Docket
Docket is a vision-first QA testing platform whose AI agents drive web, iOS, Android and desktop apps by on-screen coordinates rather than DOM selectors, so end-to-end tests survive interface changes without rewrites or scripting.
What is Docket?
Docket is an automated QA testing platform built around AI agents that use an application the way a person does. Where conventional tools bind each step to a DOM selector — an id, an XPath, a CSS class — Docket looks at the rendered screen and acts on the exact (X,Y) coordinates a human would click. That single design choice is the product's whole argument: a test tied to appearance rather than markup survives a front-end refactor, and it can reach surfaces the DOM never exposes, such as HTML5 Canvas, iframes, popups and the randomly generated class names that React and Vue produce on re-render.
The platform organises this around three behaviours the vendor calls deterministic, adaptive and intelligent. Recorded steps replay a captured journey precisely. Self-healing corrects click locations when elements drift, without a rewrite. AI steps, written in natural language, handle the parts of a flow that cannot be pinned down in advance. Tests are grouped into suites and folders, and can be generated from product requirement documents, screenshots or existing documentation, or imported in bulk.
Coverage spans real browsers, iOS simulators, Android emulators and desktop. Around the runner sit the parts that usually break end-to-end testing: dedicated mailboxes with plus addresses for email-driven journeys, handling for anti-bot systems such as Clerk and IP allowlisting, a tunnel that exposes a local environment, file upload into the agent's context, and support for two-factor authentication steps. Results reach teams as structured bug reports carrying screenshots, replayable video, network context and severity, and as email, Slack or signed webhook notifications. A documented REST API with a published OpenAPI specification drives tests, modules, apps and builds programmatically, and a GitHub app writes a commit status that can block a release.
Docket is published by Docket Technologies Inc., a two-person company from the Y Combinator Spring 2025 batch. It names eXp Realty, Centerpoint Connect and Nest Genomics as customers. No price is published anywhere.
What it does
- Write an end-to-end test in plain English, or record a real user journey, without any code
- Run the same test against web, native iOS, native Android and desktop targets from one platform
- Let tests self-heal when a button moves, a class name changes or a layout shifts
- Gate a deployment on QA results inside a CI/CD pipeline, with a manual override path
- Generate test cases from PRDs, screenshots and documentation, or bulk-import existing ones
- Collect structured bug reports with screenshots, replayable video and network context
- Exercise email-driven flows through a dedicated mailbox, and local builds through a tunnel
When to use Docket / When not to
A quick filter to help you decide if Docket is the right fit.
When to use Docket
- QA engineers and QA leads who spend more time repairing broken selectors than widening coverage
- Engineering teams shipping weekly or daily against fast-moving React or Vue front ends
- Product teams that want Jira acceptance criteria validated through the real interface before a ticket closes
- Lean startups with no dedicated QA specialist, who need coverage without adopting a scripting framework
- Teams testing HTML5 Canvas, iframes, popups or native mobile flows that DOM-based tools cannot reach
When not to use Docket
- Buyers who need a published price: no rate appears anywhere, and every button leads to a sales call
- Teams expecting self-service sign-up or an immediate free trial, neither of which the site documents
- Organisations whose processing falls under HIPAA, FISMA or GLBA, which the terms of service exclude outright
- Teams with a mature Selenium or Playwright suite that only need parallel cloud execution, a case the vendor itself sends elsewhere
- Procurement that requires a DPA, an EU Article 27 representative or a verifiable vendor certification, none of which is published
How to use Docket
A typical end-to-end flow, from setup to results.
