How it works · AI discovery and advertising
How your agent works: discover, orchestrate, optimize.
AITWIRE gives your business its own AI agent for AI discovery and advertising. Your agent finds where you could win and what’s holding you back, gets the work done with your approval, and checks what happened. With AITWIRE Managed, our team supervises your agent and carries out the approved work.
Discover → Orchestrate → Optimize
Discover
how AI presents you, and who you could win
- Your agent
- An AITWIRE analyst
- You
Orchestrate
the work, with your approval
- Your agent
- You
- Our team
Optimize
check what happened, then improve the next round
- Your agent
- Our team
- You
What your agent works on
“Does AI understand what we actually do?”
Your agent checks how up to eight AI assistants describe your services, the areas you serve and your people, against the facts you’ve approved, and gets what you control corrected.
“What should we change?”
It finds the questions where customers and clients could find you but don’t yet, and proposes what to change: clearer wording, profile corrections, a ChatGPT ad where OpenAI allows it.
“Can someone handle this safely?”
An AITWIRE analyst checks every finding. The right person at your business approves the wording, the ad and the spending limit; only the kinds of change you set to approve automatically skip that step.
“Is this bringing useful business?”
AI-answer checks, ad results and enquiries are kept apart. Our team records the enquiries your staff tell us about, and what came of them, with their source where known, and never treats them as proof of what caused them.
“Do I have to manage another tool?”
No. With AITWIRE Managed, our team runs your agent; you give the direction and the approvals. Managing it yourself isn’t open yet.
1 · Discover: how AI presents you, and who you could win
- Your agent asks up to eight AI assistants directly: ChatGPT with search, ChatGPT without search, Gemini, Perplexity, Claude, Grok, Mistral and Llama (Meta). ChatGPT counts twice: with and without web search. AITWIRE Managed includes all eight.
- It compares every answer with your approved facts: your services, the areas you serve, your team and the work you’ve documented. It shows what’s described wrongly and what’s left out.
- It shows the questions where you’re not named at all, or where a competitor you track is named instead: the questions where customers and clients could find you but don’t yet.
- Each finding shows its date, how many answers were sampled, the AI service and the model it reported. Missing information is shown as unknown, never as zero.
- An AITWIRE analyst checks every finding before it reaches you.
2 · Orchestrate: the work, with your approval
- Your agent prepares the work and says where it goes: your website, a business profile, your AI Experience Page, a ChatGPT ad where OpenAI allows it, or, for Managed clients, an offer our team sets up with you.
- Each decision arrives as one card: the AI answer, your approved fact and the proposed wording. Approve it, ask for changes or decline.
- Every piece of work moves from proposed, to approved, to published and checked live.
- Once it’s approved, our team updates your website and sends correction requests to the sites that describe you; each of those sites decides what it changes. Your agent updates your AI Experience Page and the machine-readable files it keeps for you.
- Your agent and our team change what you control, and ask others to correct what they publish. No one can edit what an AI assistant says.
- ChatGPT ads run in your own OpenAI Ads account. Each campaign is created paused and starts only after you approve the ad and its spending limit. Ads are labelled as advertisements, and buying one doesn’t change what ChatGPT says.
3 · Optimize: check what happened, then improve the next round
- Your agent asks the same questions again on a schedule and shows which answers changed and which stayed the same.
- Then it recommends the next round, including a different approach where nothing changed.
- Ad results are reported on their own, never mixed into the answer checks.
- Enquiries are kept separate too: our team records the enquiries your staff tell us about, and what came of them, with their source where known.
- A change in the answers is reported as a change, not as proof of what caused it. Approved does not mean published. Published does not mean improved.
Who does what
- Your agent
- Asks the questions; compares the answers with your facts; shows where you’re left out
- AITWIRE’s team (Managed)
- Checks every finding
- You
- Approve the facts it works from
Discover
- Your agent
- Drafts the work; updates your AI Experience Page; prepares ads, paused
- AITWIRE’s team (Managed)
- Applies website changes; sends correction requests; runs your ChatGPT ads; sets up offers
- You
- Approve the wording, the ad and the spending limit
Orchestrate
- Your agent
- Asks again; reports what changed; recommends the next round
- AITWIRE’s team (Managed)
- Records the enquiries your staff tell us about
- You
- Judge the results; set the next direction
Optimize
Included with your agent
Drafting (the wording, written for you)
Available nowYour agent drafts website wording, corrections, FAQ answers and ChatGPT ads from the facts you’ve approved, for you to approve.
Who does it: Your agent; our team checks it
ChatGPT ads (in your own account, within your limits)
Available nowRun where OpenAI allows them, in your own OpenAI Ads account, within the spending limit you approve, and reported separately from AI answers. Optional, and agreed separately.
Our fee is based on the budget you approve, not on what we spend. AITWIRE earns nothing extra by spending more of your money, and loses nothing when it finds savings. Its job is to get you the most results per approved dollar.
Who does it: Our team, through your agent
Website and profile fixes (Correction Campaigns)
Available nowApproved wording goes onto your website, and correction requests go to the sites that describe you; each of those sites decides what it changes.
Who does it: Our team
AI Experience Page (your facts, published for AI systems to read)
Available nowYour approved facts as a hosted page plus the machine-readable files AITWIRE keeps for you, on AITWIRE’s hosting.
Who does it: Your agent
Every decision on one card
Available nowThe AI answer, your approved fact and the proposed wording together: approve, ask for changes or decline.
