeks, your team driving, us alongside. At the end you have two artefacts to compare and an easy internal story either way.