- Book an introductory call through the scheduling link on the site, which is the only entry point offered
- Sign in to the application at app.docketqa.com once your account is open
- Follow the Quickstart to build a first test against your application
- Compose each step either as an AI step written in plain English or as a recorded step replayed exactly
- Give the agent application-specific guidance through Agent Rules so it interprets your interface correctly
- Group related tests into test suites and folders
- Generate further cases from PRDs, screenshots or documentation, or bulk-import cases you already have
- Run the suite on browsers, iOS simulators and Android emulators, on demand or on a schedule
- Create a Docket API key, store it as the DOCKET_API_KEY repository secret, install the GitHub app and wire the release workflow
- Set failure notifications by email, Slack or signed webhook, and adjust the action delays if timing needs tuning
Pros & Cons
Pros
- Removes selector maintenance altogether: tests survive front-end refactors instead of breaking
- Reaches surfaces DOM-based tools cannot test, including Canvas, iframes and popups
- No scripting required, so non-developers can build and maintain coverage
- Web, iOS, Android and desktop handled by a single platform and a single framework
- Genuine CI/CD integration that can block a release and route failures to a manual approval
- Documented REST API with a published OpenAPI specification, plus a readable 22-page manual
- Guided onboarding, with setup claimed at two to three hours and initial suites built by the vendor
Cons
- No pricing whatsoever is published, so the tool cannot be budgeted without a sales conversation
- No self-service sign-up: every call to action opens a scheduling page
- Call-to-action labels contradict each other, one reading 'Get Started for Free' for a demo booking
- Privacy policy, terms and cookie policy are generic and never describe QA testing at all
- Legal and privacy contact sits on signdocket.com, a domain other than the product's
- No DPA, no EU Article 27 representative, no postal address and no vendor certification that could be verified
- A two-person company with a domain registered in May 2025 offers very little track record
Pricing & Plans
Docket publishes no pricing. There is no pricing page, no rate card and no figure anywhere on the site or inside the application, and neither a permanent free plan nor a free trial is documented. The terms of service nevertheless describe a paid subscription, with billing and renewal, cancellation and fee changes, payable in US dollars through a third-party processor. A price can therefore only be obtained by requesting one from the vendor.
Data, GDPR & hosting
A consolidated view of how Docket handles your data.
GDPR overview
Implementation stops at the level of legal bases. The privacy policy, last updated 20 October 2025, states that personal information of EU and UK users is processed only on a valid GDPR or UK GDPR basis, and lists consent, performance of a contract, legal obligations, legitimate interests and vital interests. Rights of access, correction, deletion, copy and objection are announced and exercised by email at founders@signdocket.com. Beyond that, nothing: Docket never claims GDPR compliance in so many words, designates no Article 27 representative in the European Union, names no data protection officer, publishes no data processing agreement and cites no transfer mechanism such as standard contractual clauses. Transfers out of the EU are acknowledged only by a warning that destination jurisdictions may protect data less well.
Who owns the data?
Docket's security page states that customers retain ownership and control of their data, with self-service retrieval and deletion. The terms of service confirm you keep ownership of your contributions, but grant Docket a perpetual, irrevocable, worldwide, royalty-free licence over any feedback or suggestions you send. Docket also keeps data you transmit in order to manage service performance, and states you remain solely responsible for it despite routine backups. Personal information may be shared with corporate affiliates, service providers, professional advisers, authorities and business transferees. Requests to review, update or delete data go to founders@signdocket.com, an address on a domain other than the product's.
Reuse rights
Nothing in the terms restricts what you do with your own material. You keep ownership of the tests, recordings and bug reports you produce, and no clause requires Docket's permission to reuse them commercially. The restriction runs the other way: none of Docket's own content, marks or interface may be copied, republished, resold or otherwise exploited for commercial purposes without prior written permission, and any feedback or suggestion you volunteer becomes freely exploitable by the vendor. On Docket's side, the privacy policy permits usage analysis to improve the service and the creation of aggregated or de-identified data. AI features run through Anthropic, OpenAI and Google Cloud AI, and your inputs and outputs may be processed by those providers; the policy offers to opt out of AI processing where feasible on request by email, but never states whether customer data is used to train models.
