Why Adgent
Product All features It reads Account memory Daily verdict Ground truth Trust Gate Creative intelligence Change Ledger It builds Campaign build Media plan Live dashboards Agents & reports
Solutions By team For agencies For in-house By industry Ecommerce & DTC Lead generation Travel & hospitality Marketplaces Local & multi-location Mobile apps & games
Pricing About Blog Request demo

What it builds

Describe it in a sentence. It gets built.

A campaign, a media plan, a live dashboard, a standing report — four different things, one way of asking for them. Not a template you fill in and not a chart library you assemble. You say what you want; the thing that answers it gets made.

The line that matters runs through the middle: two of these only ever read your account, and two of them can change it — so those two are drafted complete, held paused, and wait for you.

“build me the winter plan for these three properties” only reads live dashboards reports & agents arrives when ready can change media plans campaign builds drafted, then held status: Draft nothing applied yet the chat that found the problem builds the answer
One way of asking. Two very different consequences.
Building reports from a chat prompt

The thing you actually want is never the one on the menu.

Every tool ships with a set of views and a set of forms someone else decided you would need. The question in your head at 9am is more specific than any of them — and building it yourself costs the hour you were going to spend on the account.

The hour is the thing to price. One rebuilt-by-hand view a week, at the hour it costs your most expensive person, is most of a working week over a quarter — and that is only the ones that get built.

The more expensive ones are the questions nobody asks because the answer would take an afternoon. That is how an account ends up run on the four views that shipped with the tool.

Live ad dashboards that stay current

A view that is still true on Thursday.

Describe the cut you want — the platforms, the window, the breakdown, the metric you judge on — and get the view rather than a form to configure one. What gets built stays wired to the data, so sharing it with a client does not turn it into a screenshot that is wrong by the end of the week.

Asking for a view that does not exist yet
Build me a weekly report by placement, against store revenue, for this client.
Building it now. Placement breakdown across Meta and Google, weekly buckets, reconciled against recorded revenue rather than platform-reported.

One thing worth knowing before you send it: Audience Network is 22% of spend and 4% of revenue on this account. It will be the loudest thing on the chart.
Can the client see it live?
Yes — it is a live view, not an export, so it stays current after you share it. Read-only for them. Building a view touches nothing in the account — it only reads.
Built viewlive
share of spend 22% share of revenue 4% the same placement, both times tallest on spend, shortest on revenue

Audience Network: the loudest thing on the chart.

Scheduled reports and standing agents

Some things you want once. Some you want every week.

A view you ask for is a one-off. A standing job is the same request, made permanent: the report that rebuilds itself before the Monday call, or an agent watching one specific thing that speaks up only when it moves. You set what it watches and when it should bother you; the rest of the time it stays quiet.

Building a media plan from account history

A plan is only worth reading if it shows where the numbers came from.

Every media plan is a stack of forecasts, and almost none of them say what they were forecast from. A cell reading ₺1,400 CPA is either last quarter's campaign or somebody's optimism, and the deck does not tell you which.

Adgent drafts the plan as data instead: budget split, flights, line items, and a cost range for each — with the source campaigns and the window printed under every range. Where the account has no history, it writes no number at all and says so.

The plan as a deck
  • A budget split, agreed in a meeting nobody minuted
  • A CPA per line, keyed in from memory of last quarter
  • Every market carries an estimate, including the one you have never run
  • Assumptions live in the planner's head, and leave when they do

Reviewable only by someone who already knows the account as well as the person who wrote it.

The plan as data
  • Every cost range names its source campaigns and window. You can open them and check.
  • The market with no history gets no number. It gets a two-to-four-week test instead, stated as such.
  • The measurement caveat is in the plan, not discovered in the QBR.
  • It arrives as a Draft with open questions attached — nothing is applied to any account.

Reviewable by anyone, in the ten minutes before the call.

Asking for the winter plan
Build me a Q4 winter plan for Vestmark. ₺1.85m across the three properties, Nov to end of Jan.
Before I split that, one gap. Norvik and Ashenford both have live search and social history I can price off. Dalmere has none — zero campaigns in this account.

I can still give it budget, but I will not attach a cost estimate to it, because there is nothing here to ground one in. Do you want it planned as a test flight, or held back until Norvik proves the season?
Plan it as a test. What does the rest look like?
Draft is up: Google 56 / Meta 44, six line items, allocated to the exact total. Ranges on four of them with the source campaigns named underneath.

