Frequently asked questions.

All plans include full access to AI Studio, Patent Illustrator, Rejection Simulator, Prosecution Intel, Patent Localizer, and Prior Art Search.

We offer Monthly ($125/seat/mo) and Yearly ($78/seat/mo) plans, both with full access to every tool. Enterprise plans with dedicated support and custom integrations are also available. All plans start with a free trial.

Yes. You can upgrade or downgrade at any time. Prorated charges apply.

Yes. Standard plans include a 7-day free trial; Enterprise customers may have access to extended trials.

Full access to all features: AI Studio, Prior Art Search, Patent Illustrator, Rejection Simulator, Prosecution Intel, and Patent Localizer.

We offer a 7-day money-back guarantee on all plans.

A session is one conversation thread, no matter how many prompts you send. Plan limits reflect a typical mix of patent work; heavier tasks and more capable models draw from your allowance faster than lightweight ones. Because Boundly is built specifically for patent work, the same job usually uses less of your allowance than it would on a general-purpose AI. Each model runs at its full context window, so long claim sets and reference documents fit without truncation.

Limits are pooled organization-wide, not per seat. For example, a 10-seat Premium team theoretically shares up to 10,000 AI Studio sessions per month, so heavy users aren’t capped by an individual quota and lighter users’ unused capacity stays in the team pool. Actual consumption may run lower depending on the model selected and the complexity of each prompt.

We send alerts as your team’s usage approaches the cap, so you’ll never be caught off guard. If you do exceed it, you can add a $50 or $100 top-up to cover the additional usage. Admin approval is required, so finance teams stay in control of spend. Anything you don’t use in the current month rolls forward to future months until it’s spent.

You keep your work. On cancellation, you can export all of your matters, prompts, and generated content. We retain a copy for 30 days in case you reactivate, then delete completely. Your data is never used for training, during or after your subscription.

Yes. Your work runs on enterprise AI model endpoints, not consumer AI. It is never used to train a model and is encrypted in transit (TLS 1.3) and at rest (AES-256). We never train on your data, not even anonymized. We never share it with third parties. Full stop. Boundly holds an active SOC 2 Type 2 attestation and is ISO 27001 certified.

No. Your prompts, documents, and generated content are never used to train any model, including third-party models we route to. This is enforced at the contract level with every model provider and documented in your DPA.

Storage and processing details depend on your deployment and setup. If you need region-specific or infrastructure-specific details, contact us and we can walk you through them directly.

Access to customer data is strictly controlled and limited to authorized personnel for support purposes only, with full audit logging.

Contact us at [email protected]. We process deletion requests within 30 days.

SOC 2 Type 2 is an independent audit of a service provider’s security controls (encryption, access management, change management, incident response) observed for months rather than at a single point in time. Boundly holds an active SOC 2 Type 2 attestation and is ISO 27001 certified. Both reports are available under NDA on request.

Your work runs on enterprise AI model endpoints, not consumer AI. It is never used to train a model and is encrypted in transit (TLS 1.3) and at rest (AES-256). No-training and data-handling terms are enforced at the contract level with every model provider, and your matters stay scoped to your organization.

We maintain a current list of subprocessors (cloud infrastructure and the model providers we route to) in our Trust Center, available on request. Every subprocessor is bound by no-training and data-handling terms at the contract level, and we notify customers of material changes before they take effect.

Boundly does not read your matters to review, improve, or train anything, and staff access to customer data is limited to authorized personnel for support purposes only, with full audit logging. The enterprise model providers we route to are bound by no-training and data-handling terms at the contract level, and your work is never used to train or improve any model.

Boundly holds an active SOC 2 Type 2 attestation and is ISO 27001 certified. Both reports, along with our DPA and subprocessor list, are available under NDA through our Trust Center. Contact us at [email protected] to request access.

Claude, GPT, Gemini, and Grok, all accessible through a single secure interface that runs on enterprise AI model endpoints, not consumer AI. As new frontier models launch, we add them so your team always has the best available.

AI-generated figures may not always be perfect on the first attempt. We suggest trying again. Results often improve. If the output still isn't right, we can connect you with a vetted patent illustrator who completes the figure for an additional fee.

We estimate that our simulator catches approximately 60% (and improving) of §112 issues based on training from real rejected applications. It is an assistive tool, not a guarantee of examiner outcomes. Support for §101, §102, and §103 is coming soon.

Yes. Patent Localizer translates and localizes filings for USPTO, EPO, JPO, CNIPA, and KIPO. AI Studio workflows run across the same jurisdictions.

Yes. Patent Localizer handles PCT national-phase entry into USPTO, EPO, JPO, KIPO, CNIPA, and more, restructuring your application to each office's drafting conventions and flagging dependent-claim rules and excess-claim charges before filing.

Yes. Patent Illustrator exports to PDF, PNG, or JPG, with additional formats available through a patent illustrator partner. AI Studio drafts copy into your editor or exports to DOCX. Prior Art Search returns a ranked search report. Patent Localizer exports localized drafts to PDF or DOCX, ready for your docketing system.

Claude, GPT, Gemini, and Grok, all from one interface. You can switch between them anytime and compare answers side by side. They run on enterprise model endpoints, not consumer AI.

No. Your work runs on enterprise model endpoints, not consumer AI. It is never used to train a model and is encrypted in transit and at rest.

Yes. You create projects for each matter, patent family, or client, and invite colleagues into the same matter so the context (drafts, threads, history) is shared. Everyone picks up where the others left off.

