Snapshot 228168

Document
Trust center
URL
https://www.eesel.ai/security
Fetched
HTTP status
200
Content type
text/html; charset=utf-8
Fetch mode
static
Size
260155 bytes
SHA-256 (raw)
6eecb32eebb0b0755474b4f3f3cecc88f62539c521df39c697c5920ee6a766e9
SHA-256 (normalized text)
4b3c9f0dcffacf24b3bb4f2eb45e40186a1f189fc763d594b6dee3b165c5497f

Normalized text

Scripts and page chrome removed; this is what change detection compares.

Trust Center
Security you can check, not just trust.
Everything we do to protect your data, written down plainly. Our security program runs on code, and its evidence goes to an independent SOC 2 auditor.
Request our security packSystem status
Compliance status
Updated November 2026
SOC 2 Type IISecurity, Availability and Confidentiality. Observation period started 1 October 2026.
Audit in progress
HIPAABusiness Associate Agreements for Enterprise customers, after a review of how you'll use eesel with health information.
Available
Security policies18 written policies, version 1.4, approved by leadership and reviewed every year.
Adopted
Data Processing AgreementA DPA covering GDPR, with Standard Contractual Clauses for international transfers.
Available
Daily control checks50 controls, checked by code every morning and recorded as audit evidence.
Running
Compliance
On the road to SOC 2 Type II
A Type II report doesn't just check that controls exist. It checks that they worked, every day, over months. Ours covers three of the five trust criteria: Security, Availability and Confidentiality.
September 2026
Program built
50 controls mapped to every SOC 2 criterion in scope, each with an owner and a defined piece of evidence.
October 2026
Policies adopted
18 security policies approved by leadership, covering everything from access control to AI governance.
1 Oct to 31 Dec 2026We are here
Observation period
Every control runs and is evidenced daily. This is the period the auditor tests.
Early 2027
Independent audit
An independent auditor tests our controls against the evidence from the observation period.
2027
Report available
Customers and prospects can read the full SOC 2 Type II report under NDA.
A compliance program written as code
We didn't buy a compliance dashboard. Our controls, policies and evidence collectors live in our own codebase, change only through reviewed pull requests, and run every morning. When a check fails, an engineer hears about it that day, not at audit time.
security controls
50
written policies
18
automated evidence checks
Daily
breach notification commitment
48h
Controls
How we protect your data
A summary of the controls behind our security program. The full policies are in our security pack.
Encryption
Our public endpoints accept only TLS 1.2 or higher.
Production databases and their backups encrypted at rest with AES-256.
File storage and message queues encrypted at rest.
Encryption keys managed in AWS KMS and rotated every year.
Access control
Least-privilege access for staff, approved before it is granted.
Access reviewed every quarter and removed within 72 hours when someone leaves.
Staff access to production data is read-only by default.
Two-factor authentication required across our source code organization.
Secure development
Every change goes through a pull request with required build, test and secret-scanning checks.
Changes to infrastructure, authentication, billing and permissions need a founder's approval.
Every code push is scanned for leaked secrets.
Dependencies are monitored for known vulnerabilities, with fix deadlines by severity.
Infrastructure
Hosted on Amazon Web Services.
All infrastructure defined as code. Every change is planned, reviewed and approved before it is applied.
Production deploys are started by a named person from reviewed code, and every deploy is recorded.
AWS threat detection (GuardDuty) and security benchmarks (Security Hub) enabled in every region.
Monitoring and logging
Production is monitored around the clock, and alerts page an on-call engineer.
Cloud account activity logged in every region, with tamper-evident log files kept in write-protected storage for a year.
Service logs kept for at least one year.
Public status page at status.eesel.ai.
Incident response
A written incident response plan covering triage, containment, root cause and fixes.
If a breach affects your data, we tell you within 48 hours of becoming aware of it.
Service incidents posted publicly on our status page.
People
Everyone with access to customer data signs confidentiality terms.
