Updated Reviewed by Charlton Swearingen, CBO
The main difference between LimaCharlie and Sublime Security is where email results go. Both connect to Microsoft 365 and Google Workspace by API with no MX change. LimaCharlie puts every message, verdict, and action in the same tenant as the customer’s EDR, identity, and cloud data, so one rule can follow a phish to the laptop that opened it.
Sublime is a strong, programmable email security product that runs as its own console.
The quick verdict
At a glance
The main differences between the platforms, side by side.
| Capability | LimaCharlie | Sublime Security |
|---|---|---|
| Platforms | Microsoft 365, Google Workspace, Gmail, and other IMAP mailboxes | Microsoft 365, Google Workspace, and IMAP [1] |
| Deployment | API, post-delivery, no MX or routing change | API post-delivery by default. Inline option on Enterprise [1] |
| Verdicts | Malicious, suspicious, graymail, or benign, with a 0 to 100 score and the signals behind it | ML and MQL detections [1] |
| Attachments | Archive unpacking, macro extraction, QR decoding, OCR, and YARA | Yes [1] |
| Custom rules | The same rule language as the rest of LimaCharlie, backtested against stored mail | MQL, plus an open-source rule feed [1] [2] |
| User-reported phishing | Abuse mailbox becomes a triage queue with SLAs, reporter replies, and AI investigation | Yes, with an AI triage agent [1] |
| Correlation with EDR, identity, and cloud | Native: same tenant, same rules, same cases | Separate product |
| Multi-tenancy | Every customer is its own tenant, managed with the rest of your fleet | Multi-tenancy API, in beta [1] |
| Pricing | Usage-based, billed daily. See current pricing → | AWS Marketplace lists $8.75 per mailbox per month (250 to 2,499 mailboxes) [3] |
Detection and triage
LimaCharlie scores every message from 0 to 100 and shows the signals behind each verdict, such as a lookalike domain, a first-time sender, or a credential form behind a link. Every flagged message also gets an AI second opinion, written back to the message once you trust it. Sublime uses ML and MQL detections with built-in AI agents.
| Capability | LimaCharlie | Sublime Security |
|---|---|---|
| AI triage | Runs through ask ai in your rules, on Anthropic, OpenAI, Google, OpenRouter, Bedrock, or Vertex, with your own instructions | Built-in AI agents [1] |
| Custom rules | Same language as endpoint and cloud rules, backtested before you enable them | MQL [1] |
| Rule content | A managed rule pack, plus your own rules | Open-source community rule feed [2] |
Response
LimaCharlie groups related messages into campaigns. You preview every copy across every mailbox, then pull them all with one confirmed action. Everything starts in alert-only mode, and you turn on automatic quarantine per customer when you trust the tuning.
| Capability | LimaCharlie | Sublime Security |
|---|---|---|
| Actions | Quarantine, trash, move to spam, restore, and warning banners, with campaign-wide clawback | Yes [1] |
| User reports | A triage queue with SLAs and reporter replies, plus an AI agent that checks sender history, similar mail, and campaigns | Yes, with an AI triage agent [1] |
Running it as an MSSP
In LimaCharlie, every customer is its own tenant, managed with the rest of your fleet, and email alerts land in the same cases as endpoint, identity, and cloud alerts. Sublime offers multi-tenancy through an API that is in beta.
| Capability | LimaCharlie | Sublime Security |
|---|---|---|
| Multi-tenancy | Built in, alongside EDR, SIEM, and cloud security | Multi-tenancy API, in beta [1] |
| Correlation | One detection and one case for a phish delivered, a link clicked, a new binary, and an impossible-travel login | Separate product |
API and AI access
Both platforms publish APIs. Sublime has a public REST API with an OpenAPI schema and a multi-tenancy API, and its AI agents write and backtest MQL detections and triage reported messages. LimaCharlie runs AI through ask ai inside any rule, on the model you choose, and its agents also see the customer’s endpoint and identity data.
| Capability | LimaCharlie | Sublime Security |
|---|---|---|
| API coverage | Every console action is an API call: REST API, LimaCharlie CLI, Python and Go SDKs, and an MCP server | REST API with an OpenAPI schema, plus a multi-tenancy API [4] |
| AI agents | AI agents started by detections, investigating across email, endpoint, identity, and cloud | ADÉ writes and backtests MQL detections, and ASA triages reported and flagged messages [4] |
| Model choice | Anthropic, OpenAI, Google, OpenRouter, Bedrock, or Vertex | Built-in agents [1] |
Pricing model
LimaCharlie Email Security is usage-based and billed daily, so you pay only for protected mailboxes, and you can add it to a customer whenever you need it. Sublime lists $8.75 per mailbox per month on AWS Marketplace for 250 to 2,499 mailboxes.
| What’s included | LimaCharlie | Sublime Security |
|---|---|---|
| Price model | Usage-based, billed daily | Per mailbox per month [3] |
| Same tenant as EDR, SIEM, and cloud | Included | Separate product |
| Buying model | Annual agreement or month to month | Listed on AWS Marketplace [3] |
Why MSSPs switch
01
Every message, verdict, and action is telemetry next to the endpoint, identity, and cloud logs. A phish, a click, a new binary, and an impossible-travel login become one detection and one case.
02
Analysts see the score and the signals behind it, so they can resolve a ticket without opening the raw email.
03
You pick the model, write the analyst instructions, and decide when the AI verdict can change the message.
04
Detection engineers write email rules in the same language as endpoint and cloud rules, and backtest them before turning them on.
It’s a very low barrier to entry. I’d say you can go in and experiment with it, and if you don’t like it, that’s fine. And you’re not stuck with it for a year... you don’t have to commit to a thousand endpoints for a year to just give it a shot.

Co-founder, Soteria
LimaCharlie was really designed with service providers in mind. It was multi-tenant. More important for me, it was designed to be built upon and extended and expanded.

CTO, Recon InfoSec
We’re very much in our SOC, like an MSSP, in the fact that we have multiple customers with multiple different levels. Also, when we’re looking at licensing, getting down to brass tacks, the model for a lot of vendors is not flexible to work with us as we scale up.

CEO, Black Hills Information Security
FAQ
Compare more tools
Book a demo with our team, or open a free tenant and connect a test endpoint today.
Sublime Security details come from the public sources below, with the date of each source. Product names belong to their owners.