4 networks · 5 AI agents · one interface
AdBeeCore brings Google Ads, Meta, Apple Search Ads and TikTok into one console for apps, websites and service businesses: it spots anomalies, proposes changes, checks them for risk and scores the outcome after the attribution window. Your team stops reconciling spreadsheets and starts making decisions.
No card · 10-minute setup · observe mode by default
Spend · 7d
$48 210
▲ 12,4 %
CPA
$3,14
▼ 8,1 %
ROAS
2,41×
▲ 0,3×
Spend vs revenue · last 14 days
Agents · latest events
Keys and tokens stay yours: the platform is an intermediary, not the owner of your access.
GAQL, OAuth2, quotas
Graph API, 200 calls/hour
ASA API v1, JWT
10 req/sec, 1,000/day
Why AdBeeCore
Most stacks are a console, an ASO subscription, a spy tool, an ETL job and a spreadsheet. AdBeeCore covers the whole loop — and lets you take the data and the agents with you.
Connect an integration and pull everything to your own server
Point storage at your own database or bucket, or stream every table to your warehouse and webhook. The schema is documented, the export runs on a schedule — no vendor lock-in, no per-row fees.
Connect any agent over MCP and let it manage ads
The whole console is exposed as an MCP server: Claude Desktop, Claude Code, Cursor or your own bot get the same typed tools — read, analyse, create and edit — inside your role and product scope.
Keywords, rankings, competitors and their creatives — no extra subscription
Track your keyword positions by country, watch competitors’ listings and creative angles, and connect App Store Connect to push metadata — all on the same data as your paid traffic.
From observe mode to autonomous edits with hard limits
Rules, schedules, hypotheses and approvals in one place: agents propose, Guard checks, the change log records — and Reflector scores the outcome after the attribution window.
Every box below is a vendor, a bill and an integration you have to maintain. AdBeeCore replaces them with one console, one data model and one login.
One product: 4 ad networks + ASO + SPY + MCP agents + own-server export.
Architecture
Decisions are not made by one clever prompt but by a pipeline with separated roles: observer, executor, guard, reflector and curator. Money moves only after a check, and every outcome comes back as a lesson.
finds anomalies
every 30 minutes, significant signals only
proposes a change
bids, budgets, statuses
blocks risk
limits in code, not in a prompt
scores the outcome
after 48–168 hours, score −100…+100
writes the lessons
weekly, into the agents’ handbook
Feedback loop: Reflector waits for the attribution window (48–168 hours) and scores each decision — the outcome becomes a few-shot example for the next ones.
Features
Campaigns, ad sets, keywords and creatives from Google, Meta, Apple and TikTok in a single data model — same metrics, same reporting currency.
Any edit, human or agent, is written with before/after values, author and timestamp. Silent mutations do not exist here.
At most one change per campaign every 72 hours. The agent does not touch bids until the network finishes counting conversions.
Keywords, rankings, custom product pages and the ASO × Apple Ads link on the same data as your paid traffic.
A creative library with per-network metrics, automatic threshold scoring and concept-level retro reports.
Slices by network, geo, device and keyword. Export to a database, webhook or deck — scheduled or on demand.
The difference
Ad consoles show numbers. AdBeeCore remembers what was there before the change, who made it and what came out of it — so the same mistake is not repeated.
Examples
Three real setups. Same product, different depth of autonomy.
Two networks connected, observe mode on. The first week produces a list of anomalies the team would have found on Friday — with numbers and the campaigns behind them.
Each client gets its own workspace and its own export to their warehouse. Members see only the products they are assigned; the agency keeps the shared playbook.
Your agent connects over MCP with a scoped token: it can read metrics, approve nothing by itself and only perform writes that pass Guard and the role check.
Pricing
One plan covers all four networks. Start free, turn autonomy on later.
See your numbers in one console.
$0 forever
For a UA team running several products.
$79 per month
Several clients, roles and access control.
$199 per month
No. It is a management layer on top of the ad consoles: it removes routine work, collects data and proposes decisions, while the final call stays with you — limits and approvals are configurable.
Yes. Observe mode is the default: agents only read data and suggest actions. Autonomous edits are enabled per network, with a daily budget cap per ad account.
Tokens and keys are encrypted with AES-256-GCM using a key from the secure environment. Decrypted values never reach logs and are never returned by the API — only the account id and network name leave the system.
Google Ads, Meta, Apple Search Ads and TikTok. Every adapter implements the same interface, so adding a network does not rewrite the console.
Connecting an account and the first sync take minutes. Historical metrics are backfilled in the background while the dashboard is already usable.
No. Sign-up is free: you get a workspace, the dashboard and observe mode. You pay for autonomous actions and team features only.
Sign-up is free and no card is required. Connect one account and watch the system catch what usually takes a week to notice.