Penetration Testing as a Service (PTaaS) is a delivery model for penetration testing. Instead of a one-off engagement that ends in a static report, pentesters assess the system on a recurring basis and publish each finding on a platform where the client’s developers and security team track it, discuss it and close it. PTaaS combines that manual work with automated scanning, and it keeps the testing scope open to changes in the system.
The model matters because of what human testing contributes. In Fluid Attacks’ assessments of client systems from January 1 to December 31, 2025, automated tools accounted for 55.8% of the risk exposure reported and pentesters for the remaining 44.2% — but for critical-severity vulnerabilities alone, 87% came from manual review.
This page covers what PTaaS includes and what it does not, how it compares with a traditional pentest and with a bounty program, what a PTaaS finding has to contain to be worth anything, and the questions that reveal whether a provider’s “continuous” testing is continuous in practice.
What is Penetration Testing as a Service (PTaaS)?
PTaaS is a service in which ethical pentesters find vulnerabilities in a system, try to exploit them and report the results continuously, through a platform rather than through a document. Fluid Attacks’ own comparison framework defines it in exactly those terms: the solution or service includes continuous penetration testing, with testers who find vulnerabilities, attempt exploitation and report what they find.
Three components make the model:
Recurring human testing: the testing team works on the system over time, not in a single window. New versions and newly discovered assets enter the scope.
A platform: findings, evidence, status, remediation guidance and the conversation between testers and developers live in one place, available while the assessment is still running.
Automation underneath: scanners cover known flaws, broad surfaces and speed; the human team works on everything else.
Be precise about what PTaaS is not. It is not a vulnerability scanner with a dashboard: if no human tests the system, the service is vulnerability scanning, whatever the interface looks like. It is not a traditional pentest uploaded to a web portal, either; moving the deliverable from a PDF to a browser changes the format, not the frequency. And it is not a bug bounty program, because the scope, the testers and the rules are agreed in advance instead of being open to whoever shows up.
The technique inside the model is manual penetration testing. PTaaS does not change how a pentester works. It changes how often that work happens, how fast the result reaches the developer, and who verifies the fix.
Why the delivery model changes the result
A delivery model decides when a vulnerability is found, how long it stays open and who confirms it is gone. Those three things, not the testing technique, are what separates PTaaS from an annual engagement.
The window between tests is where the risk lives
Software that ships weekly and gets tested yearly spends most of its life untested. Every release between two assessments is code no one looked at from an attacker’s point of view. Fluid Attacks’ analysis of continuous versus point-in-time testing makes the same argument: a point-in-time test describes the security posture of the day it ran. PTaaS closes that window by testing as the system changes.
Findings arrive while the code is still fresh
In a project-based engagement, every vulnerability — the one found in week one and the one found in week six — arrives at the same time, in the same document. Under PTaaS, each finding is published when the tester confirms it, so the developer who wrote the code is still the person who fixes it.
The platform is where remediation happens
A report describes vulnerabilities. A PTaaS platform turns them into vulnerability management: priority, assignment, status, evidence, the exchange with the tester who found the issue, and the record of every retest. The metric that matters is not how fast a provider detects a flaw but how long that flaw stays exploitable.
Automation alone does not get you there. According to Fluid Attacks’ documentation, automated scanning finds only 10% to 30% of the total risk exposure of a system; the rest needs manual testing, often because reproducing the issue requires an environment, authentication or test data. A PTaaS offering without that human layer is selling the platform, not the testing.

Fluid Attacks’ 2025 results: tools reported most vulnerabilities, pentesters reported most of the critical ones.
PTaaS at Fluid Attacks
Fluid Attacks delivers Pentesting as a Service in the Advanced plan of its all-in-one solution. What follows is what it does, with the rules and the limits.
What we test, and under what rules
PTaaS covers every attack surface that automated DAST covers — DNS records, headers, HTML content, SSL connections, unauthenticated HTTP endpoints, webhooks and WebSockets — and adds the surfaces that require a human: authenticated HTTP endpoints, GraphQL, gRPC, REST and SOAP APIs, and APIs described with OpenAPI, Postman or Swagger specifications. The testers authenticate to applications and follow those specifications, with access to the source code — an approach Fluid Attacks calls code-assisted pentesting. One documented limit: neither DAST nor PTaaS supports DOM APIs.
Testing runs under explicit rules, including in production. Fluid Attacks controls the intensity and pace of every test and uses no unauthorized brute-force or high-load techniques. Any test that could affect availability runs first in a non-production environment, and the client is notified and asked for a time slot. Tests always come from Fluid Attacks’ own IP addresses, so a false alert on a WAF or SOC can be traced. If a tester has any doubt during a test, the tester stops.
