Tempio

The AdKit alternative: one workspace over every account, and every change on the record

AdKit runs ad campaigns from an MCP client. What its own repositories state, where Tempio differs, and where AdKit is ahead.

Tempio and AdKit, row by row
TempioAdKit
What it isTempio is an AI ads command center. It connects Meta, Google and TikTok ad accounts into one normalized dashboard, and AI drafts campaign changes as reviewable change requests that pass an approval gate before they touch live spend."The ads toolbox for SaaS founders & their AI agents", as its GitHub organisation profile puts it: a hosted Model Context Protocol server plus a pack of published skills.
What it changesEvery change is written as a draft and held at the gate. Publishing to live accounts is in private beta.Its Ads MCP README states that every change starts as a draft and is applied only on your explicit approval. Its adkit_manage tool creates, edits and publishes campaigns and pulls metrics.
Ad platformsMeta, Google Ads and TikTokThe Ads MCP README names seven ad platforms today and marks two more as coming soon.
Access and approvalRead, draft and approve are separate scopes on separate keys. A read-only key never sees the approve tools.You sign in on first use against its hosted endpoint at mcp.adkit.so. The README describes multi-account access for agencies and teams; the client is MIT licensed.
Ad creativeMake ad stills from a preset and your brand kit. The model is asked for a scene with no words in it, and your headline, badge, permit line and logo are set over it from fields you typed. A scene that still shows text after one automatic retry is kept and marked, so you see it before it runs anywhere. Resize for another placement without a second generation, and attach a still to an ad as a draft.Its Ads MCP README lists adkit_studio, a tool that generates ad images and copy.

Both of these are servers your agent talks to

AdKit publishes Ads MCP as a Model Context Protocol server for managing ad campaigns "from any MCP-compatible AI agent", hosted at mcp.adkit.so, signed into on first use. Tempio's server is the same shape of thing: one endpoint, one scoped key, and your own agent doing the asking.

So the question is not which product has an MCP server. Both do, both are open about it, and both publish the tool list. The question is narrower and more useful: what will the server do without you, what does it write down when it does it, and what can you check afterwards.

What a draft is on each side

AdKit's README states that every change starts as a draft and is applied only on your explicit approval. That is the same commitment we make, stated the same way, and it is worth saying plainly rather than pretending we invented the idea.

Where the two differ is what the draft is made of. In Tempio a draft is a row: a change request with the account, the entity, the before and the after, who wrote it, who cleared it and when. Approval is a separate scope on a separate key, so a key that reads your numbers cannot approve anything, and the agent that drafted a change cannot clear its own work. The record is the product, not a side effect of it — which is why the answer to "who changed this budget" is a query rather than a memory.

Where the numbers come from

Tempio stores base counts and computes every ratio when you ask for it. A ratio with nothing to divide comes back blank rather than as a zero, money is held in integer minor units with its currency attached, and a blended figure across two currencies is labelled an estimate rather than quietly summed.

This reads as pedantry until the first time a dashboard tells you a campaign had a 0% click-through rate when what actually happened was that it had no impressions. The two are different facts and they should not look the same. AdKit's published material does not describe how it stores or computes its metrics, so this page does not say anything about how it behaves here; it is a difference we can state about ourselves and not one we can state about them.

What you can check before you commit

Everything on this page about AdKit comes from two pages anybody can open: its GitHub organisation profile and the README of its Ads MCP repository, both read on 12 September 2026 and both linked in the sources at the end of this page.

What is not on this page is AdKit's pricing, its plan names, its trial terms or anything else that lives only on its own website, because we could not open that site from the machine this page was written on. A price we cannot cite is a price we do not print, so the price row is missing rather than guessed. When the page can be read again, the figure and the date it was read go in together.

Where AdKit is ahead today

One thing, and it is not close.

It reaches more ad platforms. The Ads MCP README names seven ad platforms it manages today and marks two more as coming soon. Tempio connects three. A buyer who runs accounts on a platform Tempio does not connect is better served by a product that reaches it.

Two more used to sit on this list, and both products now ship them.

An ad library. adkit_library is listed in its README as a search across more than a million ads. Tempio's is smaller and built from the public transparency archives: Search a library of live competitor ads by platform, format, offer and days live, then save what works to a swipe file.

A creative tool. adkit_studio is listed as generating ad images and copy, and AdKit also publishes a skills pack for campaign structure, targeting, creative and budgets. Tempio's is two layers rather than one: Make ad stills from a preset and your brand kit. The model is asked for a scene with no words in it, and your headline, badge, permit line and logo are set over it from fields you typed. A scene that still shows text after one automatic retry is kept and marked, so you see it before it runs anywhere. Resize for another placement without a second generation, and attach a still to an ad as a draft.

