You can read
what the AI knows.
Every outbound tool promises AI personalisation and hands you a box you cannot open. The campaign brain is a page in your dashboard: your positioning, your proof, your tone guardrails and who to target, in fields you can edit. Every message is written from what is on that page.
Failed sends cost the customer nothing.
A dead route reroutes instead of dropping.
Your positioning,
not a personality
The brain starts as a draft. InboundIQ reads your website and pulls out your services, your differentiators and your value propositions, and you confirm or rewrite each one before anything is sent. Nothing in here was guessed and left to stand.
Company and campaign
Company, website, industry, business model, product and the goal of the campaign. Drafted from your site, then yours to correct.
Key differentiators
What the agent leads with when a prospect asks why you. A claim and the sentence that backs it, kept to things you can prove.
Company values
Tone guardrails. The agent writes in line with them and drops any angle that contradicts one, which is how it stops sounding like a template.
Your CTAs
The closing line of every email, and the single biggest lever on how many replies come back. Each one carries a score.
ICP and targeting
Primary and secondary ICP, the buying trigger, core markets, countries in and out, excluded industries, B2B or B2C.
Ideal customers
The accounts you would take tomorrow. Their lookalikes get priority over cold discovery. Add them one at a time or upload a CSV.
Three fields in.
One message out.
A first touch is your proof, your tone and your ask, assembled around what we found about one company and one person. You can see each part of it before it goes, because each part is a field you filled in.
Most providers charge for the code whether it lands or not. We bill on delivery. Want a short teardown of where yours drop off?
The claim comes from your differentiators
Not from a model's sense of what sounds persuasive. The agent leads with the one that fits what it found about this company, in the words you wrote it in.
The tone comes from your values
They are guardrails, not decoration. An angle that contradicts a value does not get written, which is why the output does not drift into the register every other cold email shares.
The ask is a CTA you chose
Every email closes on one of yours. Offers built for the prospect — an audit, a teardown, a mockup — are what the product pushes you toward, and asking for a call is what it argues against.
The follow-up
is still yours
First touches go out on their own. Replies do not. When a lead writes back, the brain drafts and you decide — which is a different relationship with an AI than approving whatever it already sent.
Thanks {{first_name}} — here is the teardown I mentioned. It covers the three points where codes usually fail for teams your size
Eleven lead variables, four tones, three lengths — and a cursor, because nothing here sends itself.
Standing instructions, written once
The brain holds a reply prompt the agent follows on every draft — "keep it under four sentences", "always offer the teardown first". Write it once and stop repeating yourself.
Or tell it what you want, this time
Type the instruction in plain words — write a short follow-up offering a demo — and the draft writes itself into the composer in front of you.
Tone and length are pickers, not prompts
Four tones, three lengths. Change one and it rewrites only the part it generated, so the sentence you typed yourself survives untouched.
Eleven lead variables, one keystroke
First name, title, company, industry, seniority, department, domain, country and the rest. They resolve against the lead in front of you, not a spreadsheet column you mapped.
It is a draft, in a box, with a cursor
Nothing the agent writes for a reply leaves without you. Edit it, keep half of it, or delete it and write your own.
The loop, in full,
including your part in it
Everyone claims their AI gets better. Here is the specific mechanism we can show you: a score on the thing that moves reply rates most, a suggestion when it slips, and a publish step with a human on it.
Every CTA carries a score
The score is how reliably that closing line turns outreach into replies. It comes from what came back, not from an opinion about the copy.
A weak one surfaces as work to do
When a CTA is underperforming, it appears on the brain's overview as an action with the score attached — not as a silent rewrite you find out about later.
You are offered alternatives, and you pick
The suggestions are lines that give the prospect something rather than asking them for time. You take one, or you keep the line you had. Neither happens without a click.
Targeting is checked before it scales
The Targeting tab shows the profiles your current ICP, geography and exclusions actually produce. If they look wrong, you change the controls, not the output.
Changes publish once a week
Saved is not live. You publish, and it takes effect within 24 hours. The weekly limit is deliberate: a campaign needs enough runway to produce a result worth reading before the strategy moves again.
Everything it knows,
you can change
The honest selling point is not that the model is clever. It is that you can open the page it reads from, disagree with a line, fix it, and know exactly which emails change as a result.
The brain decides what gets written. You see what came back.
Interested replies arrive as qualified leads. Give InboundIQ a meeting link and the genuine ones arrive as meetings instead — time agreed, invite sent, nothing in your inbox to chase.
Open the box.
Then change it.
Give InboundIQ your URL. It drafts the brain from what it reads, and every field in it is yours to correct before a single email goes out.