What it looks like in practice
What is instrumented | How it is tested | What the client gets |
|---|---|---|
Source code of the registered branches | Testers with code access, plus SAST, SCA and secret scanning | The commit where the vulnerability lives |
Web and API environments | PTaaS on authenticated endpoints and API specifications, plus DAST | The evaluated environment and the timestamp of the review |
Mobile binaries | Manual testing plus MAST | The uploaded binary version |
Cloud infrastructure | CSPM and manual review | The live state read through the API |
Every reported vulnerability | Exploitation by a pentester | Up to six images or videos as evidence, refreshed with each reattack |
Every fix | A reattack requested from the platform | Confirmation or reopening, with the full history kept |
Which system gets tested next is not decided by a calendar. Statistical models rank systems by risk — number and size of recent changes, which files changed most, how many authors touched them, the system’s vulnerability history — and five independent attack teams work from that ranking: Systems, Criticals, Escapers, Deterministic and Reattack. More than one can test the same system at once.
Findings reach developers where they already work, through integrations with IDEs, bug-tracking systems and the CI/CD pipeline. Inside the CI/CD pipeline, the CI Gate applies the organization’s own policies and blocks a release that breaks them.
What the service commits to
Fluid Attacks’ service-level agreement states three numbers that can be measured against it:
Availability: the platform and API are reachable at least 99.95% of the time.
Response: a median first response of less than 8 office hours for 90% of reattack requests, comments, events and incidents.
Accuracy: an F2 score of at least 90% on risk-exposure reports, and an F0.5 score of at least 90% on vulnerability reports.
All three are measured every quarter across every group in the organization, including deleted ones. The accuracy indicator is the unusual one: F2 weights false negatives — the vulnerabilities a provider missed — more heavily than false positives, which is the opposite of what a vendor would choose if the goal were to look good.
What “continuous” does not mean
Continuous does not mean a pentester reviews every commit, and no provider should let a buyer believe that. Fluid Attacks is explicit that it does not promise coverage: a method can test every file and still miss a critical vulnerability. Automated techniques scan 100% of applicable files in supported languages when the CLI integrations are active, but manual testing works from statistical risk prioritization instead, and the commitment is to a result — the 90% accuracy SLA — rather than to a percentage of files reviewed.
That distinction is worth carrying into any PTaaS evaluation. Continuous platform availability, recurring manual assessments, scheduled automated scanning and continuous human testing are four different things, and a provider that uses one word for all four is describing a brochure, not a service.
The reattack decides the rest
A fix that was never retested is an assumption. On Fluid Attacks’ platform, a vulnerability reported by a pentester does not close because a developer changed the code: for PTaaS, Secure Code Review, Reverse Engineering and AI SAST findings, the client requests a reattack and a pentester re-attempts exploitation. If the issue is still there, the finding stays open. Reattacks are unlimited in the Advanced plan, and the vulnerability keeps the full history of requested and verified reattacks, including the rejected ones.
Go deeper: Fluid Attacks’ analysis of continuous penetration testing explains how the testing scope adapts as new assets appear.
PTaaS, traditional penetration testing and bug bounty are not alternatives
PTaaS, a traditional engagement and a bounty program answer different questions, and the comparison buyers need is about scope, cadence and control.
PTaaS | Traditional penetration testing | Bug bounty | |
|---|---|---|---|
Cadence | Recurring, follows system changes | One engagement, usually annual or semi-annual | Open-ended, depends on researcher interest |
Who tests | A known team, under contract | A team assigned for the engagement | Anyone registered in the program |
How findings arrive | Published as they are confirmed | In a final report | As submissions, when they happen |
Scope | Agreed and adjustable | Agreed and fixed | Defined by the program, often narrow |
Retesting | Requested from the platform, done by a pentester | A separate, usually billable engagement | Not guaranteed |
Payment | Subscription | Per engagement | Per accepted finding |
Best for | Software that changes continuously | A point-in-time assessment for a specific date | Broad, opportunistic coverage on top of a program |
A bounty program rewards whoever finds something first, which produces real findings but no guarantee that anyone looked at the parts of the application that worry you most. A traditional pentest reviews an agreed scope on an agreed date. PTaaS repeats that review, with the same team keeping the context between cycles.
None of the three, PTaaS included, replaces automated scanning, and automated scanning replaces none of them. In Fluid Attacks’ benchmark of 36 third-party AppSec tools against an application seeded with 1,201 vulnerabilities, a pentester reported 1,076 of them — 89.6% recall at 100% precision — while the best-performing scanner in that test — Fluid Attacks’ own — reached 22.7% recall.
What a PTaaS finding must contain
A PTaaS finding is useful when a developer can reproduce it, understand its impact and verify the fix without going back to the provider. Anything less is an alert.
The weakness, classified: not free text, but an entry in a documented catalog with its description, impact, threat actor and remediation guidance. Fluid Attacks maps each finding to an entry in its public weakness database, such as business logic flaws, which are the clearest example of what tools miss and a tester finds.
The location: the commit for source code, the binary version for mobile, the evaluated environment and timestamp for web, APIs and cloud. A PTaaS finding without it is not actionable.
Proof of exploitation: the evidence a pentester attaches to show the issue is real, as images or videos, up to six files per finding, refreshed with each reattack and kept next to the PTaaS finding itself.
Severity in context: a score the team can prioritize with, informed by exploitability and adjusted when the system lacks the conditions that would make a known flaw exploitable.
Remediation guidance: steps the developer can act on, plus access to the tester. Fluid Attacks’ Advanced plan includes Talk to a Pentester sessions.
