AegisAI Perspectives · 148
AegisAI is AI-powered email threat detection. It connects to Microsoft 365 or Google Workspace by API, judges each email by intent and removes what fails.
See it on your own Google Workspace or Microsoft 365 tenant. No MX change.
Or get the Q3 2026 State of the AI Attack report (free, instant download)
Not ready for a demo? Check your domain.

AI-powered email threat detection checks each email for attacks. AegisAI connects to Microsoft 365 or Google Workspace through the API, judges every message by its intent, and removes what fails from every mailbox it reached1. AegisAI reads what each message asks the reader to do, and whether that fits the sender2. The API (application programming interface) is the platform's own access route into the mailbox.
If you run IT for a company of 20 to 500 people on Microsoft 365 or Google Workspace, with a small IT or security team and no SOC (security operations center, a team that watches alerts all day), this page answers the questions buyers ask first. Every answer about the product comes from a live company page, most of them the platform overview. Why the company was started is in the founding post.
Key takeaways
- AegisAI reports on its inbound email security page that the median verdict takes 1.9 seconds, and a convicted message is removed from every mailbox it reached within seconds16.
- The demo page states the product is built by the team behind Google Safe Browsing and reCAPTCHA19.
| Question | Answer | Source |
|---|---|---|
| What is AI-powered email threat detection and how does it work? | Software reads each email, checks it several ways at once and acts on the verdict. At AegisAI, a team of AI agents investigates every message17, and reputation, intent, policy and headers are checked in parallel3 | Homepage and platform overview, aegisai.com |
| Which mail platforms does it work with, and how long is setup? | Microsoft 365 and Google Workspace, read through the API after delivery6. The platform overview gives setup as five minutes with no MX change7 | Platform overview, aegisai.com |
| How are links and attachments handled, QR codes included? | Each one is opened in a sandbox (an isolated place to open a file or page safely) and the page at the end is read11 | Platform overview, aegisai.com |
| What does a verdict contain? | A confidence score and plain-language evidence4 | Platform overview, aegisai.com |
| How fast is a verdict? | A 1.9 second median, per the inbound email security page | Inbound email security, aegisai.com |
| What happens to a convicted email? | It is pulled from every mailbox, forwards included5 | Platform overview, aegisai.com |
| Can it start without removing anything? | Yes. It starts read-only in monitoring mode, and removal needs write access that admins can grant or revoke8 | Platform overview, aegisai.com |
| What if the service goes down? | Mail keeps flowing, and detection pauses until the service returns10 | Platform overview, aegisai.com |
| Which threats does it cover beyond phishing? | Business email compromise (a payment or invoice scam sent by email) and account takeover (early access), plus bulk mail and data leaving the company (early access), each in the coverage table below12,13,14,15 | Platform pages, aegisai.com |
| Who builds it, and who backs it? | The team behind Google Safe Browsing and reCAPTCHA, backed by Battery Ventures, Accel and Foundation Capital | Demo page, aegisai.com |
| How do I see it on my own mail? | Ask for a tailored demo18 | Demo page, aegisai.com |
Pick the three rows that matter most to your team and test them first.
Short on time? Watch the explainer, or book a demo.
AI-powered email threat detection uses trained models to judge whether an email is an attack, then acts on that judgment. Microsoft lists machine learning models to detect phishing and spam among its email detection technologies22.
Here is how one email moves through the product:
The email is delivered to the mailbox. AegisAI reads mail through the Microsoft 365 or Google Workspace API after delivery.
The agents read what the message asks the reader to do, and whether that fits the sender.
Reputation, intent, policy and headers are checked in parallel.
Links and attachments are opened in a sandbox, QR codes included, and the page at the end is read.
The checks end in one verdict, with a confidence score and plain-language evidence.
Convicted mail is pulled from every mailbox, forwards included.
Microsoft 365 also acts after delivery. In all organizations with cloud mailboxes, ZAP (zero-hour auto purge) retroactively detects and neutralizes malicious phishing, spam, or malware messages that were delivered to cloud mailboxes23. So reading after delivery is not unique to any one tool: ask every tool what evidence its verdict shows.
Ask for one real verdict and its evidence before you judge any tool on its model.
Yes: a detection tool must read each message to judge it. The access starts read-only in monitoring mode, and removal needs write access that admins can grant or revoke.
Grant read access first, and add write access only after your team has reviewed the verdicts.
Deployment is an API connection with no change to mail routing: no MX change (the DNS record that sends inbound mail to a server), no gateway. If the service goes down, mail keeps flowing, and detection pauses until the service returns. Plan for the pause.
Ask your mail admin to connect one tenant (your organization's Microsoft 365 or Google Workspace account) and note the time it takes.
Phishing links and malware files go through the sandbox step above. Business email compromise and account takeover each have their own check, and so do bulk mail and outbound data. The table lists each one with the page that states it.
| Threat | What it does | Source |
|---|---|---|
| Business email compromise (BEC, a payment or invoice scam sent by email) | AI agents check every payment request against the sender's history, and stop the business email compromise that authentication lets through | BEC attack prevention |
| Account takeover (early access) | Agents read sign-ins and mailbox activity per account, and raise one explained finding when behavior stops matching the owner | Account takeover protection |
| Graymail (newsletters and bulk mail that is not an attack) | Moves newsletters and bulk mail out of the inbox. There is no quarantine queue to review | Platform overview |
| Data loss (DLP, data loss prevention; early access) | Flags payroll files, API keys and customer data sent to personal or unknown addresses | Platform overview |
Match each row to an incident your team handled this year, and test those rows first.
It is built for organizations whose mail runs on Microsoft 365 or Google Workspace, the two platforms it connects to. A team that worked on email security at Google builds it. The founding post states: "Our team brings together over 25 years of combined experience working on email security at Google"21. The seed announcement of 12 September 2025 states that Accel and Foundation Capital led a $13M Seed Round for AegisAI Security20.
Read both posts if your vendor review asks who stands behind the product.
AegisAI runs the six steps above on your own mail. Connect a tenant in monitoring mode and compare what AegisAI finds with what your current tools caught9. Nothing is removed until you grant write access.
Ask for a demo, then run it read-only on your own tenant. The demo page offers a tailored demo of how the platform can help you reduce work for your team and catch more attacks.
Bring three recent suspicious emails to the demo and ask to see their verdicts.
Track three numbers from your own trial.
Compare the three numbers on day 1 and day 30.
In AegisAI, a team of AI agents investigates every message. The six steps above show how one email is judged and removed.
No. It moves newsletters and bulk mail out of the inbox. There is no quarantine queue to review.
Yes, when the email asks for money. AI agents check every payment request against the sender's history.
The founding post states that the team brings together over 25 years of combined experience working on email security at Google.