CaseBlink is an AI case preparation platform for US immigration attorneys and law firms, run by CaseBlink, Inc. of New York and founded in 2023. It organizes and labels uploaded client documents, extracts data into USCIS forms, and drafts support letters, briefs and RFE responses in the firm's style with citations to the uploaded evidence. It researches AAO decisions, the USCIS Policy Manual and case law, and assembles tabbed packets ready to file.
Its Autopilot agent, launched at the AILA 2026 conference, runs the full sequence in one click for attorney review. It covers employment, family, humanitarian and naturalization case types, and sells per seat from $100 a month, with AI drafting from the $200 Core plan. CaseBlink also runs a page for individual applicants that matches them with partner immigration firms using the platform. It uses OpenAI under an agreement for zero data retention, and Google Cloud, and reports SOC 2 Type 1 (2024) and Type 2 (2025) audits.
CaseBlink raised $2 million in March 2025 in a round led by Tower Research Ventures and says more than 100 firms use it.
Side by side on the same 15 capability axes and 12 legal signals: where the two part company, and what neither one publishes.
Vendors in the same categories as CaseBlink whose graded profiles look most alike, with a line on what each product does.
Similar is not the same, so check that each one does the job you need. No vendor pays for placement.
Ordered by how closely the graded profiles match across the 15 capability axes, inside the same categories. See how each axis is graded in the methodology.
All 15 axes, graded from public sources on the date shown. Hover a grade to see what the letter means on that axis.
How much of the product is actually AI. Whether the machine learning is the mechanism the buyer is paying for or a feature layered onto conventional software, and whether the vendor is specific about which is which.
CaseBlink is AI case preparation for US immigration attorneys. Its models tag, label and extract data from client documents and check them for missing context, duplicates and inconsistencies. They search AAO and BIA decisions, case law and the USCIS Policy Manual, fill USCIS forms from intake and uploaded documents, and draft support letters, reference letters and briefs in the firm's voice. They read an RFE notice to map evidence to each request and draft the response.
The Autopilot agent runs that sequence from upload to a tabbed filing packet in one pass. The Core plan at $200 per seat is where the AI lives. The Lite plan at $100 covers categorization, exhibit lists and one click PDF export, which are largely workflow, and the Pro plan in beta adds case management around the AI. For a firm buying Core or above, removing the models leaves very little.
Whether the vendor publishes measured accuracy on citations and assertions, grounds output to primary sources, and says plainly what its system does when it does not know. Legal has a documented public record of fabricated citations reaching filed briefs, so an untested claim of accuracy is not evidence.
CaseBlink says every output cites directly to the client's uploaded case documents and its research, so the attorney can check each statement at its source. It names what the research draws on: AAO and BIA decisions, case law, the USCIS Policy Manual and government resources. Its quality checker reviews uploaded evidence for missing context and inconsistencies before drafting. No number is published: no citation accuracy rate, no rate of unsupported statements, no test set and no approval outcome.
The Service Agreement lists ways AI output can go wrong, including errors, misleading content and weak reasoning on complex tasks, and disclaims any warranty of the accuracy of output.
What the system decides on its own, what a lawyer must approve, and whether the vendor documents where the review point sits. A tool that drafts under review and a tool that files without one are different products and different risks.
Autopilot organizes documents, reviews evidence, conducts research, writes drafts and assembles the filing packet in one click, then hands the attorney a complete packet to review before anything is filed; CaseBlink does not file on its own. Each statement in a draft cites to the uploaded evidence or research behind it, which gives the reviewing attorney a way to check the work line by line. The page for individual applicants says an expert attorney prepares the case using CaseBlink under their supervision.
No control inside the automated run is published: no confidence signal on a weak draft, no point where Autopilot stops for a decision, and no route for flagging an error back to the vendor. The Service Agreement says CaseBlink may review output before it is returned but does not promise to, and places the risk of relying on output with the customer.