New hires are screened before they start.
Security training is required when people join and every year after.
Work devices must have disk encryption, automatic updates, a firewall and screen lock.
Vendor management
A register of every vendor that handles customer data, with an owner for each.
Each vendor's security reviewed every year.
A published list of our subprocessors, below.
AI and your data
Your data runs your agents. Nothing else.
AI is the product, so we hold it to the strictest rules in our program.
01
Never used for training
eesel does not train or fine-tune models on your data. AI providers process it only to generate your agent's responses, under commercial API terms that do not permit training on API data.
02
Isolated per workspace
Each workspace's data is kept separate. An agent can only search the knowledge you've given it, and automated tests check that boundary on every code change.
03
Only the actions you allow
Agents can only use the tools you turn on. Any action can be set to wait for a person's approval, and outbound requests are limited to domains you allow.
04
Credentials stay out of the AI
When you connect an app, the key goes straight to our servers through a secure form. The AI model never sees your passwords or API keys.
05
Optional PII redaction
Turn on redaction to strip personal details such as emails, phone numbers and card numbers from synced helpdesk tickets before they are stored and indexed.
06
Every run on the record
Every conversation and action your agents take is logged in your workspace's activity view, so you can see exactly what happened and why.
Data lifecycle
Where your data lives, and how it leaves
Hosting
Production runs on AWS in the United States (N. Virginia). Free and Teammate plans store data there. EU data residency is on the Enterprise plan: data stored in AWS Ireland with an EU search index, and the AI running in the EU.
Backups
Databases are backed up automatically every day and backups are kept for 35 days, encrypted. File storage is versioned.
Recovery
Our recovery targets are 4 hours to restore service and no more than 24 hours of data loss. Restore testing is part of our SOC 2 controls.
Deletion
Delete your workspace and its documents, search index, integrations, credentials, agents and conversation history are wiped from our live systems straight away. Encrypted backups roll off within 35 days, and security logs expire on a fixed schedule.
Subprocessors
Who helps us run eesel
The third parties that process customer data on our behalf. This list is the same one in our Privacy Policy, section 8.2.
Subprocessor	What it does
Amazon Web Services	Cloud hosting, databases and file storage. AI models through Bedrock, for workspaces on the Bedrock option
Anthropic	AI models
Baseten	Runs open AI models such as DeepSeek, with zero data retention
Browserbase	Hosted browsers for the agent's browser, when you switch it on, including apps you give it a login for. Until mid-November 2026 also renders HTML files you download as PDF
DeepInfra	Runs open AI models such as DeepSeek, with zero data retention
Firecrawl	Reading websites you add as knowledge
Fireworks AI	Runs open AI models such as DeepSeek, with zero data retention
Google	AI models, including removing personal details from text
Intercom	eesel's own customer support inbox
Meta (WhatsApp)	WhatsApp for your team, when you switch it on, where your teammates chat with their agent on eesel's WhatsApp number. Connecting your own WhatsApp Business number is an app you connect, not a subprocessor
Microsoft Azure	Runs DeepSeek models that score finished conversations
OpenAI	AI models
Pinecone	Search index of the knowledge you connect
Pipedream	Connections to the apps you choose to connect
PostHog	Product analytics, and session replay of the dashboard with all text and typing masked
SendGrid (Twilio)	Sending and receiving email
Slack	eesel's own team workspace, for internal alerts and support
Together AI	Runs open AI models such as DeepSeek, with zero data retention
Typesafe AI	Checks an action an agent is about to take against what it was asked to do
Vercel	Hosts the dashboard and the sandboxes where agents run
Vercel (AI Gateway)	Routes requests to the AI model providers on this list
Auth0 (Okta)	Sign-in to the dashboard
Datadog	Monitoring and logs
Google Analytics	Website and dashboard analytics
Stripe	Payments
Apps you connect yourself, such as Zendesk, Shopify or Slack, are not subprocessors. They are systems you chose, and eesel reads from and writes to them on your instruction. See the full list, with locations and a feed of changes
Documents
Everything your security review needs
Email us and we'll send what you need, usually within one business day. Some documents are shared under NDA.
