Aderant vs Intapp: how they compare in 2026
Aderant and Intapp meet at the bill and part ways around it. Both sell timekeeping and outside counsel guideline tools to large law firms, Aderant through iTimekeep and Onyx and Intapp through Intapp Time and Terms, and both now use AI to draft time entries. Beyond that overlap, Aderant runs the firm's finances in Expert Sierra, an AWS hosted system that more than 275 firms have adopted, and its Agent Center prepares billing appeals, collections, rates and forecasts for staff to approve. Intapp runs the risk side, from intake and conflicts to ethical walls, and its Celeste platform lets administrators choose Anthropic or OpenAI models while Walls for AI applies the firm's barriers to Harvey, Copilot and Legora. Intapp's AI Addendum puts its no training commitment in published contract terms, while Aderant states its training exclusions in a June 2025 article. Firms whose priority is collecting faster and pricing work should look first at Aderant. Firms whose priority is governing which people and AI tools see which matters should look first at Intapp.
Which one fits
Choose Aderant if
- Collections, appeals and rates are where your finance team loses time. Aderant's Agent Center offers seven agents that prepare electronic billing appeals, prioritize collections, explain rates, recommend codes and build general ledger forecasts for the team to review and decide.
- You are moving financial management off your own servers. Expert Sierra runs on AWS for a single fixed three year subscription with hosting and maintenance included, and Aderant says more than 275 firms have adopted it.
- Your lawyers use Harvey and you want that work on the timesheet. Completed Harvey work arrives in iTimekeep as a draft entry with matter context, duration and a narrative, which the lawyer edits and submits through normal approvals.
- Docketing sits with the same vendor as billing. CompuLaw supplies more than 3,000 court rule sets maintained by Aderant employed attorneys, alongside Milana and Deadlines.com.
- You want customer results on collections and time capture. Michael Best & Friedrich reports its collection cycle falling from 55 to 47 days, and Bernstein Shur cut time entry lag from three days to 0.7 days.
Choose Intapp if
- Your lawyers use several AI tools and the walls must hold in all of them. Intapp Walls for AI applies a firm's ethical walls to people, agents and AI tools, naming Harvey, Microsoft Copilot and Celeste, and Legora joined in August 2026.
- Clients ask for AI terms in writing. Intapp's AI Addendum, effective 13 April 2026 and published with its product terms, says it will not use customer data to train or develop any AI models unless the customer agrees in writing.
- You want to choose the model. Administrators can power Celeste with Anthropic Claude or OpenAI GPT models, and the subprocessor list names Amazon Bedrock and Microsoft Azure AI Services by product and region.
- Intake and conflicts are the bottleneck. Intapp has sold intake and conflicts software to law firms for two decades, and Compliance with Celeste, launched in August 2026, adds agents for intake, conflicts and terms.
- Procurement wants to verify certifications at source. Intapp names ISO 27001, 27017, 27018 and 27701, SOC 1 and SOC 2 and CSA STAR, with Schellman as auditor and certificates linked from its compliance page.
At a glance
All 15 axes, side by side
The same grid applied to every vendor in the index, graded from public sources. 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.
MADDI supplies model based capabilities within Aderant's established law firm business systems. askMADDI answers plain language questions in Stridyn Analytics and the cloud general ledger, assists with guideline rules in Onyx and suggests time narratives in iTimekeep. Agent Center adds seven specialist agents that prepare billing appeals, prioritize collections, draft evaluations, recommend codes, explain rates and build forecasts. Four agents were already in an early access program when Aderant announced it on 25 August 2026, and three more became available at ILTACON that month, each using data from the products a firm licenses. Expert Sierra still runs accounting, matters, billing and expense workflows without those agents, so the models power operational capabilities within a wider platform rather than replacing the system of record.
Models do real work on a platform that ran for two decades without them. Intapp Celeste, generally available since 15 July 2026, is branded as the agentic AI platform for professional firms and is sold with individual products, such as Time with Celeste, which drafts complete time entries from the work it sees, and Compliance with Celeste, launched on 19 August 2026 with agents for intake, conflicts and terms. Intapp Assist is built into Time, Terms and DealCloud, with Assist SKUs for Conflicts and Intake in the October 2026 product terms, and its Ask Intapp feature answers questions about client obligations in Microsoft Teams with links to source records. The larger half is not models: intake, conflicts, AML, walls, timekeeping, billing, workspaces and relationship management are systems of record a firm would still buy without the generative features.
