31 questions, answered from the data
- How many platforms does the register list?
- Fourteen: Astelia, Averlon, Backline, Cogent Security, Dux, Maze, Onit, Qualys, Rapid7, Rival, Seemplicity, Tenable, Vicarius and Zafran, in alphabetical order. Each has a profile with the same seven topics. See /platforms/ for the list and /method/ for how a platform is added.
- Does the site order platforms by merit?
- No. Platforms are listed alphabetically, and no page carries a rating, a total or a pick. Facts sit side by side under seven topics, and the reader decides which rows matter. The rule is stated in plain words on /method/, and the guides at /guides/ name what to check, not what to buy.
- When were the facts read?
- 10 Nov 2026 for the first edition. Every fact shows its own read date beside its source link. Vendors change their pages, so a fact describes the page on that day. Corrections are listed with their date on /corrections/, and the method is on /method/.
- What does none found mean in a profile?
- It means a page was read on the stated date and the thing was not on it, for example named ticketing integrations. It is a statement about the pages read, not about the product. Each entry names the pages. Examples are on /platforms/astelia/ and /platforms/dux/, and the labels are explained on /method/.
- What are the seven topics?
- Proves exploitability in your environment, ships the fix, finds the owner, speed on a new disclosure, agents you can shape, works on the stack you have, and proof. Every profile uses them in that order. The guide at /guides/what-to-ask-in-a-remediation-platform-trial/ gives one test for each.
- What is a vendor's claim label?
- It marks a statement taken from the vendor's own site or release, such as a customer result or a speed figure. The register quotes it and gives the page. A third-party source label marks a statement from an outside outlet. Read /guides/reading-a-vendors-remediation-claim/ for how to test the claim.
- What does the Backlog Calculator compute?
- Analyst hours and working days. With its defaults of 20,000 findings, 25 minutes each, 3 analysts and 30 queue hours a week, the result is 8,333 hours and 463 working days. Every input is editable, and the assumptions are listed on /calculator/. It counts analysis time only, not fix time.
- Does the calculator store what I enter?
- No. The calculator runs in your browser and the numbers are not sent anywhere or stored. The site sets no cookies and runs no analytics, as the privacy page at /privacy/ states. The only record of a visit is the request log kept by the content delivery network.
- How do I find a good minutes-per-finding number?
- Time a sample. Take 30 findings your team closed last month, measure the minutes from opening to a recorded decision, and use the median. The 25-minute default is an input to replace. The method is in /guides/how-to-measure-a-backlog/, with a table of results at 5,000, 20,000 and 50,000 findings.
- What is agentic vulnerability management?
- Software agents carry a finding through steps a person used to do, within limits the customer sets. The test is what ships: a pull request, a patch job, a rule or a scoped ticket. The definition and a table of the outputs vendors state are in /guides/agentic-vulnerability-management-defined-by-what-ships/.
- Which platforms say they open pull requests?
- Averlon, Backline and Rival state that they generate pull requests for human review, as read on 10 Nov 2026. Cogent Security describes executing fixes through existing tools. Each claim and its source are on the profiles, and /guides/what-an-agent-can-fix-without-a-human/ sorts them by output.
- Which platforms say they deploy patches?
- Qualys, Tenable and Vicarius state that they deploy patches through their own patch tooling. Astelia states that it drives patches through the customer's patch management tool where one exists. All are vendor claims read on 10 Nov 2026. See /platforms/qualys/, /platforms/tenable/ and /platforms/vicarius/.
- Which platforms let customers build their own agents?
- Astelia describes customers writing a workflow in natural language for a Planner agent to build, and Rival, Tenable and Zafran state a customer-built agent feature. Several others record none found on 10 Nov 2026. The agents topic on each profile at /platforms/ holds the quotations.
- Do remediation platforms replace the scanner?
- The agentic platforms in the register mostly sit after the scanner and read its findings; Astelia and Backline state this directly. Qualys, Rapid7 and Tenable are scanner vendors that also describe agents. See /guides/agentic-vulnerability-management-defined-by-what-ships/ for where each type sits.
- Which platforms publish prices?
- None publishes a price list on the pages read on 10 Nov 2026. Maze states pricing by cloud infrastructure size, Rapid7 by usage per asset type and Vicarius by managed endpoint count. Most others record none found. The profile pages at /platforms/ show the line, and /guides/what-to-ask-in-a-remediation-platform-trial/ covers asking.
- How fast do attackers exploit new vulnerabilities?
- Rapid7's Q3 2026 report puts the average at three days from public disclosure, with 71% of tracked exploitation inside seven days. That is Rapid7's research, read 10 Nov 2026. The consequences for policy are in /guides/the-remediation-sla-that-holds/.