Named, dated evidence that the product works in production at real firms or legal departments. Case studies with figures and identified customers count. Unattributed testimonials and launch announcements do not.
Logos of firms including Dorsey & Whitney, Dickinson Wright, Fakhoury Global Immigration, Clark Lau and Cline Williams appear on CaseBlink's site, which runs a testimonials page with named attorneys. Its page for individual applicants refers to partner firms that review cases prepared on the platform. It reports more than 100 firms, 1,000 users and 3,000 cases prepared, up from more than 50 firms and 1,000 cases in March 2025, and announced a $2 million funding round on its blog.
No outcome evidence is published: the claim of filing five times faster has no method or sample behind it, and no approval rates, RFE rates or time per petition are reported for any named firm.
How client confidences are handled: attorney client privilege and work product treatment, segregation of one client matter from another, whether client data trains any model, and what the vendor commits to in writing rather than in marketing.
The Service Agreement (effective 15 July 2026) says third party AI tools acting for CaseBlink retain no customer data and that output generated from customer data is itself customer data. Section 5 carries mutual confidentiality obligations, including notice before any compelled disclosure, and the data processing agreement has customer personal data deleted within 90 days of termination. The security page adds isolated tenant environments, data purged immediately after each model operation, no model training on client content, and a zero data retention agreement with OpenAI available on request.
Section 1.4 licenses CaseBlink to copy, create derivative works of and otherwise use customer data to improve and develop its products and to create anonymized aggregated data, on a platform holding asylum declarations and medical evidence. Neither the agreement nor the security page addresses attorney client privilege or work product.
Whether the vendor is clear that it supplies a tool rather than legal advice, who its audience is, and how it addresses unauthorized practice of law, competence and supervision duties, and jurisdiction limits. ABA Formal Opinion 512 is the reference point. Where the advice line is not the duty a product raises, the axis is read through the nearest professional duty it does raise: judicial conduct rules and the reviewing duty for products sold only to courts, and the duty to bill for time actually spent for products that draft time entries.
Section 6.2 of the Service Agreement says that material provided by CaseBlink or its representatives is for general information, is not legal advice and creates no attorney client relationship, and the website terms repeat it; the agreement describes the service as a tool for legal professionals and puts responsibility for reviewing output on the customer. Section 6.3 lists the limits of AI output in plain terms. The product is sold to immigration attorneys, paralegals and corporate immigration teams, and every draft is framed as work the attorney reviews before filing.
The exception is the page for individual applicants, which takes a foreign national's intake form and documents, structures them into a case file for partner firms to assess, and matches the applicant with a vetted partner firm. It says the attorney prepares the case under their supervision, but carries no not legal advice statement of its own and does not say whether partner firms pay for the introduction.
Published governance over model behavior: who owns it inside the vendor, what is tested before release, and what is disclosed about disparate output across matter types, parties, or populations.
Section 6.3 of the Service Agreement lists the ways AI output can fail, including factual errors, a lack of empathy, bias carried over from training data and weak reasoning on tasks that need judgment. It makes the customer responsible for any bias audits, notices or consents that apply where AI processes personal data. What sits behind that on CaseBlink's side is not described. No one is named as accountable for the drafting and research models, and nothing says how a change to prompts, templates or the underlying model is tested before firms see it.
No evaluation is published of how drafts and RFE responses perform across visa categories, nationalities or the languages of source documents. The security page's 52 controls cover information security, not model behavior, on a product whose drafts argue that a person qualifies for a visa.
Retention, deletion, access control, and what happens to prompts and documents after they are processed. Whether the vendor states its subprocessors and its incident practice, or leaves the buyer to assume.
The Service Agreement (effective 15 July 2026) binds the third party AI tools acting for CaseBlink to retain no customer data, and the security page says data is purged immediately after each model operation. The data processing agreement has customer personal data deleted within 90 days of termination, commits CaseBlink to remediate a personal data breach promptly and notify the firm without undue delay with assistance on its own notices, lists its subprocessors, Google LLC, OpenAI OpCo LLC and Slack Technologies LLC, with 15 days' notice before any new one gains access and a right to object, and grants audit rights that a current SOC 2 Type 2 report can satisfy.