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.
The Compliance Agent identifies requirements in outside counsel guidelines and cites the underlying language when helping create and test billing rules. Other agents keep a record of the information used and work prepared, while askMADDI answers from the firm's financial and operational data. Aderant describes safeguards, validation and performance monitoring, but does not supply an error rate, test set or worked example that demonstrates how a reviewer opens every supporting source. The relevant failures include an incorrect ledger answer, an unsuitable billing code or a misread client requirement. These are financial and contractual assertions rather than case-law citations. Source grounding is claimed for specific tasks, without enough detail to independently assess its reliability across the agent suite.
Grounding is real and documented, limitations are published with real candor, and nothing is measured. Ask Intapp is described as answering only from the firm's own Intapp Terms data and returning answers with direct links to the source information, so the answer is bounded to a known source and traceable to it. The AI transparency statement publishes five limitations. Outputs may be inaccurate, incomplete or context dependent; performance depends on the quality and context of input data; no specific accuracy level is guaranteed unless stated; outputs are no substitute for professional judgment or expertise; and outputs are not guaranteed to be error free, uninterrupted, consistent, up to date, accurate, complete or free from bias. A separate AI Disclaimer sits in the product terms. No accuracy figure, error rate, test set, benchmark or evaluation is published for Assist, Celeste or any other AI feature, and no check beyond the source link is described. A reader can see what the vendor declines to promise, but not how often the products are right.
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.
Agent Center prepares, prioritizes and drafts work for the firm's team to review and decide. Agents use authorized data from licensed products and follow existing product permissions and firm controls. Their record of source information and prepared work gives reviewers context for accepting or changing a result. The Harvey integration provides a concrete approval path: completed legal work becomes a draft iTimekeep entry, then a lawyer reviews, edits and submits it through the firm's normal process. The agent descriptions stop at recommendations, draft messages and forecasts. They do not specify escalation thresholds, mandatory stops for uncertain answers or recovery after an incorrect recommendation. The firm's reviewers make the final operational decisions.
Intapp states the autonomy boundary and builds approval into Celeste. The AI transparency statement says its products are not meant for fully autonomous decision making, lists automated decision making as a prohibited use, and puts responsibility for checking outputs and final decisions on the customer and user, adding that Intapp applies no human oversight to a user's use of the products. The Celeste governance page says explicit approval is required before every write, and the Celeste page says it requires human approval for sensitive actions. Nothing published says what happens after an output proves wrong, and no confidence threshold or configurable autonomy setting is described.
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.
Chartwell Law's June 2026 customer story describes moving a nearly 300 attorney firm from internal infrastructure to Expert Sierra, with Chief Information and Operations Officer Maor Levy explaining the consolidation of billing, timekeeping and matter information. Michael Best & Friedrich reports its collection cycle falling from 55 to 47 days, and Buchanan Ingersoll & Rooney's story followed in August 2026. iTimekeep customers report gains too: Butzel Long tripled time entry velocity, and Bernstein Shur cut time entry lag from three days to 0.7 days. Expert Sierra testimonials attribute a reduction of more than half in prebill time to Holland & Knight and halved receipt processing time to Kostelanetz. Aderant says more than 275 firms have adopted Sierra, that its Net Promoter result stays above 90, and that it has had no failed enterprise implementations. None of the figures carries a measurement method, and none concerns Agent Center or askMADDI.
Intapp is listed on NASDAQ, and its first quarter of fiscal 2027, reported on 27 October 2026, brought revenue of $161.3 million, cloud ARR of $515.3 million, 908 clients above $100,000 in ARR and cloud net revenue retention of 123 percent, with Baker Donelson, Bryan Cave Leighton Paisner, Practus, Shoosmiths and Troutman Pepper Locke named among legal wins. Its homepage carries named customer results: Mintz reports 35 to 40 percent faster prebill review, McCarter & English about 50 percent faster draft invoice review, Loeb & Loeb up to four times faster request processing and Sandberg Phoenix a threefold increase in matter openings. Fredrikson & Byron's case study describes centralizing client guidelines in Intapp Terms. The figures carry no method or date, and none is tied to Assist or Celeste.
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.