Data retention & training
Hosting summary
Docket does not say where customer data is hosted. No country, no region and no hosting provider is named. The security page states only that the service runs in an SSAE-18 and SOC 2 certified data centre, which describes the provider's certification rather than a location, and the privacy policy adds only that the company is headquartered in the United States and may use service providers there and in other countries, with a warning that destination jurisdictions may protect personal data less well. No transfer mechanism such as standard contractual clauses is cited, and no subprocessor list is published. What is documented concerns protection rather than place: 256-bit TLS in transit, 256-bit AES at rest, role-based access control, least privilege, session timeouts, usage logging and anomaly monitoring, data minimisation, monthly vulnerability scans and application security testing on every release cycle. Buyers with a data residency requirement will have to obtain the answer directly from the vendor.
Things to keep in mind
Risks and trade-offs to weigh before adopting Docket.
- No published price and no self-service path: budgeting and comparison are impossible without engaging a salesperson
- One call-to-action reads 'Get Started for Free' but opens a demo booking, and the 'Watch Demo' button links nowhere
- The legal documents never describe QA testing, although the tool handles customer interfaces, test credentials and application data
- Legal and GDPR contact sits on signdocket.com, a leftover from a previous brand, which makes it easy to doubt the right recipient
- The consent banner claims no tracking or analytics while the cookie policy documents Segment and PostHog, and the site loads Google Tag Manager
- Certification is easy to misread: only the data centre is described as SSAE-18 and SOC 2 certified, and the linked trust portal never answered
- Vendor-reported gains of 60, 80 and 85 percent come from Docket's own case studies with no published method, as do the competitor comparisons
Setup & Integrations
Technical difficulty
Two different levels. Creating tests is deliberately easy: plain-language objectives or recorded journeys, no scripting framework, and the vendor claims most teams finish setup in two to three hours with hands-on help building the first suites. Wiring the pipeline is another matter and needs an engineer: an API key stored as a GitHub repository secret, a GitHub app installation and a release workflow with commit-status permissions. The advanced timing settings are flagged for power users only, and the local tunnel expects a command-line tool.
Deployment
Integrations
Supported languages
Behind Docket
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Docket.
Frequently asked questions
How does Docket differ from a selector-based testing tool?
Which platforms can Docket test?
Do I need to write code to create a test?
Can it test HTML5 Canvas or a single-page application?
Does Docket plug into a CI/CD pipeline?
Is there an API?
How much does Docket cost?
How long does it take to get running?
What is the minimum age to use the service?
Is Docket SOC 2 certified?
Should you pick Docket?
Docket makes a sharp, legible bet: stop binding tests to markup and let an agent look at the screen instead. For teams whose real cost is not writing tests but repairing them after every release, that bet addresses the right problem, and it opens surfaces — Canvas, iframes, popups, native mobile flows — that selector-based suites reach badly or not at all. The surrounding platform is more complete than a 2025 product usually manages: multi-platform runs, a GitHub app that can hold back a release, a documented REST API with an OpenAPI specification, dedicated mailboxes, anti-bot handling and a tunnel to local environments.
The reservations are commercial and contractual rather than technical. Nothing about price is public, and every path forward is a sales call, which suits an enterprise buyer and blocks everyone else. The legal apparatus lags noticeably behind the product: the privacy policy, terms and cookie policy are generic templates dated October 2025 that never mention QA testing, screenshots or customer application data, and they route legal and privacy contact to an address on a former brand's domain. There is no data processing agreement, no EU representative, no postal address, and the only certification claim on the site belongs to the hosting provider, not to Docket. The consent banner even asserts there is no analytics while the cookie policy documents two analytics vendors.
Read it as a genuinely interesting engineering proposition from a two-person, Y Combinator-backed company with less than two years behind it. Worth a pilot on a difficult interface; worth a careful procurement review before anything wider.
- Choosing a selection results in a full page refresh.
- Opens in a new window.