Protect: Keep all of it secure.
AI & Cybersecurity
You automated it. Now make sure it is safe.
Security testing and hardening aligned to the OWASP Top 10 for websites and web apps, and AI security reviews against the OWASP Top 10 for LLM applications: prompt injection, data leakage and unsafe agent actions.
The problem it removes. Every new AI agent and integration is a new door, and nobody has checked whether it locks.
How it works
- 1
Agree the scope in writing
Exactly which sites, apps and AI features are in scope, and written permission to test them. Nothing is tested without it.
- 2
Test against the OWASP Top 10
Injection, broken access control, misconfiguration and the other common failures, checked methodically.
- 3
Test the AI features
In the AI review, prompt injection, data leakage and unsafe agent actions, against the OWASP Top 10 for LLM applications.
- 4
Report in plain language
Every finding ranked by risk, with what it means for the business and exactly how to fix it.
- 5
Fix and retest
In the hardening and AI packages we help apply the fixes, then retest to confirm they hold.
Three standard packages
Standard packages start from ₹3.0L. Custom engagements are scoped separately. Scope and price are agreed in writing before any work starts. The figures here are in rupees; each market has its own price list, shown on its market page.
Starter
1 week
from₹3.0L
Vulnerability assessment
- OWASP Top 10 aligned testing
- Injection and XSS checks
- Misconfiguration review
- Prioritised report in plain language
- Walkthrough call on the findings
Growth
2 weeks
from₹6.0L
Security hardening
Everything in Vulnerability assessment, plus:
- CSP, HSTS and security headers
- Input sanitisation and bot protection
- GDPR and DPDP data-handling review
- Retest after the fixes
Premium
Scoped, from 3 weeks
from₹10.0L
AI security review
Everything in Security hardening, plus:
- Prompt injection testing
- Data leakage review of AI features
- Agent permission and tool-use review
- OWASP Top 10 for LLM applications audit
- Remediation support and retest
What the price covers
- Discovery, scoping and the written plan
- Design, build, integration and testing
- Setup on your own accounts, and the handover with documentation
- The support period stated in the package
What it does not
Running costs stay on your own accounts and are billed to you by the providers, not by us: WhatsApp Business Platform messaging, AI usage, phone numbers and call minutes, CRM or automation plans, hosting and the domain. We list the ones your system needs, with what drives each one, before you commit.
Advertising budget, content or media you want written or produced, and anything added after the scope is signed are quoted separately.
What we need from you
- Written authorisation to test the systems in scope
- A test account for each user role
- A contact for urgent findings
- Access to a staging copy, where one exists
Your data
Testing is done only with written authorisation and within the agreed scope. Findings are shared only with you, and any data seen during testing is not kept.
Everything runs on accounts you own, and is handed over with documentation. You can keep us on, or not.
Built with
- OWASP Top 10
- OWASP Top 10 for LLM Applications
- Burp Suite
- OWASP ZAP
- Nuclei
- Mozilla Observatory
Chosen per project for what you already use. Naming a tool here implies no partnership.
Related work
Client project
Manufacturing: hydraulic presses and rubber moulding, Faridabad, India
Barsat Engineers
A 30-year manufacturer with ZED Gold certification and no digital presence, getting a website, a WhatsApp enquiry system and secure foundations.
Read the case studyBuilt by Soarin Clevan
AI systems for revenue, automation and security, UK, UAE and Australia
Mantron
A website built to diagnose, quantify and convert without a salesperson: a free lead audit, a revenue calculator and a full lead pipeline.
Read the case studyQuestions we are actually asked.
Soarin Clevan builds AI automation systems that find leads, answer every enquiry, follow up and get you paid.
The OWASP Top 10 is a widely used list of the most critical web application security risks, published by the Open Worldwide Application Security Project. There is a separate Top 10 for applications built on large language models.
Prompt injection is an attack where text given to an AI system, by a user or hidden in a document or web page, overrides its instructions, for example to leak data or take an action it should not. We test for it in every AI security review.
Testing is scoped and paced to avoid disruption, and anything intrusive is agreed in advance or run against a staging copy.
The vulnerability assessment is a methodical test against the OWASP Top 10. Deeper, exploit-led penetration testing is scoped separately where a business needs it.
It connects to the rest of the system
- AI Lead GenerationYour pipeline fills while you run the business.
- AI ReceptionistA front desk on WhatsApp and your website that never logs off.
- AI Voice AgentEvery call answered. Every callback made. Even at 2am.
- Automated Follow-upEvery enquiry answered in under a minute, day or night.
- Payment RecoveryGet paid on time without chasing anyone yourself.
- Business AutomationAutomate the work that should not need a person.
- Websites & SearchA site that converts, found on Google and named by AI.
Tell us where the system is leaking.
A short description is enough to start. We reply with first thoughts, the questions that matter, and whether we are the right people for it.
- 1First thoughts on what you describe
- 2The questions that actually matter
- 3A straight answer on whether we fit
- Markets
- India, Australia, United Arab Emirates, United Kingdom, Brazil, South Africa