In a June 2025 article, Aderant states that confidential, intellectual property, personal and sensitive information is excluded from machine learning training, and that it uses aggregated, synthetic or customer partnered material with masking and filtering instead. That statement sits in an article rather than a contract. Section 5(b) of the API terms, version 11 September 2020, gives Aderant a perpetual license to submitted content for providing, securing and improving the APIs and related services, without an express model training exclusion, and no subscription agreement, data processing agreement or AI terms are published. The privacy notice separately excludes personal data processed on customers' behalf. Agents follow product permissions, but the specific matter wall configuration and model provider handling are not explained.
Intapp's AI Addendum, effective 13 April 2026, says it will not use customer data to train or develop any AI models unless the customer agrees in writing, and rules out eyes on human review of customer data for abuse monitoring. The Celeste governance page adds that prompts and outputs are never used for model training, that reads run within each user's own permissions, and that ethical walls, insider lists and information barriers are enforced through Intapp Walls for AI each time a playbook runs, including playbooks called from other AI tools. Wall enforcement depends on holding Walls for AI, and Intapp Assist, the assistant inside Time, Terms and DealCloud, has no matching wall statement. Privilege and work product are not addressed by name, and the retention period after the agreement ends is not stated.
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 AI supports business operations inside law firms: time narratives, billing appeals, collections, pricing, financial forecasts and associate evaluations. The product audience is finance, operations, talent and legal staff within a firm. Human review is part of the agent workflow and the Harvey time-entry integration. Neither those descriptions nor the API terms establish a specific boundary between software assistance and regulated legal advice. The API terms require lawful use and respect for third-party rights, without connecting the AI features to lawyer competence, professional supervision or jurisdiction-specific duties. A firm's approval of a generated time narrative or interpretation of a client's billing guideline remains a professional decision; the product supplies no separate ethics or advice framework for it.
Intapp's products run intake, conflicts, AML checks, ethical walls, outside counsel guideline compliance, timekeeping, prebilling and relationship management, and its AI features answer questions about a firm's own obligations and internal data. They give no legal advice, their users are a firm's professional staff rather than the public, and there is no legal research output or client advice surface. The AI transparency statement says outputs are no substitute for professional judgment or expertise and that users must check all outputs before relying on them. The ethical walls page describes the product as fulfilling a firm's ethical duty to its clients and profession. No rule of professional conduct, bar guidance, ethics opinion or court practice direction is named.
AI Governance and Bias Disclosure
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.
Aderant organizes its AI position around four principles: firm centered design, responsible development, legal expertise and trust. A June 2025 article describes controlled orchestration, validation, quality monitoring and a data strategy using aggregated, synthetic or customer partnered data with masking, anonymization and filtering. Lisa Erickson is identified as the executive responsible for product management and AI, without a formal governance responsibility matrix or release approval process. The vi suite added AI performance reviews, sentiment analysis and summaries in March 2026, and the Talent Agent drafts associate evaluations from performance history and feedback, with no assessment of how either performs across employee groups. Aderant publishes no bias test protocols or results.
Intapp's AI transparency statement, effective 2 August 2026, covers all its AI products. It explains how they process input data with rule based logic, statistical models and machine learning, describes them as low risk AI systems under the EU AI Act, commits to telling users when they deal with an AI system and to marking AI generated output, lists prohibited uses including automated decision making, and offers a complaints route. The AI Addendum adds a contractual commitment to monitor and evaluate model performance against testing benchmarks. No evaluation results, red team findings or bias measurements are published, no one is named as accountable for model behavior, and the statement says of itself that it is not legally binding, so the binding commitments sit in the addendum.
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.
Agent access follows the permissions and controls in the firm's licensed Aderant products. Expert Sierra uses AWS hosting, security monitoring and auditing, and Aderant says every AI run it powers is observable. Product specific SOC 2 examinations cover security, availability and confidentiality. Aderant lists its subprocessors openly and offers security whitepapers, penetration test reports and diagrams through a trust center access process. The open product material gives no specific deadlines for prompt retention, agent log deletion, customer data return or incident notice. The privacy notice, last updated 8 March 2024, gives general retention and individual rights provisions for data Aderant controls, while excluding its processing on customers' behalf.
