OSINT and exposed credentials
Correlate public-source signals and exposed credentials related to the scope to investigate entry points and attack paths.
Ares by Vultus · Pentest AI
Autonomous pentesting with AI.For digital surfaces.
Investigate your digital surface like an adversary. AI agents, OSINT, exposed credential signals, and mobile testing connect exposures to attack paths with demonstrable impact.
Multi-agent adversarial architecture and a proprietary harness with more than 18 years of experience and Vultus Red Team TTPs. Authorized scope, technical evidence, and human supervision.
Read the Ares Manifesto ↓- Multi-agent AI
- Digital surfaces
- 18+ years of Red Team experience
From surface to attack path
What is autonomous AI pentesting?
An offensive investigation in which AI agents gather context, formulate hypotheses, and validate attack paths across an authorized digital surface.
Ares connects discovery, exploitation, and evidence to demonstrate how exposure can become impact. Vultus specialists guide operations and translate findings into remediation priorities.
Applications, APIs, and integrations
Investigate authentication, permissions, and business flows, connecting application behavior to adversarial hypotheses.
Mobile and reverse engineering
Combine virtualized environments, reverse engineering, and advanced language models to extract and analyze information relevant to testing.
AI Pentesting in Action · A Vultus Series
ARES KILLCHAINS BREAKDOWN
Kill chains executed by Ares, explained by Vultus. See how an initial exposure connects to attack paths and potential business impact.
A kill chain is the sequence of stages in an attack. In this series, documented offensive simulations show how Ares tests hypotheses, gathers evidence, and guides remediation.
From exposed credentials to cloud risk
A simulation that started with an exposed credential and reconstructed paths to personal data and cloud infrastructure. See how Ares connected evidence to demonstrate potential impact.
Watch on YouTube ↗PortugueseOne request, one attack chain
An operation initiated by a natural-language request, with 7 hours and 30 minutes of autonomous execution. The episode explains how attack hypotheses became a documented exploitation chain.
Watch on YouTube ↗PortugueseVibe coding: the AI-created vulnerability
A flaw introduced through vibe coding became the starting point for an offensive simulation. Learn how Ares identified the gap and gathered evidence to guide remediation.
Watch on YouTube ↗PortugueseFrom profile photo to domain
A profile photo upload in a forgotten application opened an attack path in an autonomous simulation. See how Ares connected initial access to escalation all the way to domain administrator.
Watch on YouTube ↗Portuguese
Ares Manifesto · Vultus's Vision
The era of autonomous attacks has begun
AI has changed the speed of offensive discovery. Defense must keep pace.
Flaws once difficult to connect become part of attack paths that can be investigated at scale. The Ares Manifesto presents Vultus's vision for responding to this shift.
- Continuously investigate the digital surface.
- Demonstrate the impact of attack paths, beyond listing flaws.
- Combine autonomy, governance, and specialist review.
Case studies and press
Public case studies of autonomous AI pentesting
Experiences published by Security Leaders show autonomous testing in different sectors. EXAME presents the technology and Vultus's expansion.
Reproduction of the article · EXAME / BusinessFeatured · EXAME
Vultus receives R$8 million to scale AI security testing
The article introduces Ares and Vultus's strategy to turn offensive security experience and proprietary technology into testing at greater scale.
Read the report in EXAMESecurity Leaders ·
MV
38 vulnerabilities in five hours
To meet a tight deadline, the healthtech assessed a critical scope with Ares. According to the report, eight findings were critical and two attack chains were demonstrated, with information for reproduction and remediation.
Read the case study in Security Leaders ↗PortugueseSecurity Leaders ·
Unidas
26 vulnerabilities in three hours
During a testing window planned to reduce operational impact, the rental company found flaws not identified in the traditional assessment. The article reports six critical findings and evidence that helped guide remediation.
Read the case study in Security Leaders ↗PortugueseSecurity Leaders ·
Arezzo&CO
89 vulnerabilities in six hours
A critical application that had already undergone traditional assessments was investigated by autonomous agents. According to the published account, the test identified 21 critical and 29 high-severity vulnerabilities, with documented paths to guide the team's remediation.
Read the case study in Security Leaders ↗PortugueseThe reported results relate to specific scopes. Read each article for the full context.
How Ares works
Autonomy with control and specialist review
The starting point is an approved scope, with defined execution limits and objectives. The investigation records tested paths and findings so the team can review the evidence and take action.