The security page adds AES 256 encryption at rest and in transit, isolated tenants, single sign on, role based access, audit logs for sensitive operations, continuous threat monitoring and disaster recovery backups. The breach clause sets no fixed hour window.
What the vendor stands behind contractually when its output is wrong. Indemnities, caps, carve outs, insurance, and whether any of it is published or only reachable through a negotiated agreement.
Section 7 of the Service Agreement caps CaseBlink's total liability at the fees paid in the twelve months before the claim, cumulative rather than per incident. It raises the cap to twice that amount for breaches of the data security terms or confidentiality involving customer data. Section 8 has CaseBlink indemnify the firm against third party claims that the service infringes intellectual property. Output is carved out of all of this.
The agreement disclaims any warranty of the accuracy or authenticity of output and any liability for actions taken based on it, and has the customer indemnify CaseBlink for claims arising from its use of output. A petition denied because a drafted brief misstated evidence or cited a decision wrongly would fall on the firm.
How deeply the product reaches into the systems legal work already lives in: document management such as iManage and NetDocuments, Word and Outlook, contract lifecycle management, matter management, e-billing, and court filing systems.
Drafts leave CaseBlink as Word files and filing packets as tabbed PDF. It supports single sign on through SAML 2.0 and OIDC with Okta, Azure AD and Google Workspace, including central provisioning and deprovisioning, and the Service Agreement refers to API access. Its answer to case management is to build its own. The Pro plan, in beta at $300 per seat, adds client intake, profiles, a data repository, calendars and dependents tracking inside CaseBlink, and Enterprise offers custom integrations and a white label client portal.
No connection is named to the immigration case management systems firms already use, such as INSZoom, Docketwise, eImmigration, LawLogix or Imagility, or to document management systems, and the API is not documented. A firm keeping its existing case system would move documents in and drafts out by hand.
Where the software runs and where the data sits. Multi tenant cloud, single tenant, private deployment, on premises, and whether region of residence is a published option or an enterprise conversation.
Google Cloud Platform hosts the service in isolated tenant environments, with encryption at rest and in transit, continuous monitoring and replicated disaster recovery backups, as CaseBlink's security page describes. The data processing agreement's subprocessor schedule places every listed processor, Google, OpenAI and Slack, in the United States, so a firm on a standard plan knows its data is processed in the US. The Enterprise plan, for 15 or more users, adds private hosting, a white label portal and data residency options.
The isolated tenant design addresses separation between firms. Which regions the Enterprise residency option covers, and whether private hosting means a dedicated cloud project or on premises, is not published.
Independent attestation a buyer can pull without a sales call: SOC 2, ISO 27001, penetration test summaries, a trust center with current reports and named scope rather than a badge image.
A SOC 2 Type 1 audit in 2024 and a SOC 2 Type 2 audit in 2025 are complete, with annual audits, regular third party penetration tests, a vulnerability disclosure program and staff security training, according to CaseBlink's security page. It lists its controls by family, access control, business continuity, change management, incident response and information security, 52 in all. The audit reports and its zero data retention agreement with OpenAI can be requested through its trust center at trust.caseblink.com.
The data processing agreement lets a current SOC 2 Type 2 report stand in for a customer audit, which ties the attestation to the contract. The security page describes the platform as SOC 2 Type 2 compliant and certified, though SOC 2 produces an auditor's report rather than a certificate. No auditor or report period is named publicly.
Which models sit underneath, whose they are, where they run, and whether the vendor commits to telling customers when that changes. A legal buyer inherits every dependency it cannot see.
