The AI penetration testing market runs between two poles: fully-autonomous agents on one end and human-validated engagements on the other. Planck Operator holds the middle, a steerable, human-on-the-loop agent with exploit-proven findings and a self-serve on-ramp. Here is how it lines up against the field.
Most vendors sit at one end. On the autonomous end, agents run with minimal human input; on the human end, people validate everything. Each pole trades away something the other keeps. Planck Operator is built to keep both: autonomous breadth and human steering.
RunSybil (~$40M funding, third-party estimate) sits here: point-and-forget breadth with minimal steering. Strong on known vulnerability classes; weaker on context-specific business logic. XBOW (~$120M) started near this pole and, as of mid-2026, pulled back toward human-in-the-loop.
A named third category: steerable and human-on-the-loop. It runs autonomously for continuous breadth, and a human can direct it at business logic and edge cases. Findings are exploit-proven, pricing is published, and a free Recon tier is self-serve.
Terra (~$30M funding, third-party estimate) sits at the human-validated pole. Cobalt, Synack, HackerOne, and Bugcrowd run human PTaaS marketplaces with engagement medians roughly $30k to $105k (third-party estimates). Depth and a signed letter, on a fixed calendar.
Also in the field: Horizon3 / NodeZero (~$250M funding, third-party estimate) is a well-capitalized incumbent with sales-led, per-IP pricing; Astra ($1,999/yr) is the closest low-end self-serve analog. Planck's ownable position is the steerable, human-on-the-loop, exploit-proof, self-serve middle. Funding figures are third-party estimates as of 2026.
A page for each, written fairly: where the competitor is strong, where Planck differs, and how the two fit together.
Horizon3's NodeZero is a well-capitalized incumbent with sales-led, per-IP pricing. See how Planck compares on pricing model and steerable, exploit-proof findings.
AutomatedPentera is an established automated security validation platform. See how Planck's steerable agent and published, self-serve pricing line up.
AutonomousXBOW (~$120M) pulled back from pure self-serve autonomy toward human-in-the-loop in mid-2026. See where Planck's steerable middle sits.
PTaaSCobalt runs a human PTaaS marketplace of time-boxed engagements. See how Planck adds continuous, self-serve coverage between them.
Fully-autonomousRunSybil sits at the fully-autonomous pole. See how Planck keeps autonomy's breadth while letting a human steer the run.
A summary read across the field. Vendor characterizations reflect publicly described models as of 2026; where a figure is an estimate it is noted on the linked comparison pages.
| Planck Operator | NodeZero | Pentera | XBOW | Cobalt | RunSybil | |
|---|---|---|---|---|---|---|
| Self-serve on-ramp | Free Recon | Sales-led | Sales-led | Limited | Scoping call | Sales-led |
| Self-serve | Yes, free Recon tier | No | No | Limited | No | Limited |
| Steerable / human-on-the-loop | Yes, core model | Automated | Automated | Added mid-2026 | Human-delivered | Minimal steering |
| Exploit-proof reporting | Request/response, repro, CVSS v3.1 | Proof-oriented | Validation-oriented | Autonomous findings | Human report | Autonomous findings |
| Continuous | Yes | Yes | Yes | Yes | Time-boxed | Yes |
| Standards | WSTG, API Top 10, ASVS, PTES, NIST 800-115, ATT&CK, CVSS v3.1 | Recognized frameworks | Recognized frameworks | Recognized frameworks | Recognized frameworks | Recognized frameworks |
This matrix summarizes market positioning, not a benchmark. For the detail behind each column, see the individual comparison pages above.
Point Operator at a domain, steer it where it matters, and read exploit-proven findings, and start self-serve today with a free Recon tier.