- What deadline does CISA set for known exploited vulnerabilities?
- Two weeks for CVEs assigned after 2021 and six months for older ones, as the default under BOD 22-01 for federal agencies. The directive allows adjustment in cases of grave risk. Private companies are not bound by it, but /guides/the-remediation-sla-that-holds/ shows how to borrow the structure.
- How should mean time to remediate be reported?
- As a median and a 90th percentile per tier, with the count and age of open items past SLA, and with the clock start and stop stated. A mean of closed items alone hides the long tail. The full definition is in /guides/mean-time-to-remediate-measured-honestly/.
- How should approval work for agent changes?
- Set it per action type: read-only steps run alone, reversible changes run in staged rings, irreversible changes wait for a person, and every approval is logged with the evidence shown. Cogent Security, Onit and Qualys describe such gates. See /guides/how-approval-should-work/ for the table.
- How do I check whether ownership mapping works?
- Take 30 findings from last month, write down the team that actually closed each, and compare with the platform's answer. Count right team, wrong team and no answer. Onit claims about half of tickets bounce for wrong owners, which is its own observation. See /guides/ownership-mapping/.
- What should a lean team check first?
- How many findings the platform closes before a person reads them, and what evidence it shows. At 25 minutes a finding, 20,000 findings take about 8,333 analyst hours. The calculator at /calculator/ sizes your own queue.
- What should an enterprise with ServiceNow check?
- Whether each scanner you run and each ServiceNow module has a named integration, and how the platform resolves an owner for an asset with no CMDB entry. The guide /guides/the-remediation-sla-that-holds/ covers the clock choices.
- What should a regulated company ask for?
- An export of one exception and one fix with the full trail: evidence, approver, timestamps and the re-check after the change. NIST SP 800-40 Rev. 4 lists identifying, prioritizing, acquiring, installing and verifying as the steps an auditor expects. See /guides/how-approval-should-work/.
- What should a managed service provider check?
- Tenant separation, per-customer approval rules and per-customer reports, before any agent. Maze states that single-tenant hosting is available (vendor's claim, read 10 Nov 2026). The stack topic on each profile lists integrations.
- Are there side-by-side pages?
- Yes. Each page puts the sourced facts of two platforms in the same rows under the seven topics and ends without a conclusion. Pairs are named alphabetically. The index is at /compare/, and each page shows the read date, 10 Nov 2026, next to every fact.
- What is the difference between reachable and exploitable?
- Reachable means an attacker can get to the vulnerable code or service. Exploitable means they can use the flaw there, given version, configuration and controls. A flaw can be reachable and not exploitable. Definitions and sourced examples are at /glossary/#reachability and /glossary/exploitability/.
- What happens when no patch exists?
- Teams use controls such as a firewall rule, a web application firewall rule, a disabled feature or a virtual patch, record them as an exception with an expiry date, and plan the fix. PHP 8.1 and older had no upstream fix for PHP2Shell on 21 Oct 2026.
- Where can I see recent platform news?
- On /news/, which lists 17 dated items from 2 Sep to 18 Nov 2026, newest first, each linked to the original announcement. They include Cogent Security acquiring Specular on 10 Nov 2026 and Rapid7's Q3 2026 threat update on 21 Oct 2026. Order is by date, not importance.
- How do I report an error?
- Write to editor@automatedvulnerabilityremediation.com with the page URL and the line that no longer matches its source. Corrections are made on the page and logged with their date on /corrections/. Vendors may send a newer source for any fact about their own platform; the entry is updated with a new read date.
- Does the site use cookies or analytics?
- No cookies, no analytics and no third-party scripts, as of 10 Nov 2026. The only record of a visit is the content delivery network's request log, which holds the IP address, time and page requested. Details are on /privacy/, and the calculator runs entirely in the browser.
- Can I reuse the Remediation Register data?
- Yes, with the cite line shown on the register data page: Automated Vulnerability Remediation, Remediation Register, edition 10 Nov 2026. The file is available as CSV and JSON, sorted by vendor and topic, with no ratings. The terms are on /terms/, and the edition date matters because rows describe pages as read on that day.
Related pages
- News
- Remediation Register: the data, downloadable
- How the register works, and what it leaves out
- The Backlog Calculator
Sources
3 sources, each with its read date
- Astelia Security: www.astelia.io/platform, read 10 Nov 2026
- Averlon: www.averlon.ai/platform, read 10 Nov 2026
- Backline Security: backline.ai/product/, read 10 Nov 2026
Last reviewed 10 Nov 2026.