Our AI agents · Outside only

AI agents work your firm from outside. Fixed checks decide what is true.

Every paid assessment is human-verified before release. Agent against agent: attackers point AI at firms, and LeakTrace reads them first.

This page says what each AI agent does, what it can never do, what waits for your approval, and where you can read its work.

  • Your AI security agent, on your dashboard, uses AI to read your findings and answer you.
  • On a paid assessment, the Discover AI agent chooses follow-up checks.
  • The Close AI agent chooses when to re-check a fix you have marked done, and where your DNS host's own screens go among the fixed steps for an email record.
  • On monitoring, the Monitor AI agent sorts the changes the fixed checks find, so you see first the one that needs you.
  • The first read, the hourly watch and the daily re-check are fixed outside checks, and no AI decides a finding.
  • If the AI fails or reaches its limit, the fixed path runs as it did before.
  • A person verifies every paid assessment before release.

Our AI agents work from outside only. They never log in, never install anything, and never touch your internal systems. The one change they can make, to your email records, happens only when you approve it.

Check my firm How it works
Outside onlyNo log-insNothing installedHuman-verified
Works from
Outside only
Log-ins
None
Changes
Only when you approve
Findings
From fixed checks only
Release
Human-verified
Its work
Dated, on your dashboard
01 · What each AI agent does

Discover, Validate, Close, Monitor.

DiscoverOn a paid assessment, the Discover AI agent reads what your firm shows the public internet with fixed outside checks: your email records, your website, your host names, look-alike domains, and firm addresses in monitored breach databases.It records each finding with the evidence it relied on and the date it was read.It then chooses follow-up checks: which to run on what the first read listed, in what order, and when to stop. The result of each check is what is stored.The AI writes no finding.
ValidateEvery finding is kept with the record that proves it and a check you can run yourself. A finding with no evidence is not released.Then a person reads the assessment: every paid assessment is human-verified before release. Validate is a person's step. We do not call it an agent.
CloseOn a paid assessment, the Close AI agent gives you the fix for each finding, step by step.For a fix to your own email records, where we hold a guide for your DNS host, it sets the steps out with that host's own screens.Every line in those steps is written by us, and AI chooses only where each one goes.When you mark a fix done, AI chooses whether to re-check now or wait for the change to show from outside.The finding is recorded closed only when a fresh read from outside shows it, with the date and the evidence.On monitoring, a finding that is gone from the daily re-check is recorded closed only after a fresh read from outside agrees.The closure is dated on your dashboard, with its evidence.
MonitorOn monitoring, fixed checks re-check your email settings, look-alike domains and new host names every hour, and everything else in full every day.When they find a change, the Monitor AI agent puts first the change that needs you, adds a plain reason when it has one, and says when a new look-alike domain is worth a takedown request.It cannot change a severity and cannot hold back a High or Critical change. You are alerted when something changes.On monitoring, each look-alike domain worth a second look is followed from found to gone: it is re-checked twice a day in public records, and if you tell us it is not yours, the request to its registrar is drafted for you to send.
Your AI security agentAsk it a question in plain English and it answers from your own firm's findings, and names the finding it used.It drafts a note about any finding, for you to send. It proposes a re-check and runs it only when you approve. When it cannot answer, a person replies.
How oftenFixed checks run before your outside reading is written. The Discover AI agent chooses follow-up checks only after you buy.On a paid assessment, the Close AI agent's re-check queue runs every 30 minutes for a finding you have marked fixed.AI may hold a re-check back for a set wait, inside the same retry window.The Monitor AI agent's fixed checks run every hour and every day while you are on monitoring, and the first month of monitoring comes with the assessment.

