C
CaseGlide
CaseGlide is a litigation intelligence platform for corporate legal departments and property and casualty claims organisations, built on matter management and e-billing and layered with three named AI products. Case Clerk AI reads the status reports defence counsel already send, extracts the material facts, updates the case file and syncs it to the claims system; Chronicle AI assembles a case chronology from depositions, independent medical examinations, demand packages and expert reports; Chambers AI answers questions from the customer's own closed and open case history.
Each carries a published guardrail: no fact enters the file without a traceable source, every chronology entry links to its source document by page and line, and Chambers draws on no one else's portfolio. The vendor states that it does not predict verdicts or score outcome risk. The platform scores exposure, posture and deadlines across a docket, flags settlement candidates, and measures defence counsel on reporting quality, cycle time and results.
Council, a separate ten-week analytics engagement run from a data extract, sits alongside the platform. CaseGlide, LLC is based in Tampa, Florida.
Capability grades
All 15 axes, graded from public sources on the date shown. Hover a grade to see what the letter means on that axis.
AI Centrality
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.
The AI does the reading and a matter management platform does the rest. Three named products carry the intelligence: Case Clerk AI turns defence counsel status reports into structured case facts, Chronicle AI assembles chronologies from the litigation record, and Chambers AI answers questions from the customer's own case history. Underneath them the published agreement describes matter management services for exchanging, monitoring, auditing and reporting on litigation with law firms, and e-billing services for invoice data, which are substantial products in their own right and would still run without a model. Verified 20 September 2026.
Citation Accuracy and Hallucination Disclosure
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.
Grounding is documented product by product and accuracy is never measured. Every fact Case Clerk AI extracts stays traceable to the report it came from, with the vendor's stated rule that nothing enters the file without a source; every Chronicle AI entry links back to its source document by page and line; Chambers AI answers cite the case files they draw on. The vendor argues the point publicly, telling buyers to test traceability by clicking any output and asking to see the document, and saying that explanations and benchmark scores are not sources.
No accuracy rate, error rate or evaluation is published, which the vendor treats as the right trade: it says reliable legal AI shows sources, not scores. Verified 20 September 2026.
Autonomy and Oversight Model
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.
Each product is bounded by a published rule about what it may not do, and the boundaries are categorical rather than aspirational. Case Clerk AI may not put a fact in the file without a traceable source; Chronicle AI entries must link to a source document, page and line; Chambers AI answers only from the customer's own case history and never benchmarks the customer against another portfolio; and the platform as a whole does not predict verdicts or score outcome risk, which the vendor states repeatedly and explains as a design choice, on the basis that unverifiable inference is where fabricated output hides.
The oversight model follows from that: each product does one narrow checkable job, the source trail is the review surface, and the vendor's own formulation is that the AI reads and the customer's team decides. Verified 20 September 2026.
Operational and Outcome Evidence
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.
Customers are shown as logos and results are shown as targets. Ten carrier and enterprise marks appear on the homepage, including American Integrity, Pure Insurance, Tower Hill, Velocity Risk and Schindler, under a claim of P&C carriers and Fortune 500 legal departments, with no deployment described and no customer quoted. The numbers presented are explicitly aspirational, labelled as targeted reductions of 10 per cent in defence spend, 5 per cent in settlements and 15 per cent in litigation volume, and the industry figure that frames them, 135 nuclear verdicts totalling 31.3 billion dollars in 2024, is market data rather than product evidence.
Every product screen on the site is labelled as a fictional matter or fictional firm, which is candid and is also a demo rather than evidence. Verified 20 September 2026.
Privilege and Confidentiality Posture
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 confidentiality machinery is contractual and it is strong, with one gap that the vendor's own buying guidance names. The published terms, dated 12 August 2020, bind both parties to hold confidential information, including third-party confidential information such as material from law firms, to use it only for the purposes of the agreement, to restrict access to personnel with a need to know, and to return or destroy it on termination; compelled disclosure requires written notice to the other party first and a reasonable effort to obtain a protective order.