Who does it: You
Analyst Check (a person checks every finding)
Available nowAn AITWIRE analyst checks each finding before it reaches you.
Who does it: Our team
Results (the same questions, asked again)
Available nowWhich answers changed and which didn’t, with the date, sample size, AI service and model.
Who does it: Your agent
Agent Findings (the next round)
Available nowWhat to do next, including a different approach where nothing changed.
Who does it: Your agent; our team checks it
Enquiries (recorded by our team)
Available nowOur team records the enquiries your staff tell us about, and what came of them, with their source where known, kept apart from AI answers and ad results.
Who does it: Our team
Offers and Redemptions (offers you approve, redemptions your staff confirm)
Available nowOffers, for Managed clients today: our team sets up and tracks each offer by hand — you approve the offer, the ad and the spending limit; the ad runs on ChatGPT, labelled; your staff confirm each redemption by code or link, and our team reports redemptions separately from ad results. A redemption shows an offer was used, not that the customer wouldn’t have come anyway. Some offers aren’t allowed (alcohol, for example). Ads are optional and agreed separately.
Who does it: Our team
Two examples
Growth: the brokerage that isn’t named
A brokerage sells a lot of condos in one Ottawa neighbourhood. When sellers ask an AI assistant for the best brokerage to sell a condo there, the answers name three other brokerages. Nothing they say is wrong; this brokerage just isn’t in them.
- Discover.
- The weekly check shows it: 14 of 20 ChatGPT answers name other brokerages and leave this one out. The agent compares that with the approved facts and finds them thin: they don’t mention the team’s condo sales in that neighbourhood, or that it helps sellers who live out of town. An AITWIRE analyst checks the finding.
- Orchestrate.
- The agent drafts clearer seller-services wording for the website, and updates for the business profiles that describe the brokerage. If OpenAI accepts brokerages for ChatGPT ads, it also prepares an ad for seller consultations, never for an individual listing. The broker of record approves the wording, and nothing goes out before that. Our team updates the site and sends the profile updates, and the agent updates the brokerage’s AI Experience Page.
- Optimize.
- A few weeks later the agent asks the same questions again and shows how many answers now name the brokerage, and how many still don’t. Where nothing changed, it recommends a different approach for the next round. Seller enquiries are recorded separately, with their source where known, and never counted as proof that AI sent them.
No factual error was needed to start. The agent found where the brokerage could win, and carried the work through.
Growth: the heat-pump option homeowners haven’t considered
An Ottawa heat-pump installer fits cold-climate heat pumps in older, oil-heated homes, after a home assessment. Homeowners replacing an oil furnace rarely think of a heat pump. When they ask AI assistants for their options, the answers list furnaces, seldom mention heat pumps, and never name this installer. Nothing they say about the installer is wrong; they just don’t mention what it does.
- Discover.
- The weekly check shows it: in 12 of 20 ChatGPT answers to “What are my options to replace an oil furnace in an older Ottawa home?”, heat pumps aren’t mentioned and the installer is never named. The agent compares that with the approved facts: assessments for older homes are a listed service, and the installation records include older homes. An AITWIRE analyst checks the finding.
- Orchestrate.
- The agent drafts a clearer service page explaining heat-pump assessments for older homes, and a ChatGPT ad for home assessments, where OpenAI allows it, with a spending limit. The owner approves each piece, and nothing goes out before that. Our team updates the service page, and the agent updates the installer’s AI Experience Page. The ad is created paused in the installer’s own OpenAI Ads account and starts once approved.
- Optimize.
- A few weeks later the agent asks the same questions again, shows what changed and recommends the next round. The ad’s results are reported on their own, because an ad doesn’t change what ChatGPT says. Assessment enquiries are recorded separately by our team, and never counted as proof that AI sent them.
No page promises suitability or savings: the assessment decides what suits each home.
Public AI
The AI you don’t control: ChatGPT, Claude, Gemini and the rest. No one can edit what they say. Your agent checks how they answer about you against the facts you approve, and checks again after each change.
AITWIRE · Private AI
The AI you run or buy — internal copilots, support bots, website chat. Here you own it, so you govern it: tested against the information you approve and the claims you prohibit, continuously.
Your agent asks up to eight AI assistants directly: ChatGPT with search, ChatGPT without search, Gemini, Perplexity, Claude, Grok, Mistral and Llama (Meta). ChatGPT counts twice: with and without web search.
The Five Dimensions
How AI answers about you — scored every AI Answer Update Cycle
Sample data for illustration. Your real scores appear on your AI Monitor dashboard after your first AI Answer Update Cycle.
AI is already shaping what your customers and clients believe before they visit your website.
When someone asks AI about your category, it may recommend you, leave you out, or repeat out-of-date information. Your agent finds where you could win and what’s holding you back, gets the work done with your approval, and checks what happened.
Evidence you can check.
Every number shows its date, sample size, the AI service and the model it reported. Missing information is shown as unknown, never as zero. A change is reported as a change, not as proof of what caused it.
Built for local businesses whose customers and clients ask AI who to hire: real estate first, then major home projects such as heat pumps, and law firms.
Tell us which customers and clients you want to win.
Pilot places are open. Each pilot is measured against a scorecard agreed with the client before it starts.
Prefer to run it yourself? Managing your own agent isn’t open yet. Join the self-service waitlist