SOC 2 Type II report
After audit
Available under NDA once the audit is complete. Ask us for a status update in the meantime.
Request
Security policy pack
Under NDA
All 18 of our security policies in one document.
Request
Data Processing Agreement
On request
Our DPA, including Standard Contractual Clauses.
Request
Security questionnaires
On request
Send us your questionnaire (CAIQ, SIG Lite or your own) and we'll fill it in.
Request
Mutual NDA
On request
Our standard mutual NDA, or we'll sign yours.
Request
security.txt
Public
Our machine-readable security contact, published to the RFC 9116 standard.
Open
Responsible disclosure
Found a vulnerability? Tell us.
Email engineering@eesel.app with the details and how to reproduce it. We acknowledge credible reports within one business day and keep you updated until it's fixed. Please give us a reasonable chance to fix the issue before you share it, and don't access other customers' data or disrupt the service while testing.
engineering@eesel.app
Security FAQ
Our SOC 2 Type II audit is in progress. The observation period runs from 1 October to 31 December 2026 and covers Security, Availability and Confidentiality. The independent audit follows in early 2027, and the report will be available under NDA once it's issued. We're happy to walk you through our controls in the meantime.
Yes. We sign Business Associate Agreements with Enterprise customers, after a review of how you'll use eesel with health information. Talk to us before you connect any protected health information so we can set up your workspace correctly. A workspace under a BAA runs every AI model call on AWS Bedrock in eesel's own AWS account. Tools that would send its content to a vendor without a BAA, such as web search, are turned off.
No. eesel never trains or fine-tunes models on your data. AI providers process it only to generate your agent's responses, under API terms that exclude training.
eesel picks the model for each task and moves to better ones as they come out. Agents run on open-weight models such as DeepSeek, run for eesel by Baseten, Together AI, Fireworks AI and DeepInfra with zero data retention, or on Anthropic's models. DeepSeek, the company, receives no data. Both go through Vercel's AI Gateway. OpenAI handles search embeddings and filters which tickets reach an agent. When redaction is on, Google removes personal details from synced tickets before they are indexed. On the Bedrock option, every model call runs on AWS Bedrock in eesel's own AWS account instead.
Yes, as an add-on. Your workspace's AI model calls then run on AWS Bedrock in eesel's own AWS account, not with the other AI providers on our subprocessor list. Tools that would send your content to other services, such as web search, are turned off. Bedrock can run in the United States, or in the EU on the Enterprise plan if you need EU processing. It is the same setup our HIPAA workspaces use. AWS is still a subprocessor, and your agents still run in sandboxes hosted by Vercel.
On AWS in the United States (N. Virginia). Workspaces hosted in the EU are stored on AWS in Ireland, with their search index in the EU. Free and Teammate plans are stored in the US; EU data residency, with the AI running in the EU too, is on the Enterprise plan.
Yes. Our public endpoints accept only TLS 1.2 or higher. Our databases, their backups, file storage and message queues are encrypted at rest with AES-256, using keys managed in AWS KMS.
Yes. Deleting your workspace wipes its documents, search index, integrations, credentials, agents and conversation history from our live systems straight away. Encrypted backups roll off within 35 days, and security logs expire on a fixed schedule. You can also ask us to delete your data and we'll confirm when it's done.
Within 48 hours of becoming aware of a breach that affects your data. Service incidents are posted on status.eesel.ai.
Yes. We offer a DPA with Standard Contractual Clauses for international data transfers. Email us and we'll send it.
Yes. Send it to engineering@eesel.app and we'll work through it with you.
Still have more questions?
Running a security review?
Talk to the people who run our security program. We'll send our security pack, answer your questionnaire and get on a call with your security team.
Email our security teamBook a demo