One app, two jobs: manage every Google Business Profile in bulk, and monitor how Google, ChatGPT and other LLMs see your locations — then fix what they find. Powered by AI agents that never sleep.
Every panel above the fold — the reviews, the questions, the bot working the queue, the location dialogs — is the real interface, running on fictional locations. Scroll back up any time and take control.
Three brands, 32 countries, one map. Colour is health, size is reach.
Reviews and questions arrive in the customer's language; the agent answers in kind.
You set how much the agents do — from on-demand drafts to publishing on their own.
Pull every location from Google or a workbook. Nothing is edited on Google directly — you hold a copy you control.
Hours, categories, attributes, descriptions, photos, posts. A preview shows will-change, already-matches and blocked before anything moves.
Someone can prepare a hundred changes and someone else can release them. Enforced in the database, not the UI.
Ten edits per location per minute — Google's ceiling, not ours. Blocked profiles are refused before they queue, so a run never fails forever on the same row.
Every agent is specialised, trained on its own material and limited to its own job. A draft never reaches Google without passing a second agent that runs on a different model — and, below gear 3, a person.
They also work as a crew. Agents hand work to each other, watch each other's output and step in when something looks off: the Proof-reading Agent checks every draft, the Policy Guard can stop any action, the Validation Agent refuses what Google would refuse, and the Watch Agent reports what actually changed after a publish. No agent acts unchecked — and every check is in the logbook.
Every agent starts with an onboarding training on your own material: your locations and services, approved replies and posts, your brand's tone and voice, the terms and promises it must never use, and the rules of your industry. That is how it understands the context before it writes a single word.
Then comes a calibration run. On a first batch of real reviews, questions and profile fixes the agents draft everything — and nothing goes out. You approve or decline each piece and, where something is not as expected, say why: too formal, wrong term, never offer a refund, not how we talk to customers in Zurich. Every decision feeds the agent's next learning round, and the confidence it shows in the logbook is calibrated on exactly this feedback. The more detail you give, the faster the drafts sound like you — the agents are eager to learn ;-)
Answers the public questions on your profiles — the ones strangers answer first if you don't.
Replies to every kind of review, in the reviewer's language and your tone.
Finds the gap between what people ask and where you are not the answer.
Reads every draft the FAQ and Review agents produce — on a different model — before anyone else sees it.
Checks every field against Google's real rules before a profile can enter a publish run.
Watches Google for changes you did not make — suggested edits, hours drift, duplicates.
Drafts posts, offers and photo descriptions per location and season.
The line that no other agent may cross — and the one that decides what the gear allows.
The same gear switch you saw under the logo governs the whole workspace. Change it per brand, per market, or per kind of task.
Moving between gears takes one click and applies from the next item onwards — nothing that is already published changes, and every publish stays logged and reversible from its snapshot.
Set it where it makes sense: gear 3 for 5★ reviews and opening-hours questions, gear 2 for anything under 3★, gear 1 for a brand that is new to the platform or for a regulated market. In gear 3 the agents still act on their own only above the confidence threshold you set in the logbook; every draft below it waits for a person.
Nothing happens on its own. The agents watch, flag and measure; ask for a draft whenever you want one. Every publish is a person's click.
Agents draft replies, answers and fixes the moment they are needed — one item at a time, in the customer's language. A person approves before anything reaches Google.
Agents reply, answer and fix within the rules you set — tone, no-go topics, score thresholds — and publish. Everything is logged and reversible from the snapshot.
The database is the source of truth and publishing is a separate, deliberate act. That one decision is what makes two hundred locations manageable by a team instead of a hero.
Google's hard ceiling. Cannot be raised. It governs bulk publishing across every location you have.
We target 250 and pace every run under it.
An honest deferral message when a run would exceed it.
Basic API Access is granted per project and is not transferable. Start it first.
| 1 · Standby | 2 · Drafting | 3 · Autopilot | |
|---|---|---|---|
| Reviews answered | by you, on request drafts | agent drafts · you approve | agent drafts and publishes |
| Q&A answered | by you | agent drafts · you approve | agent publishes from profile facts |
| Profile fixes (category, description, hours, photos) | suggested only | staged for approval | applied and queued for publish |
| Publishing to Google | always paced, always snapshotted, always logged — regardless of gear | ||
| Reversible | ✓ | ✓ | ✓ from snapshot |
Tell us a little about the business and we come back with a pricing proposal that fits it — per location, volume tiers, and the gears you actually need. No card, no self-serve trial that quietly bills you.
A proposal and a walkthrough on your own locations follow within two working days. No card, no self-serve trial that quietly bills you.
BusinessProfileAgent.com is a web application for managing many Google Business Profile locations in one app. It imports every location through the Google Business Profile API, lets you edit profiles one by one or in bulk, and publishes approved changes back to Google. AI agents built on large language models (LLMs) answer customer questions, draft replies to reviews, optimise each business profile, and create posts and other publications — at the level of autonomy you set, from draft-only to autopilot.
No. You grant access through Google's own OAuth screen; we never see a password. You can revoke the grant from your Google account at any time, and we run the disassociation flow Google's policy requires.
A profile Google would refuse — no primary category, an invalid phone — never enters a publish run. It is flagged with the exact reason and a link to fix it, so a run never fails forever on the same row.
Not yet. The data model separates each platform from the location record so Bing can be added without restructuring, and Bing Places has no review API, so review features are Google-only by nature. Bing Places does offer its own Google Business Profile sync: it imports your listings from Google and keeps them updated weekly, biweekly or monthly. Once your Google Business Profiles are up to date, that makes the start on Bing much easier — with the caveat that categories and UTM tags may not carry over exactly.
Only in gear 3, and only within the rules you set. In gear 2 it drafts and a person approves. In gear 1 it drafts when asked. Every reply is logged and can be edited or deleted afterwards.
In the EU, in your own workspace. Your locations, reviews and history stay yours; workbook and JSON export put custody with you at any time.