The security article adds safeguards for non-public personal information, encryption, access clearances and background checks. The live definition of Confidential Information still carries a drafter's parenthetical asking whether third parties, especially law firms, are covered. What is missing is the clause the vendor tells buyers to demand of any AI litigation vendor: an explicit bar on customer data training models and a zero-retention commitment.
The published agreement contains neither, and reserves the right to use customer data in aggregated and anonymised form. Verified 20 September 2026.
UPL and Professional Responsibility Posture
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.
The vendor takes a clear position on what its output is and leaves the legal judgement where it belongs. The published line is that CaseGlide makes the record legible and checkable and leaves the judgement with the customer's team, reinforced by the refusal to predict verdicts or score outcome risk, so the product informs a settlement decision without purporting to make it. The estate also engages the professional duties around it at length, publishing guidance on the verification duty, on keeping client data out of consumer-tier tools, and on how AI use bears on privilege.
What is not published is a statement of who inside a claims or legal department may rely on an extracted fact without checking it, or any treatment of the supervision question for the non-lawyer claims staff who use the same file. Verified 20 September 2026.
AI Governance and Bias Disclosure
Published governance over model behaviour: who owns it inside the vendor, what is tested before release, and what is disclosed about disparate output across matter types, parties, or populations.
The guardrails are published and the governance behind them is not. Each product carries a stated constraint, the platform refuses outcome prediction as a design choice, and the vendor publishes a framework for what buyers should demand of any AI vendor's data handling, which is a real public position on how AI should be built for this work. Nobody is named as accountable for model behaviour inside CaseGlide, nothing describes what is tested before a change ships, no evaluation results are published, and nothing addresses whether extraction quality varies by report style, firm or matter type.
Checked the homepage, the platform page, two answer-library pages, the terms of service and the security section on 20 September 2026. Verified 20 September 2026.
AI Safety and Data Stewardship
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 agreement, dated 12 August 2020, carries most of this, which is where a buyer can enforce it. CaseGlide commits to administrative, physical and technical safeguards and an information security programme for non-public personal information under the Gramm-Leach-Bliley Act, security clearances assigning specific access privileges, encryption of data collected, transmitted and stored, business continuity and backup, and background checks on personnel with access to customer data.
On a security breach it commits to notify the customer promptly, estimate the effect, specify corrective action, investigate and report. Alongside that, SOC 2 audit coverage is claimed with documentation available on request. The gaps are retention and the supply chain: no retention window is stated for case data during the term, no subprocessor list is published, and no model provider is named. Verified 20 September 2026.
AI Liability and Recourse
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.
A real published allocation of risk, with the accuracy of the output carved out of it. The terms, dated 12 August 2020, give a services warranty of professional quality with re-performance as the exclusive remedy, an intellectual property indemnity running from CaseGlide to the customer with defined remedies, a mutual liability cap at the fees paid in the preceding twelve months, a one-year limitation period, and an insurance schedule the customer can rely on, including five million dollars of errors and omissions cover, five million of umbrella liability and commercial crime cover.
Against that, the general disclaimer expressly excludes any warranty of data accuracy, so a wrong extraction or a missed deadline in a synced file is not covered by anything published. Verified 20 September 2026.
Practice Systems Integration Depth
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.
The central integration is asserted and never specified. Syncing the case file back to the customer's claims system is presented as a defining feature on every page, and the estate says elsewhere that CaseGlide connects to accounts payable and ERP systems through APIs, but no claims system, policy administration system or ERP is named anywhere, and no API documentation, connector list or implementation detail was located.
The Council engagement is offered specifically as a way to get value without credentials or API connections, which tells a buyer something about how much integration work the platform normally needs. Checked the homepage, the platform page, the answer library index, two answer pages and the terms on 20 September 2026. Verified 20 September 2026.
Deployment Model and Data Residency
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.
Nothing published states where the platform runs or where litigation data is held. No hosting provider, region, tenancy model or residency option appears on any page or in the agreement; the only related commitments are business continuity, disaster recovery and backup capabilities in accordance with industry standard practices. For a platform holding privileged defence counsel work product for carriers and Fortune 500 legal departments, a security reviewer has nothing published to start from.