What AI decides, what proves the finding, what needs your approval

Discover
What AI decidesAfter the fixed first read, which follow-up checks to run on what it listed, in what order, and when to stop. It can point a check only at a host name the first read listed under your own domain, an email record the first read could not read, or the look-alike check with the name of a service your firm uses.
What proves the findingThe result of the check, stored with its evidence and the date it was read. The AI writes no finding and no evidence.
What needs your approvalNothing. It reads from outside and changes nothing.
Close
What AI decidesFor a finding on your own email records: where your DNS host's own screen lines go among our fixed steps. After you mark a fix done, whether to re-check now or wait for the change to show from outside.
What proves the findingEvery line of the steps is written by us: our fixed steps, unchanged and in order, and your DNS host's own guide. The AI writes no text. A finding is recorded closed only when a fresh outside read shows it.
What needs your approvalAny change to your email records, one step at a time. You mark a fix done. The AI agent never does.
Monitor
What AI decidesWhen the hourly watch or the daily full re-check finds a change: which change you see first, a plain reason for it when it has one, and whether a new look-alike domain is worth a takedown request.
What proves the findingThe change as the fixed check stored it. A reason beside it is a sentence we wrote for that kind of change: the AI picks one or none. The AI agent cannot change a severity and cannot hold back a High or Critical change.
What needs your approvalA takedown request is drafted only for a look-alike worth a second look that you say is not yours, and you send it.
Your AI security agent
What AI decidesWhich of your own findings answer the question you asked, and when to propose a re-check.
What proves the findingIt names the finding it used. When it cannot answer, a person replies.
What needs your approvalA re-check runs only when you approve. A note about a finding is drafted for you to send.
02 · What they can never do

Outside only means outside only.

They never log in to anything of yours, never install anything, and never touch your internal systems.

They send nothing in your name: a note about a finding and the request about a look-alike domain are drafted for you to send. They change nothing on their own.

No AI agent can state a finding without a fixed outside check behind it, and AI cannot change a severity or edit evidence.

Every limit is written down in the outside reading policy.

  • Log in to anything of yours
  • Install anything
  • Touch your internal systems
  • Send anything in your name: notes and requests are drafted for you to send
  • Change anything on their own
  • Release an assessment without a person
  • State a finding without a fixed outside check behind it
  • Let AI change a severity
  • Let AI edit evidence
  • Point a check at anything the fixed checks did not already list for your firm
  • Use AI on a firm that has not bought: these AI agents work for paying firms only
  • Hold back a High or Critical change
03 · What needs your approval

Nothing of yours changes until you say so.

Four things wait for you. A change to your email records is made only when you confirm each step. A re-check your AI security agent proposes runs only when you approve it.

No AI agent marks a fix done: you do. A takedown request is drafted only for a look-alike worth a second look that you say is not yours.

The email record, one click per step, works where your domain's DNS is at Cloudflare and you supply a token for that one use. Anywhere else you get the exact record to paste.

Each step you confirm is recorded with the value before and after. No token is kept: it is used once and discarded.

04 · Who verifies

A person, every paid release.

A person. Every paid assessment is held until a person has read it, and nothing is released on its own: it is human-verified before release.

On a paid assessment, a person reads your findings and your fix steps before you do.

Release without a personNever
A finding with no evidenceNot released
DeliveryInside 72 hours of confirmation
05 · How to see their work

Every line dated, with the stage that did it.

Your dashboard lists what each AI agent did for your firm, newest first. Every line carries its date and the stage that did it, and a closed finding carries the evidence of the re-check.

How often a check runs is stated only for a record the checks really run for. With monitoring off or paused, the list says so and shows only what happened.

On your dashboard, under AI agent activity
  • DiscoverRead your firm from outside.
  • ValidateA person verified your assessment before it was released to you.
  • CloseConfirmed closed from outside, with the evidence of the re-check.
  • MonitorThe daily full re-check ran and found nothing new.
The kinds of line the list holds. Yours carry your own dates and findings.
06 · Which page next
The stepsFrom your request to a fix you can confirmHow it works → The methodHow each finding is read, ranked and provedMethodology → For a firmCheck my firm, then the External Cybersecurity AssessmentBusiness →