Status and history: current state, report date, and every reattack requested and verified.
PTaaS reporting should also produce an executive summary and a technical export.

Every PTaaS finding maps to a catalog entry like this one, with impact, threat and fix.
How to evaluate a PTaaS provider
Most PTaaS marketing uses the same adjectives. These ten questions separate providers, because each has an answer that can be checked.
Criterion | The question to ask |
|---|---|
Human testing | Do human testers assess the system, and what share of findings comes from them? |
Tester consistency | Are the testers employees under direct contract, and do they stay with my account between cycles? |
Expertise | How many offensive security certifications does the team hold, and can you name them? |
Cadence | What triggers a new manual assessment: a calendar, a release, a risk model? |
Attack surface | Do you test authenticated endpoints and API specifications, and what do you not support? |
Evidence | Does every finding include proof of exploitation I can reproduce? |
Chained exploitation | Do your testers combine vulnerabilities to show the real impact, or report them one by one? |
Retesting | Who verifies a fix, how many retests are included, and how fast do they start? |
Accuracy commitment | Is there a contractual accuracy figure, how is it measured, and does it weight false negatives? |
Operational security | Who controls the channel between the tester and my system, and under what agreement? |
Two of these PTaaS criteria deserve emphasis. Chained exploitation turns a list of medium-severity issues into a demonstrated path to impact, and it is where automation still falls short: a Fluid Attacks review of published research on LLM-based pentesting agents found that 83.3% of chained-exploitation samples stalled before completing a multi-vulnerability chain, and about 56.7% of CVE-exploitation samples identified the right CVE but failed to build a working payload.
Expertise is the other. Advanced exploitation is research work, not tool work. As Andrés Roldán, VP of Hacking at Fluid Attacks, put it in his review of the OffSec Exploitation Expert certification, “new research is needed to develop an exploit for a modern Windows OS build.” Fluid Attacks’ pentesters hold international certifications in multiple security testing techniques, and CREST has accredited the company for penetration testing.
NIST’s SP 800-53 Rev. 5 catalog (2020) lists penetration testing as a control of its own, CA-8, with enhancements for an independent testing team and red team exercises. The PCI Security Standards Council makes penetration testing mandatory in PCI DSS requirement 11.4, and 11.4.1 asks the entity to define, document and implement the methodology and to retain the results for at least 12 months. Worth saying plainly: PTaaS can produce the evidence those frameworks ask for, and the same holds for GDPR and HIPAA programs, but buying PTaaS does not make an organization compliant.
Conclusions
PTaaS is a delivery model, not a testing technique, and that is where its value sits. It puts human testers on the system repeatedly instead of once a year, publishes each finding when it is confirmed instead of in a quarterly document, and verifies fixes with a retest instead of an assumption. The parts of a PTaaS offering worth checking are the operational ones: who tests, how often, with what evidence, under what accuracy commitment, and who confirms the fix.
The data supports putting humans in the PTaaS loop. Automated scanning covers only a 10% to 30% share of what is there, and in Fluid Attacks’ 2025 results, 87% of critical-severity vulnerabilities came from manual review. A PTaaS service that cannot explain its human component is a scanner with a better interface.
Frequently asked questions
Is PTaaS really continuous penetration testing?
It depends on the provider, and the word hides four different things. Continuous platform availability, recurring manual assessments, scheduled automated scanning and continuous human testing are not the same. Ask what triggers a new manual assessment, whether new versions and newly discovered assets enter the scope, and how findings reach you between cycles.
What is the difference between PTaaS and a bug bounty program?
PTaaS is a contracted service with an agreed scope and a known team; a bounty program is an open invitation paid per accepted finding. A bounty gives opportunistic coverage with no guarantee that anyone examined what concerns you most. The service model gives methodical, repeated review of a scope you define, with retests included.
Does PTaaS include manual penetration testing or only automated scanning?
Manual testing is what makes it PTaaS. If no human tests the system, the service is vulnerability scanning delivered through a platform. Automated scanning finds only 10% to 30% of a system’s total risk exposure, according to Fluid Attacks’ documentation, because the rest needs an environment, authentication, test data or an understanding of what the application should do.
Does PTaaS satisfy PCI DSS penetration testing requirements?
It can provide the testing and the evidence, but it does not by itself make you compliant. PCI DSS requirement 11.4 makes penetration testing mandatory, and 11.4.1 asks the entity to define, document and implement the methodology and retain results for at least 12 months. A provider should show how its methodology, scope and reports map to those conditions.
Does Fluid Attacks guarantee the accuracy of its PTaaS findings?
Yes, contractually. The service-level agreement of the Advanced plan carries a 90% accuracy commitment, measured with an F2 score of at least 90% on risk-exposure reports and an F0.5 score of at least 90% on vulnerability reports, alongside 99.95% platform availability and a median first response under 8 office hours for 90% of reattack requests. All three indicators are measured each quarter.
Test your software at the pace you release it
Our PTaaS pentesters and scanners assess your systems as they change, and every fix you ship gets re-attacked until the vulnerability is actually gone. Start free trial · Contact us.
Get started with Fluid Attacks' PTaaS right now















