A trial answers seven questions, one per topic, and each needs a test on your own findings
Set up the trial before the first call
Fix the data and the baseline first. Export a sample of 200 open findings that your team has already triaged, with the disposition each received. Record your current median time to decide, the share with no owner, and the number of tickets reassigned. The trial is a comparison against these, not against a demo.
Ask the vendor which integrations it needs and how long they take, since a trial that spends ten days on setup answers little about fixing.
Seven tests, one per topic
The register groups its facts under seven topics. Each has a test.
| Topic | Test | Evidence that answers it |
|---|---|---|
| Proves exploitability in your environment | Take 20 findings the platform dismissed and check each against your configuration | The evidence shown per dismissed finding, and whether you agree |
| Ships the fix | Ask for fixes on 10 findings of three types | Pull requests, patch jobs or rules you could apply, with test results |
| Finds the owner | Run 30 findings through ownership resolution | Share assigned to the right team on arrival |
| Speed on a new disclosure | Time a real disclosure during the trial | Minutes from disclosure to a list of your exposed assets |
| Agents you can shape | Change an approval rule and an agent behavior | The setting, the log entry and the undo |
| Works on the stack you have | Connect every scanner and ticketing tool you use | Named integrations working, not listed |
| Proof | Ask for a reference customer of your size and the method behind a headline figure | A call, and a written definition of the figure |
Questions to put in writing
- Which of the findings in my sample would the platform close without a person, and under what approval setting?
- What does the platform do when it cannot determine exploitability or ownership, and where does that item go?
- Which actions does the agent take on my systems, with what permissions, and how are they revoked?
- Where is my data processed and stored, and is it used to train a model? Maze, for example, states it does not train or fine-tune models on customer code or findings and offers single-tenant hosting (vendor's claim, read 10 Nov 2026).
- What is the price basis? Maze states pricing by the size of cloud infrastructure and Vicarius by managed endpoint count; Rapid7 describes usage-based pricing by asset type. Most vendors in the register publish no price.
Decide on the baseline, not the demo
At the end of the trial, repeat the baseline measures on the same sample. The useful results are numbers: minutes per finding, owner accuracy, findings removed with evidence you accepted, fixes you would merge. Write them down against the seven topics and compare with what the vendor's pages claimed.
Where the trial result and the vendor page disagree, the register is the place to send the source. Corrections are accepted at editor@automatedvulnerabilityremediation.com.
Agree the exit before you start
Write the pass and fail thresholds before the trial begins. For example: at least 80% of dismissed findings have evidence you accept, at least 70% of tickets reach the correct owner on arrival, and every approved change appears in the log with a working undo. The numbers are yours; the point is that they exist before the demo.
Also agree what happens to your data at the end. Ask for deletion in writing, and for the list of systems the platform was given access to, so each credential can be revoked on the last day.
Watch for what a trial cannot show
Two weeks will not show a quarterly patch cycle, a reorganization, or an audit. Ask for references that cover them, and ask what the vendor has seen go wrong in similar environments.
A trial also cannot show support quality after the contract. Read the documentation during the trial, since it shows what the product does when the sales team is not in the room. Where documentation is behind a login, as the register records for one vendor, ask for read access for the trial period.
Related pages
- Guides
- Agentic vulnerability management, defined by what ships
- Vulnerability ownership at scale: grouping, routing and expiry
- AI vulnerability management in the weekly remediation meeting
Sources
4 sources, each with its read date
- Qualys, TruRisk Eliminate, read 10 Nov 2026
- Maze, cloud platform page, read 10 Nov 2026
- Vicarius, pricing, read 10 Nov 2026
- Rapid7 Exposure Command pricing, read 10 Nov 2026
Last reviewed 10 Nov 2026.