Checked the homepage, the platform page, the answer library, the terms of service and the security section on 20 September 2026. Verified 20 September 2026.
Security Certifications and Trust Center
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.
The certification is stated with a route to the evidence, which is more than most. The site says CaseGlide is SOC 2 audited, that security documentation is available to the buyer's team on request, and that the vendor walks through it as part of every evaluation, under a heading that treats litigation data as privileged. The agreement backs it with substantive security obligations. What the top band asks for is still absent: no auditor is named, no report period or scope is published, there is no trust portal, and the claim is audited rather than a stated Type 2 certification with dates. Verified 20 September 2026.
Model Supply Chain Disclosure
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 vendor tells buyers to make every AI vendor name its models, and does not name its own. Its published guidance says a buyer should expect a two-part answer covering which foundation models the vendor uses and under what enterprise terms, and separately how customer data is treated. On its own estate the AI is described only by what each product reads and produces: no model, version or provider is identified, nothing states where inference runs, and no commitment to notify customers when the models change was located.
Checked the homepage, the platform page, two answer-library pages and the terms on 20 September 2026. Verified 20 September 2026.
Commercial Transparency
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.
No price and no unit of charge is published. The agreement sets out how commercial terms work, fees in an ordering document, net 30 payment, one-year terms with automatic renewal and 180 days' notice not to renew, late charges at one and a half per cent a month, but never what anything costs or what it is charged per. The estate publishes a page on how to price matter management and e-billing software including costs, without pricing its own, and the Council engagement is quoted only as ten weeks and four to five hours of the customer's time.
Checked the homepage, the platform page, the Council and answer library pages and the terms on 20 September 2026. Verified 20 September 2026.
Firm and Practice Coverage
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.
Two buyers are addressed on their own pages and the work is described precisely for each: corporate legal departments, aimed at general counsel, VPs of litigation and legal operations running a litigation portfolio, and property and casualty claims organisations, aimed at claims and litigation leaders working litigated claims alongside a claims system. The subject matter is defended litigation of the kind that produces nuclear verdicts, and the published answer library maps the territory in detail, from venue and jurisdiction risk to counsel selection, billing guideline enforcement and settlement timing.
The limits are not drawn: nothing states which lines of business, matter types or jurisdictions the platform does not serve, or how it behaves where defence counsel do not report in prose. Verified 20 September 2026.
6 public documents
The public pages on file for CaseGlide, 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.
-
caseglide.com/platform3 signals
Ethical Walls and Matter Segregation, Primary Law Corpus Provenance, Court Disclosure Support
Read Sep 20, 2026
-
caseglide.com/terms-of-service3 signals
Client Data in Training, Prompt and Output Retention, Third Party Request and Subpoena Notice
Read Sep 20, 2026
-
Refusal and Uncertainty Behaviour, Outside Counsel Guideline Readiness
Read Sep 20, 2026
-
caseglide.com/answers1 signal
Billing and Fee Posture
Read Sep 20, 2026
-
Bar Guidance Alignment
Read Sep 20, 2026
-
Fabricated Citation Record
Read Sep 20, 2026
Legal Signals
What each signal meansA 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.
Client Data in Training
Can material a lawyer puts into this product be used to train a model?
A published agreement or policy exists and none of it addresses the question either way, or the document that would answer it could not be read and the summary names the retrieval limit. The summary states which shape the silence takes: an improvement right granted that never names training, or no improvement right granted at all.
The published Terms of Service, dated 12 August 2020, never mention training, models or AI. They give the customer ownership of its information and reserve CaseGlide's right to use customer data "to the extent such data is aggregated and anonymized": a data use right granted that never names training, and does not exclude it either. CaseGlide's own buyer guidance, updated July 2026, tells buyers to require an explicit contractual bar on their case data training any model, and says CaseGlide should be held to the same standard as any vendor.
The published terms predate Case Clerk AI, Chronicle AI and Chambers AI and contain no such bar. They also state that a signed Ordering Document prevails over them, and that document is not published.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
Retention is acknowledged in public materials with no stated period.
