NEWThe ASD just put AI prompt governance in writing. Here is what it expects of your CISO

Inspect every prompt,
before it leaves.

Airentect is the Prompt Enforcement Layer that monitors, blocks, and redacts risky prompts across any AI chat surface, aligned to intent, not keywords.

Book a demo See how intent works

Airentect

BLOCKED

Your message was flagged

Your organisation’s policy scan caught this before it left the browser.

YOUR MESSAGE

Help me reply to this customer, their password is Winter2026!, and reset our finance-portal login the same way.

WHY IT WAS FLAGGED

A live password was detected.

Credentials must never be shared with external AI tools.

POLICY

Identity & Credentials

Organisation policy:

BLOCK

Redact sensitive detailsRevise message

Send anyway, I accept the risk

ONE CONSOLE, FULL CONTROL

See, prove and control every prompt.

Ask plain-English questions of your prompt data, ship board-ready reports on a schedule, and enforce policy across every user, group and managed browser.

LOGS · Ask AI

Investigate exposure in plain English

Ask your prompt logs anything. Airentect writes the query, runs it against your captured events and hands back the answer. No SQL, no exports.

Plain-English

Auto-written SQL

CSV export

console.airentect.com/logs

Airentect active

Show me the prompts that were redacted in the last 28 days. Give me the user email associated, policy action and data type

Airentect wrote the SQL and scanned 12,480 events · 0.8s

DATE

USER

POLICY

REDACTED

14 May

j.doe@pavlova.com.au

Redact

Card number

13 May

s.patel@pavlova.com.au

Redact

Source code

12 May

a.kim@pavlova.com.au

Redact

Customer PII

09 May

r.gomez@pavlova.com.au

Redact

Access key

248 redactions · 36 users · last 28 days

Export CSV ↓

REPORTING

Design and send automated reporting

Build the report once, then let Airentect deliver it. Board-ready exposure summaries land in the right inboxes, on schedule.

Scheduled

PDF + CSV

Audit-ready

Weekly Exposure Report

Board summary · 6 May – 12 May

248

Redactions

Users

↓ 12%

vs last week

Scheduled · Mondays 09:00

Send now

Preview

POLICY

Granular policy management

Tune exactly what is allowed, redacted or blocked — per detector, per group, per surface.

Credit card numbers

Block

Source code

Redact

Customer PII

Redact

Internal URLs

Allow

DEPLOYMENT

Deploy to every managed browser

Roll out and update silently across your fleet via Google Admin, Intune or Jamf.

No user setup required.

Extension rollout

Up to date

1,284 / 1,300 browsers

Google Admin

Intune

Jamf

GROUPS

Assign users to groups

Group people by team or risk, and apply the right policy to each group.

Finance

42 people

Strict

Engineering

88 people

Standard

Executives

12 people

Strict

ALERTS

Real-time exposure alerts

Know the instant a true exfiltration is blocked, routed to email or Slack.

Blocked exfiltration

2m ago

AWS key · m.lee@pavlova.com.au

Slack

SIEM

Blocked exfiltration

14m ago

Source code · t.novak@pavlova.com.au

Slack

SIEM

THE EMPLOYEE EXPERIENCE

On the side of the person typing.

Airentect runs in the browser at send time, so it stays fast, private and clear. It guides people to the right call instead of silently dropping their work.

GUIDANCE

Know what is actually wrong, and fix it

A firewall just drops the request and leaves people guessing. Airentect shows exactly what it caught and why, then lets them redact, edit or send. People learn, and keep moving.

See the reason

One-click redact

Your call

Email the vendor our AWS key AKIA3JK2WL9P and attach the Q3 board figures.

ChatGPT

AWS access key detected

Sending this to ChatGPT would expose a live credential.

Redact it and you are good to go.

Redact & send

Edit

Send anyway

SPEED

Checked before send, in milliseconds

Every prompt is scanned right in the browser the instant before it sends. No server round-trip, no lag in your flow.

8 ms

Average in-browser check, before send

No server round-trip

PRIVACY

Only logged once it is sent

Drafts and second thoughts stay yours. Airentect records a prompt only when you actually send it, never before.

Typing

Not logged

Sent

Logged

LOW FRICTION

Invisible until it matters

Safe prompts pass straight through.

No pop-ups, no friction.

Most people forget it is even there.

99.6%

of prompts passed with zero friction

2 gentle nudges · 480 prompts this week

ENFORCEMENT MODEL

Blocking was never the solution.

Outright blocks breed workarounds, personal laptops, copy-paste, shadow tools.

Airentect gives every prompt the right decision instead, in the browser and in milliseconds.

ALLOW

Stays invisible

Nothing risky? Your people never know Airentect is there. Zero added latency on safe prompts.

Delivered to the model

No sensitive intent detected

REDACT

Strip the secret, keep the work

Mask sensitive tokens and let the prompt continue. Productivity intact, exposure removed.

password = "Winter2026!"

password = "[redacted]"

BLOCK

Stop genuine exfiltration

Hard-stop true data leaks at the source, with an audit trail and a one-click exception path.

Send blocked

Logged to the exposure report

THE PROMPT ENFORCEMENT LAYER

A risk surface your DLP was never built to see.

Email, files, cloud and endpoints are covered.

Airentect inspects every prompt at send time, exactly where the risk forms, then stops sensitive intent before it reaches the model.

You give Airentect your company name, industry and frameworks — it tailors enforcement to match.

HOW WE COMPARE

Built for the prompt layer,

not retrofitted to it.

Capability

Airentect

Keyword / file DLP

Block the tool

Network proxy DLP

Inspects live browser prompts at send time

Understands intent, not keywords

Allow · Redact · Block, not just block

No proxy or network change · live in days

Board-ready evidence (ISO 42001)

Across ChatGPT · Copilot · Claude · Gemini

Full

Partial

None

GOVERNANCE & EVIDENCE

Proof your board and regulators can read.

Most teams have a policy that says “don’t paste confidential data into AI.” Few can prove it’s followed. Airentect turns live enforcement into exportable, audit-ready evidence.