The data processing agreement's subprocessor schedule names OpenAI OpCo, LLC as the third party AI provider, processing in the United States, alongside Google LLC for hosting and Slack Technologies LLC. CaseBlink must give at least 15 days' notice before any new subprocessor can access customer personal data, with seven days for the firm to object on reasonable grounds. The security page describes a zero data retention agreement with OpenAI, available on request, under which data is not retained after each operation and is not used for training.
A firm can tell its clients from a document it signs which companies see their passports, employment records and asylum declarations. The specific OpenAI models and versions behind drafting, research and extraction are not named, and a change of model within OpenAI would not trigger the notice.
Whether a buyer can learn what this costs without entering a sales process: published rates, the unit being charged, what sits behind an enterprise tier, and what implementation adds.
Prices, units and plan contents are published. Lite costs $100 per seat per month ($92 billed annually) for document categorization, labeling, quality checks, exhibit lists and tabbed PDF export; Core costs $200 per seat per month ($183 annually) and adds AI research, drafting, form filling, RFE responses and the Autopilot agent; Pro costs $300 per seat per month ($275 annually) and adds case management, now in beta.
Every plan includes unlimited cases and a 14 day free trial, annual billing includes one month free, and Enterprise, for 15 or more users with private hosting, a white label portal and dedicated support, is priced on request. The Service Agreement sets out the rest of the commercial terms: one seat per named person, reasonable use limits, trial cancellation, automatic renewal and notice of fee changes. Cases are unlimited per seat, in a category often priced per case.
Who the product is actually built for. AmLaw, midlaw, small firm and solo, in house departments, government and courts, and which practice areas are supported rather than merely claimed.
CaseBlink is built only for US immigration law. It covers employment visas (H-1B, L-1, O-1, TN, E-2 and E-3), employment green cards (EB-1A, EB-1B, EB-1C, EB-2 NIW, PERM and EB-3), family based petitions, humanitarian cases including asylum, DACA, TPS and U and T visas, naturalization, and RFE responses across them. It is sold to solo practitioners, high volume immigration practices, paralegals and corporate immigration departments.
Plans are per seat, from a single user up to Enterprise for 15 or more, and named customers run from boutique immigration firms to Dorsey & Whitney. A separate service matches foreign nationals with partner firms that use the platform. Nothing published says where drafting or research is less reliable, for instance humanitarian narratives against employment petitions, or which forms Autopilot cannot yet assemble end to end. Removal defense is not mentioned at all.
The public pages on file for CaseBlink, with the recorded signals each one supports and the date it was last read. Open any of them and check the reading against the record.
Prompt and Output Retention, Third Party Request and Subpoena Notice, Outside Counsel Guideline Readiness
Read Sep 18, 2026
Client Data in Training, Ethical Walls and Matter Segregation
Read Sep 18, 2026
Primary Law Corpus Provenance
Read Sep 18, 2026
Billing and Fee Posture
Read Sep 18, 2026
Fabricated Citation Record
Read Sep 18, 2026
Per seat per month, unlimited cases subject to one professional's ordinary caseload: Lite $100 (packet assembly, no AI drafting), Core $200 (AI drafting, case preparation, Autopilot agent), Pro $300 (case management, beta); annual billing gives one month free; Enterprise custom, per seat or case cap. Source: https://www.caseblink.com/pricing and the Service Agreement (15 July 2026).
Implementation: None published. All plans include a 14 day free trial; a payment method is taken at sign up.
Firm skills in CaseBlink can now start on their own when a case date comes close, such as an RFE or NOID response coming due, a document about to expire, or a priority date coming current on a chosen Visa Bulletin chart. Each automation is written as a sentence, and before it is saved a preview shows which cases it would pick up or skip today, and why.
A signal records what public sources say on the date shown. It is not a grade and it is not a recommendation. Where a signal reads Not addressed, it means the index did not locate the material in public sources on that date, which is a statement about disclosure rather than about the product.
Can material a lawyer puts into this product be used to train a model?
A public policy or trust page states no training on customer content, with no matching term located in the published agreement.