Explore Ares →01 · Define the scopeAlign the authorized surfaces, objectives, and testing boundaries.
02 · Test attack hypothesesInvestigate real attack paths and validate exploitation in a controlled manner.
03 · Prioritize and remediateReview evidence with specialists and direct the response according to business impact.
Proprietary technology · Applied offensive expertise
Multi-agent adversarial architecture with a proprietary harness
Multiple AI agents operate within an architecture designed for offensive investigation. The harness is the orchestration layer built by Vultus to apply its Red Team knowledge to testing.
This technology brings together more than 18 years of offensive security experience and the tactics, techniques, and procedures (TTPs) of the Vultus Red Team. The goal is to test attack paths relevant to the environment and produce evidence that helps prioritize the response.
01 / contextSurface
authorizedOSINT · applications · APIs · mobile
authorizedOSINT · applications · APIs · mobile
02 / investigationHarness
Vultus
Vultus
HypothesesValidationContext
03 / decisionEvidence
of impactExpert review · remediation
of impactExpert review · remediation
18+ years of experience
Offensive knowledge accumulated by the Vultus Red Team and incorporated into Ares's proprietary harness.
Dozens of active clients
Brazilian technology used by companies, with public case studies documenting autonomous testing in practice.
Vultus recognized by ISG
Leader quadrant in Risk-Based Vulnerability Management in ISG Provider Lens™ Cybersecurity Brazil 2025.
See the ISG recognition ↓Mobile investigation
Reverse engineering and AI for deeper testing.
Ares also tests mobile applications using virtualization and reverse engineering environments supported by advanced language models.
From artifacts to attack hypothesis
Extracting and analyzing application information helps agents understand its structure and behavior. This context informs new hypotheses and the investigation of exposures and attack paths.
Results feed into the same adversarial operation: validation, technical evidence, and remediation guidance, within the defined scope.
Offensive experience · Independent recognition
Knowledge that turns evidence into decisions
Vultus in the Leader quadrant for Risk-Based Vulnerability Management.
Recognition in ISG Provider Lens™ Cybersecurity Brazil 2025 reinforces Vultus's approach: combining technical depth with business insight to prioritize risk.
This Red Team and vulnerability management experience guides the development of Ares's proprietary harness.
A complementary operation
AI pentesting, scanners, and manual penetration testing: what is the difference?
Scanners help identify potential vulnerabilities. Penetration testing aims to demonstrate how a flaw can be exploited and what its impact is. In Ares, agents test offensive hypotheses and document evidence to support that validation.
Autonomous investigation also uses context, hypotheses, and attack chaining. Pentesters and red teamers guide objectives, boundaries, and deeper investigation, combining agent capabilities with Vultus's offensive security experience.
When to use
Offensive validation at the pace of change
- Evolving digital surfaces with multiple assets and integrations.
- OSINT investigations and exposed credentials related to the scope.
- Application, API, and mobile testing before and after significant changes.
- Recurring validation, retesting, and attack path prioritization.
Frequently asked questions
Questions about autonomous pentesting
Which digital surfaces can Ares test?
The operation can cover applications, APIs, integrations, and mobile, as well as OSINT investigation and exposed credentials related to the scope. For mobile, Ares combines virtualization, reverse engineering, and advanced language models to extract and analyze information relevant to the test. Targets, access, and objectives are defined for each operation.
Does AI pentesting replace specialists?
Ares operations combine automation with human supervision. Specialists maintain contextual analysis, validation, and deeper technical investigation when necessary.
Is autonomous pentesting the same as a vulnerability scanner?
Ares aims to validate exploitable risks through attack hypotheses and technical evidence. Scanners help identify possible flaws; offensive validation deepens understanding of how those flaws may affect the environment.
How is the testing scope defined?
Surfaces, objectives, and execution limits are agreed upon before the operation. The demonstration is conducted within approved scope and overseen by Vultus.
What does multi-agent adversarial architecture mean?
It means multiple AI agents operate within an architecture designed to test attack hypotheses. In Ares, Vultus's proprietary harness coordinates their work and incorporates its Red Team's experience and TTPs, within an authorized scope and under human supervision.
Can Ares be used between periodic pentests?
Yes. The aim is to maintain a recurring layer of offensive validation, including application changes and retesting after fixes, complementing periodic assessments.
Talk to Vultus
Understand where AI pentesting fits in your environment
Discover Ares and define a demonstration scope with our specialists.