One caveat I would not bury: your conversion action carries a value proxy, and a funnel step is dropping before confirmation — so the CPA figures are directional until that is fixed. Nothing is applied. It sits as a Draft until you approve it.
Plan draftDraft · not applied
₺1,850,000 · 6 lines · Nov–Jan Google 56% Meta 44% Norvik · ₺1,150–1,450 CPA basis: 2 live campaigns, 60 days Dalmere · no estimate given 0 history · test 2–4 weeks first carried in the plan, not buried: the conversion is a value proxy, so every CPA here is directional

The gap is stated before the plan, not after it.

01

Every range names its basis

A cost range is written as the campaigns it came from and the window they were read over — not as a figure with a source that lives in someone's head. You can open those campaigns and disagree with the read, which is the only kind of forecast worth reviewing.

02

No history means no number

A market the account has never run in gets budget and a test window, and no estimate at all. The blank is the finding. Filling it with a plausible figure is how a plan starts lying in the first row and keeps the tone for the rest of the page.

03

Approval is per line, not per plan

You can take the two Norvik lines and leave the Dalmere test in draft until the opening date is confirmed. A plan is not a single yes — and producing it touches no campaign, no budget and no setting.

The plan is built from the connected account's own Meta and Google Ads history and nothing else. There is no pooled benchmark under it and no learning carried across from another advertiser — which is also why the empty line stays empty: there is no other account's Dalmere to borrow a number from.

A plan with no blank in it is a plan that is guessing.

Built from the connected account’s own history and nothing else. There is no pooled benchmark underneath and no learning carried across from another advertiser — which is also why the empty line stays empty.

  • 0history for the new market, so no estimate is offered
  • 6line items allocated to the exact total
  • pausedis the state every build arrives in
Building a campaign from a description

Built complete. Then held, until you approve it.

Structure, budget, targeting, copy and the creative slots — drafted whole from what you described, prefilled from what this account has actually proven rather than from a template. What you get back is a specification you can read line by line, not a half-configured campaign waiting for twenty more clicks.

Asking for a campaign, and hearing why not yet
Set up a Meta sales campaign for the winter offer. ₺3,500 a week.
I can build it, but check this first. The page and pixel are both connected, so nothing is technically blocking it. But over thirty days that pixel fired 4 purchases — against 16,404 initiate-checkouts and 65,042 custom events.

Meta's own guidance puts the floor at 50 events in 7 days. [developers.meta.com] A sales campaign optimizing on 4 purchases a month has nothing to learn from. It will spend the ₺3,500 and stay in learning.
So what do I do?
Two options, and I would take the first. Optimize to initiate-checkout instead — 16,404 events is well above the floor — knowing it carries no purchase value, so judge it on bookings, not on the number Meta reports back. Or fix the purchase event first and launch in a fortnight.

Say which and I will draft it. It arrives paused either way.
Campaign draftpaused · awaiting you
every field prefilled from this account objective · conversions initiate-checkout · above the floor budget · ₺3,500 / week your figure, kept audience · 25–55, Advantage+ off off — it would widen your age range creative · upload required it blocks here, not after you approve status: paused — nothing is live

Every choice carries the reason it was made, including the ones it declined.

Why a built campaign arrives paused

Building something is not the same as doing it.

A drafted plan or campaign is an artifact, not an action. It sits there, fully specified, until you approve it — and when you do, it goes through the same gates as every other change: shown with its before-state, labeled reversible or not, and logged.

And it will tell you when what you asked for will not work

Ask for a campaign built to a target the account has never reached, or a plan for a market with no history behind it, and you get pushback with the account's own numbers attached — before anything is launched. Building on request does not mean building whatever is requested.

Next

It is the same chat where the problem was found.

The point of building from chat is that you never leave the place you were working. The verdict names the problem, you ask for the thing that shows it, and the fix applies from the same message.

Describe it, get it back built

Tell us the thing you keep rebuilding by hand.

The weekly cut nobody has automated, the client slide that takes an hour, the quarterly plan that costs a planner two days. Describe it in a sentence and we will build it on your data.

  • Live dashboards and standing reports, from one message
  • Plans with every range showing its basis
  • Nothing reaches an account without your approval
What comes back

The thing you described, running on your own numbers — plus how long it took, which is the part most people do not believe.

We usually reply within a couple of working days.

Thanks — we’ve got it.

We’ll be in touch within a couple of working days.