Retention is addressed at the end of the relationship and nowhere else. The agreement, dated 12 August 2020, requires each party to return or destroy the other's confidential information on termination or expiry, and access to the platform ends immediately, but no retention period is stated for case files, extracted facts or chronologies during the term, and no deletion timetable is given. The vendor's own buyer guidance recommends zero data retention with named deletion terms in writing, which the published agreement does not provide.
Ethical Walls and Matter Segregation
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.
Separation between customers is asserted as a product guardrail rather than documented as a control: Chambers AI answers from the customer's own case history only and never benchmarks against another portfolio, and the vendor tells buyers to require that their data is never pooled. How that isolation is enforced is not described, and nothing addresses separation inside a customer, which matters where a claims organisation and its panel firms use the same file and a conflict would bar one team from a matter.
Third Party Request and Subpoena Notice
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 agreement, dated 12 August 2020, commits to notice before compelled disclosure: where a party must disclose confidential information to comply with a court order or applicable law, it must first give written notice to the other party and make a reasonable effort to obtain a protective order. That is a contractual commitment a customer can enforce rather than a policy statement. No transparency report or record of requests received is published.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
No located public material identifies the corpus behind the product’s answers.
The corpus is the customer's own litigation record and the vendor says so plainly, describing no black boxes and no borrowed data, with answers drawn only from the customer's own case files. No external legal corpus, licensed dataset or cross-customer benchmark is used or claimed, so questions of licensing and coverage do not arise in the usual form.
Good Law Verification
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.
Checked the homepage, the platform page, two answer-library pages and the terms on 20 September 2026. The product reads defence counsel work product and case files rather than citing legal authority, so no subsequent-history check arises and none is described.
Refusal and Uncertainty Behaviour
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.
There is no described no-answer path, and the vendor's published position is that the guardrail sits elsewhere: each product does one narrow job on the customer's own data, every output traces to a source a person can check in seconds, and the platform refuses a whole class of question by never predicting verdicts or scoring outcome risk. No confidence score, uncertainty flag or abstention behaviour is described for the extraction and chronology work itself.
Fabricated Citation Record
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.
Searched the AI Hallucination Cases database maintained by Damien Charlotin on 20 September 2026 on the name CaseGlide. No court order, opinion or disciplinary record naming the product was located. The vendor itself publishes a running account of the sanctions imposed on others, which is commentary rather than a record about this product.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Public materials refer to professional responsibility in general terms without naming guidance.
The estate engages the duties directly and reaches them through courts rather than through bar guidance. Published pages work through the verification duty at the point of signing and filing, the privilege consequences of running client material through consumer-tier tools, and the standards a legal department should impose on its panel firms, citing sanction decisions, a Delaware order putting a firm's AI policies on the docket and the joint-responsibility clause of Rule 11.
Those are court decisions and a procedural rule; no bar ethics opinion, professional conduct rule or bar-issued AI guidance is named anywhere.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
The buyer is the payer, not the biller. The product exists to review, author or benchmark legal fees, so the record of fees is the product itself rather than a byproduct of AI assisted lawyering. Typically sold to in-house teams paying outside counsel.
The fee record is the subject matter on one side of this product. The platform includes e-billing for defence counsel invoices, and the published material works through guideline enforcement, the billing violations to watch for, rate increase requests, alternative fee arrangements and whether insurance defence should be flat fee or hourly, all from the position of the organisation paying the bill rather than the lawyer sending it.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
The material exists behind a sales conversation or an executed agreement.
The material a customer would hand to its own stakeholders exists and is unusually good, covering what grounds the AI, how to test traceability, what to require in a master agreement and what AI standards to put in outside counsel guidelines, alongside the offer of security documentation on request. What it does not contain is the list itself: no subprocessor list and no model provider is published, which is the first item the vendor's own guidance tells buyers to demand.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Some elements of the record are available, short of a document level export.
Most of a disclosure record exists as a by-product of the design. Every extracted fact links to the defence counsel report it came from, every chronology entry links to its source document by page and line, and Chambers answers name the case files behind them, so what the AI produced and what it rests on can be shown. What is not recorded is which model produced a given output or who verified it, and no export built for a court's AI disclosure requirement is described.