Lesson 2 of 8
The six tool types
The finder sorts tools into six types because each one answers a different question. Vendors cross the lines freely, so it helps to know where the lines are.
Why types before vendors
Ten vendors on this site claim between two and six of the types. Several have renamed their products in the last eighteen months. Picus launched its Autonomous Exposure Validation Platform on 7 July 2026, SafeBreach launched a CTEM platform on 22 April 2026, and Cymulate repositioned around what it calls agentic cyber defense engineering on 1 June 2026. Product names now tell you more about marketing cycles than about what a tool tests.
Types are steadier. A tool either tests controls one technique at a time or it does not. It either executes attack chains or it models them. Pick the type that fits your question, then check which vendors document it.
1. Breach and attack simulation
Scripted, safe attack scenarios run against your controls from agents or simulators you deploy. The output grades each control: prevented, detected or missed. It is the oldest type and the most mature for testing EDR, email gateways, web proxies, WAFs, DLP and SIEM rules. It says little about whether an attacker could chain weaknesses together. Read the full note.
2. Adversarial exposure validation
Gartner's umbrella category, which takes in both control testing and path testing. A product sold as AEV usually started as one or the other and added the second half. The question to ask is which half came first, because that half is usually deeper. Read the full note.
3. Automated pentesting
Software that runs a penetration test without a person driving each step: discovery, exploitation, credential reuse and lateral movement. It proves that a chain works by executing it. Most products here run without endpoint agents, from a host you provide or from the vendor's cloud. Read the full note.
4. Attack path analysis
Mapping how weaknesses connect into routes toward valuable assets. Some tools execute the paths, which overlaps with automated pentesting. Others model them in a graph built from configuration and identity data, which is safer and wider but is a prediction rather than proof. XM Cyber describes its approach as validating attack paths in a modelled environment rather than running attacks. Read the full note.
5. Exposure management platforms
Tools that gather vulnerabilities, misconfigurations, identity issues and attack surface data, then rank them by risk. They are the backbone of a CTEM programme. On their own they rank rather than prove, which is why several now add a validation module or import results from a validation tool. Read the full note.
6. Managed validation services
A vendor or partner runs the testing for you and hands over findings. It is a delivery model that can sit on any of the first four types. It suits teams without people to operate a platform. NetSPI's breach and attack simulation product became an expert-led service in 2026, which shows how thin the line between software and service has become. Read the full note.
Where the types overlap
The map on the finder home page shows that most vendors document three or more types. That is not a contradiction. A platform can run control tests from agents and path tests from a separate module, and sell both under one name.
The overlap does create two traps. The first is assuming that a vendor strong in one type is equally strong in another it lists. The research records on this site mark a claim as partial when the vendor's own material describes it narrowly, as a beta, or only in passing. Take partial claims as a reason to test that capability in the pilot, not as a reason to rule the vendor out.
The second trap is buying two tools that do the same job. A team that already owns an exposure management platform with attack path modelling may not need a second path tool; it may need control testing. A team with a mature BAS deployment may need executed paths, not more scenarios.
How the types map to questions
A short way to hold all six in your head:
- Do our controls work? Breach and attack simulation.
- Could an attacker get through, and how? Automated pentesting, or attack path analysis if you cannot run attacks.
- Both, under one programme? Adversarial exposure validation.
- What should we fix first across everything? Exposure management platforms.
- Who will run it? If nobody, managed validation services.
A note on labels
Vendors use the labels that sell. In 2026 that means "AEV", "CTEM" and "agentic". None of these is wrong, but none tells you what runs on your network. When a vendor describes its product, translate each claim into one of the two kinds of evidence from lesson 1: control evidence or path evidence. Then ask for a demonstration of each on your own estate.
The next lesson places all six types inside the CTEM programme model, which is where most buyers meet them.
Previous: What exposure validation is. Next: Where validation fits in CTEM.