It is a place to standardize how your firm works. Save a proven prompt once and the whole team can reuse it, or keep your own personal workflows. It turns one practitioner’s best approach into a repeatable standard.

Yes. AI Studio is the hub. An AI agent works alongside you, and Patent Illustrator, Prior Art Search, the Rejection Simulator, and cross-jurisdiction localization all open as capabilities on one canvas.

Yes. Prosecution Intel is reachable from AI Studio and gives you the examiner’s history as a built-in data source, so a response can take into consideration how that examiner actually decides cases.

Yes. AI Studio can populate IDS forms straight from your matter, and more form automation is on the way.

We search 96M+ global patents and applications, publicly indexed web publications (academic papers, technical disclosures, product documentation), and the FDA Orange Book and Purple Book for pharma and biologics references.

Both. Search is hybrid. Semantic ranking catches synonyms and adjacent disclosures, and keyword matching anchors the exact terms that matter (model numbers, defined terms, chemical names).

Yes. If the strongest prior art is in Japanese, Chinese, Korean, or German, it surfaces alongside everything else, presented in English, with the original document one click away.

Yes. Searches run on enterprise model endpoints, not consumer AI. Your queries, draft claims, and uploaded references are never used to train a model and are encrypted in transit and at rest.

Yes. Prior Art Search is already there. Pull results straight into a draft, an office-action response, or a memo inside AI Studio without re-uploading or switching context.

No. Prior Art Search is assistive. It accelerates and broadens your review, and you and your team stay responsible for clearance decisions and final analysis.

Major offices including USPTO, JPO, KIPO, EPO, CGPDTM, CNIPA, CIPO, IP Australia, UKIPO, IPOS, and ILPO, plus PCT national phase, with additional jurisdictions on request. Coverage spans 15+ languages including English, German, French, Japanese, Chinese (Simplified and Traditional), Korean, Spanish, Portuguese, Italian, Dutch, Russian, Arabic, and Hebrew.

Translate converts the language only, which is what you want for incoming work like office actions, prior art, FTO research, and litigation. Localize translates and then restructures the draft to match the target office’s drafting conventions, which is what you want for outgoing filings like national phase, direct foreign filings, and EPO validation.

Yes. Claims are restructured into EPO two-part form per EPC Rule 43(1), with known features moved into the preamble and novel features placed behind "characterised by." Dependent claim chains are reworked to comply with JPO and CNIPA dependency rules, and excess claim counts are flagged before filing.

It is jurisdiction-aware. Beyond language, each office’s preferred claim format and specification structure are applied automatically, functional claim language is recast for offices that interpret it differently, and terminology is locked across a patent family so nothing drifts between the priority document, continuations, and foreign counterparts.

Yes. Your work runs on enterprise AI model endpoints, not consumer AI. It is never used to train a model and is encrypted in transit and at rest.

Yes. Localization is built into AI Studio, so you can translate or localize any draft you are already working on without re-uploading or switching context. Prior art search, figures, rejection simulation, and examiner intelligence are reachable from the same workspace, and your matters keep their context across all of them.

Section 112 rejections today, covering written description and enablement (112(a)), indefiniteness (112(b)), and means-plus-function (112(f)). Support for sections 101, 102, and 103 is coming soon.

Upload your full application, claims, specification, and figures together. The simulator analyzes them as one complete unit, the way an examiner reviews a file, rather than looking at the claims in isolation.

The simulator analyzes each claim against rejection patterns drawn from real USPTO office actions, then flags the specific claim language, points to the relevant spec citation, and suggests a tailored fix that references your actual wording.

Yes. Your application runs on enterprise AI model endpoints, not consumer AI. It is never used to train a model and is encrypted in transit and at rest.

Yes. The simulator lives inside AI Studio, so you can take a flagged issue, draft an amendment, and apply it without leaving the workspace. Prosecution Intel is reachable from the same place when you are drafting an office-action response.

A patent illustrator turns an invention into the formal figures a patent application needs, drawn to USPTO standards (37 CFR 1.84) with consistent reference numerals and lead lines. Boundly does this with AI in seconds, and a vetted human illustrator handles the figures that need one.

Yes. Boundly generates black-and-white line figures to USPTO formatting standards (margins, line quality, reference numerals), and you review and approve every figure before filing. For unusual or intricate drawings, a vetted human illustrator can take over.

A rough inventor sketch, a screenshot of a CAD file, or a photo. For each image you get two options to choose from, and you can regenerate if neither fits. You can also drop in a single PDF of many drawings and split it into individual figures.

Often only for the hard figures. AI clears the straightforward majority, and when a figure is genuinely intricate a vetted human illustrator from our marketplace is one click away, so you pay a person only for the figures that need one.

It is a flat monthly rate instead of paying per figure, where per-figure services often run $100 to $300. See our breakdown of what patent drawings cost for the full picture, and the pricing page for current plans.

Yes. The illustrator is one of the capabilities you can open inside AI Studio, alongside the editor, prior art search, and cross-jurisdiction localization. Your figures stay tied to the matter you are working on, so context carries across tools and colleagues collaborate on the same project. The standalone tool also remains available for now.

No. Boundly is an assistive tool. It accelerates drafting, illustration, prior art, and prosecution work, but every output requires attorney review. Final responsibility for filings and decisions rests with the user and their qualified counsel.

Qualified IP professionals can apply to join our Partner Program. Accepted partners earn recurring revenue on every referral, with monthly payouts via Stripe and a real-time tracking dashboard.

No. Boundly is an assistive AI tool only. All outputs are estimates based on training data. Final responsibility for all filings and decisions rests solely with the user and their qualified counsel.