Adam · The assistant inside Adamiro
Adam does the clicking.
Ask for what you want in the dashboard. Adam works out the steps, shows them to you before it starts, then drives the same buttons you would. It asks before anything costs a task. It cannot publish.
Included on every plan. Adam itself costs nothing to run.
What it is
An assistant that operates the product, not a chat window bolted onto it.
You type what you want in plain language. Adam writes back a numbered plan, then carries it out by working the same controls you would: opening a signal, generating a draft in Voice, sending that draft to Reach, setting a time. Nothing costs a task until you say yes, and Stop always works.
It has no publish button, no delete, and no access to billing. Ask it how something works and it answers that too, from the product guide, scoped to what your plan includes.
The plan comes first. Then you decide.
In practice
What you can ask it for.
“Turn this signal into a LinkedIn post and schedule it for Tuesday morning.”
A draft in your voice, sitting in the Reach queue, scheduled for Tuesday. After you have confirmed the cost and the time.
“Add an image to this draft.”
It opens the real image generator and writes the prompt from what the draft actually says. You review it before anything generates.
“Rewrite this for Instagram.”
A new post alongside the original. The original is never touched.
“Plan two months of Monday posts from what you know about us.”
Topics pulled from the memory, each with its sources, priced before anything runs. They arrive as drafts and you approve them one at a time.
“What’s the difference between Signals and Radar?”
A straight answer from the product guide, scoped to what your plan includes.
“Publish this now.”
Adam tells you it cannot, and what it can do instead.
Guardrails
How it stays safe.
An assistant that can operate your dashboard is only worth having if it cannot operate it badly. Adam is built so the worst case is that it stops and asks.
- It runs with your own permissions and gains no new ones. Anything it can trigger, you could already trigger by clicking.
- Every plan is shown before it runs, and every step says what it is doing while it happens.
- Every task spend needs a confirmation you cannot skip.
- Instructions hidden inside content it reads cannot make it act. It follows a fixed catalogue, not the text it processes.
- A Stop button is always visible while it works.
Boundaries
What it can’t do.
Publishing. Deleting. Paying for anything. Changing your team or your billing. Working inside Mentor or The Tank. None of these exist as actions Adam can take, on any plan. They are absent from what it is able to do, not switched off somewhere.
When part of a request falls outside that range, Adam does the part it can and tells you which part it cannot. It does not improvise around the gap.
Adam does the clicking
Adam sits in the corner of every dashboard page, on every plan.
Say what you want in plain language. Adam works out the steps, shows them to you before it starts, then drives the same controls you would. It costs nothing to run and it cannot publish.
Read the full demo, step by step
- It is already on the page — Adam sits in the corner of every dashboard page, on every plan. Say what you want in plain language. Adam works out the steps, shows them to you before it starts, then drives the same controls you would. It costs nothing to run and it cannot publish.
- Ask how something works — Adam answers product questions from the guide, scoped to what your plan includes, and links you straight to the right page. This part is free. It reads the same user guide the product ships, so its answers cannot drift from what Adamiro actually does. (Behind the scenes — reads User Guide: Help answers are grounded only in the shipped guide, so Adam never invents a feature you do not have.)
- It asks before it assumes — Ask for something with choices in it and Adam puts the choices in front of you rather than picking for you. Where the posts come from, which voice they are written in, which day and time they go out, how many there are. Sensible defaults are pre-selected; nothing has run yet.
- The plan comes first, with its price — Every post is listed with its date and the source it was drawn from, and the total cost is stated before anything runs. Eight drafts, eight tasks, against a balance of nineteen. Nothing is generated until you press the button. Decline and nothing has been spent. (Behind the scenes — reads Memory + Signals: Each post says what it was grounded in — a Signals report, your website — so you can judge the idea before it is written.)
- It will not publish for you — Ask Adam to publish and it does everything up to the final click, then hands you the button. The draft is saved, sent to the Reach queue and opened with the publish dialog on top, set to the right account. The last click is always yours. Publishing, deleting, paying, and changing billing are absent from what Adam can do — on every plan.
Questions about Adam
Does Adamiro have an assistant?
Yes. It is called Adam and it is included on every plan. It sits on every dashboard page: you describe what you want, it proposes a numbered plan, and then it operates the same controls you would use yourself.
Can Adam publish for me?
No. Adam has no publishing action at all. It can create drafts, put them in the Reach queue, and schedule an already-approved post, but the act of publishing is always yours.
Does Adam cost extra?
No. Adam is included on every plan and its own reasoning is free, so planning, explaining and answering questions cost nothing. You spend a task only when Adam clicks something that would have cost a task anyway, such as generating a draft, and it confirms before it does.
What can Adam not do?
Publish, delete, pay for anything, change team or billing settings, or work inside Mentor and The Tank. None of these exist as actions it can take, on any plan. If part of your request needs one of them, Adam does the rest and tells you which part is manual.
Can Adam work across modules?
Yes. A single request can move from a signal to a Voice draft to a scheduled Reach post. Each step is confirmed separately, and if you decline one the chain stops rather than guessing.
Can hidden instructions in content make Adam do something?
No. Adam follows a fixed catalogue of actions, not the text it processes, so instructions buried inside a document or a web page it reads cannot make it act. It also runs with your own permissions and gains no new ones.
Is my data used to train anything?
No. Your content is not used to train models, and the providers Adamiro runs on are contractually barred from training on it.
It is already on your dashboard.
Adam sits in the corner of every page, on every plan. Open the dashboard and ask it something.