The security page says client data is never stored or trained on and that there is no model training on client content. No matching prohibition is in the Service Agreement (15 July 2026), whose section 1.4 licenses CaseBlink to use customer data to improve and develop its products and services; the agreement does bind third party AI providers to retain no customer data.
How long does the product keep what a lawyer typed, and can that be set to zero?
A specific retention period is published and the customer cannot change it.
Fixed terms are published. The Service Agreement says third party AI providers acting for CaseBlink retain no customer data, the security page says data is purged after each AI operation, and the DPA requires deletion of customer personal data within 90 days of termination. Case files, including AI output, are kept in the account for the life of the subscription.
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Segregation is asserted in public materials with no published detail on how it is enforced.
Segregation is claimed at tenant level without detail at matter level. The security page lists isolated tenant environments and fine grained access controls by role, and the Pro plan adds team management; nothing describes restricting access to particular cases or clients within a firm.
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Terms commit to notice where lawfully permitted. No transparency report located.
The Service Agreement commits CaseBlink to give written notice before any disclosure required by subpoena or order, to let the customer intervene, and to cooperate in seeking protection; the DPA commits to notify the customer of government access requests. No transparency report is published.
Where does the law in this product come from, and does the vendor have the right to use it?
Sources are identified without stating the license or rights basis.
The research sources are named without a license or provenance statement. CaseBlink says its immigration research draws on AAO decisions, the USCIS Policy Manual and case law; where the case law comes from and on what terms is not stated.
Does the product tell you when the authority it just cited has been overruled?
No located public material addresses whether authority is checked for subsequent history.
No check that cited AAO decisions or case law remain good law is described on the product, security, pricing and Autopilot pages, in llms.txt or in the Service Agreement.
What does the product do when the answer is not in the corpus?
No located public material addresses what the product does when it cannot ground an answer.
CaseBlink flags missing context and quality gaps in uploaded documents, but no behavior for signaling uncertainty in drafts or declining to draft when evidence is missing is described.
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
No court order, opinion or disciplinary record addressing fabricated or hallucinated legal citations produced by this product has been located as of the date shown. This is a statement about the public record on that one subject, not a finding about the product, and this signal is not a litigation history.
The AI Hallucination Cases database maintained by Damien Charlotin, and trade press reporting, record no court or agency finding addressing fabricated content in output from CaseBlink. This is a statement about the public record rather than a finding about the product, and it covers fabricated content only, not litigation history of any other kind.
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
No located public material engages with bar or ethics guidance.
No ethics opinion, court AI order or rule, or bar guidance is referred to on the product, security and applicant pages, in llms.txt or the Service Agreement, or on the vendor's blog.
Does the vendor address what happens to the bill when the work takes an hour instead of six?
Public materials claim time savings without addressing billing or disclosure, and the product sits inside a fee relationship between a lawyer and a client where those savings would change the bill.
Speed and efficiency are marketed without billing or disclosure guidance. CaseBlink says cases are filed weeks ahead and five times faster and that firms win more cases with less effort; nothing addresses how firms should reflect AI assisted preparation in fees or tell clients about it.
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
A subprocessor and model provider list plus client facing disclosure material is published or available without an agreement in place.
The DPA in the public Service Agreement lists every subprocessor with its role and country, naming OpenAI as the AI model provider and Google as cloud host, with change notice and objection rights, and the security page offers the SOC 2 reports and the zero data retention agreement with OpenAI through the Trust Center, all without an agreement in place.
If a judge’s standing order requires an AI disclosure, can the product produce one?
No located public material addresses court disclosure or verification certification.
Output cites to uploaded evidence and the platform keeps audit logs of sensitive operations, but nothing records which parts of a filing were AI generated or supports disclosure of AI use to USCIS or a court.
Every grade and every signal on this index is drawn from public sources and dated. If a record is wrong, out of date, or missing an artifact the index did not locate, send the source and it will be reviewed and the record redated. Vendors are welcome to submit documentation. Nothing on this index is for sale, including a listing, a placement, or a grade.