If reach is the job you are hiring for, the rest of this page is not going to change your mind, and it should not.

What each product is for

AdKit describes itself as a toolbox. The organisation profile reads "The ads toolbox for SaaS founders & their AI agents", and the repositories under it back that up: an MCP server that manages campaigns across seven ad platforms, a published skills pack covering advertising and marketing strategy, campaign structure, targeting, creative and budgets, and a set of conversion-tracking libraries for JavaScript and TypeScript. It is a wide product, aimed at a founder who wants one place to reach a lot of surfaces.

Tempio is narrower on purpose. It is a command center for the accounts you already run: connect them, normalise them into one org-scoped workspace, and hand your own agent the numbers and the tools. Fewer platforms, more depth on each one, and an approval gate that every write passes through whether a person cleared it or a rule did.

Neither of those is the better answer in the abstract. Breadth wins when your spend is spread across surfaces nobody has built a good adapter for. Depth wins when most of your money sits on two or three platforms and the thing that costs you is not reach but the change somebody made on a Tuesday that nobody can now explain.

The tool list, and what it tells you

AdKit publishes ten tools: adkit_projects to list and select a project, adkit_status for connected platforms and accounts, adkit_setup for connection links, adkit_help for platform support and parameters, adkit_manage to create, edit and publish campaigns and pull metrics, adkit_library to search more than a million ads, adkit_studio to generate ad images and copy, adkit_upload_file, adkit_feedback and adkit_checkin.

A short tool list is a design decision, not a limitation. Ten tools with a help tool beside them is a surface a model can hold in its head, which is why the pattern keeps appearing: one manage verb, one help verb, and a router underneath. We landed on the same idea from the other direction and it is one of the few places where two products converging is a signal rather than a coincidence.

What a tool list cannot tell you is what happens between the call and the account. That is the part worth reading the rest of this page for.

The gate, in practice

On Tempio, a write from a tool is never a write. It is a change request: a row with the organisation, the account, the entity, the operation, the values before and after, the guardrails that were checked when it was written, and the identity that cleared it. Approval is its own call, with its own scope, on its own key.

That has three consequences you feel within a week.

The first is that autonomy becomes a dial rather than a switch. A rule you wrote can clear a class of change — a budget move inside a band you set, on an account you named — and everything else waits. The record does not care whether the approver was a person or a rule; it records which, and the change stays reversible either way.

The second is that a bad instruction is cheap. An agent that misreads a brief and drafts eleven budget changes has produced eleven rows to reject, not eleven live changes to unwind across three platforms at nine at night.

The third is that the audit question has an answer. Every draft, every plan and every connection action lands on the record, scoped to your organisation, and the question "who changed this, and when, and on whose authority" is a query rather than an argument.

Planning above execution

The other structural difference is that Tempio has a planning layer and AdKit's published material does not describe one. Plan a flight across your connected accounts — brief, budgets, dates — then turn the whole plan into drafts in one action and watch it run on one timeline, paced against plan every day, with phases that scale budget under the cap you set. The drafts a materialised plan produces arrive at the gate as one batch rather than as forty unrelated changes.

This is the part of the product that exists because of how founders actually work. Nobody decides one budget. They decide a quarter, and then spend three weeks translating that decision into individual changes on individual platforms, losing the shape of it somewhere around the second week.

Who each one suits

AdKit suits a founder who wants reach: more ad platforms than we connect, an ad library to research what is running, and a creative tool to draft the assets. It is a toolbox, it says so, and the breadth is the point.

Tempio suits somebody whose spend is concentrated on a few platforms and whose real problem is control: several accounts, possibly several businesses, an agent they want doing the legwork, and no appetite for a change nobody can explain. Fewer surfaces, a harder line on what is allowed to happen without you, and a record that survives the argument.

If you want both, the honest answer is that you can run both. They are MCP servers. Your agent can hold two.

Sources

Every statement on this page about AdKit comes from one of these two pages, read on 12 September 2026.

AdKit's own website could not be opened from the machine this page was written on, so nothing that lives only there — its pricing, its plan names, its trial terms — appears above.

Questions

Put frontier AI to work on every ad account you run.

Tempio connects your ad accounts into one workspace and hands your own AI the numbers and the tools. Every change it writes is a draft until you approve it, or until a rule you wrote does.

Request early access
  • Read and draft by default. Approve is a separate permission you grant.