Intapp publishes security and data policy across its AI products, with independently audited certifications reachable from its compliance page. The AI Addendum, effective 13 April 2026, bars training on customer data unless the customer agrees in writing, rules out eyes on human review of customer data for abuse monitoring, and commits Intapp to monitor and evaluate model performance against testing benchmarks. The Celeste governance page says prompts and outputs are never used for model training and that client data is stored for the duration of the client agreement. The subprocessor list names each subprocessor by product, activity and location with change notices, and listed safeguards include encryption in transit and at rest, role based permissions, data minimization, audit logging and secure development. Neither the addendum nor the Celeste pages give a retention period after the agreement ends or an incident notice deadline.
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.
Aderant's API terms allocate losses for use of the Expert APIs and related developer services. They supply the APIs as is, disclaim reliability and content warranties, exclude consequential losses and cap liability at the amount paid for the applicable APIs during the preceding year. Business users indemnify Aderant for misuse, terms violations and content routed through their integration. The terms expressly recognize separate licenses and other product agreements, which can control conflicts. They therefore provide a standard limitation for integration exposure rather than a warranty on MADDI output across the suite. No equivalent open subscription clause establishes AI accuracy remedies, an AI indemnity or insurance available to the customer for a wrong billing recommendation.
Intapp's Artificial Intelligence Addendum, effective 13 April 2026, is published with its product terms and leaves the risk of AI output with the customer. Section 9(a)(v) states that Intapp is in no event liable for inaccuracies, errors or omissions, and the AI transparency statement adds that outputs are not guaranteed to be error free, accurate, complete or free from bias, with the customer and user responsible for decisions taken. The Intapp Assist product terms, effective 21 October 2026, sit alongside the addendum. The master agreement the addendum depends on is not public, so no indemnity, liability cap, service level or insurance position for AI output can be read before a commercial conversation.
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 Harvey partnership, announced in December 2025, moves completed work into iTimekeep as draft entries containing matter context, duration and a generated narrative, and sends context from a firm's Aderant systems back to Harvey. Lawyers can edit the draft before submitting it through their existing billing approvals. A First AML connection brings due diligence checks into Expert and Expert Sierra through Aderant's professional services team, and a multi year renewal of the LawPay partnership covers payments, with a BillBlast integration announced as coming. Expert Sierra connects practice and financial records within the Stridyn platform. The Expert API terms describe credentials, usage limits and permission to build client integrations, but do not serve as an endpoint reference or installation guide, and installation steps and permission mappings are not documented.
Integration runs deep into the practice stack. Celeste lists MCP and Snowflake connectors, operates inside Microsoft Copilot and reaches ChatGPT Enterprise through a plug-in released on 17 September 2026. Under the Harvey partnership announced on 23 February 2026, Walls for AI policies apply to Harvey Assistant, Vault and Workflows, and the Legora partnership of 25 August 2026 brings Legora into Walls for AI and turns Legora usage into Intapp Time entries, in a phased rollout. Ask Intapp runs as a Microsoft Teams app, Intapp Workspaces delivers matter workspaces in Microsoft 365, and the Intapp Integration Service runs on Boomi. Data providers named as subprocessors include Dun & Bradstreet, S&P Global Market Intelligence, Grata, ESRI, Plaid and Gresham, and Moody's data joined Celeste in July 2026. No API or developer documentation is published, so connector depth rests on the partner announcements.
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.
Expert Sierra is Aderant's AWS-hosted cloud practice and financial system. The self-hosted Expert product remains a distinct deployment, while Sierra adds Stridyn, newer financial applications and AI capabilities. Its cloud service includes infrastructure maintenance, security monitoring and platform upgrades, allowing customers to retain existing customizations and integrations during migration. The product page identifies AWS but does not specify a tenant-isolation model, available countries or where MADDI inference occurs relative to stored firm data. Aderant's worldwide offices and the privacy notice's treatment of website information say nothing about where cloud customer data resides. The hosting provider and delivery model are clear; the regional and tenancy details required for a sensitive matter deployment remain unsettled.
Hosting regions are named for each product across the United States, Ireland, the Netherlands, the United Arab Emirates, Singapore, Australia and Canada, depending on the customer's hosting location, with routing exceptions published, such as United Arab Emirates email traffic passing through the European Union and Singapore traffic through Australia. AI features run in the United States or the European Union according to the customer's hosting location, and customers outside both are processed in the European Union. DealCloud has four regional instances with separate login endpoints, and the integration layer processes data in the customer's chosen region. Celeste stores client data for the duration of the client agreement. Tenancy is not described: nothing says whether customers are single tenant or multitenant, or whether isolation is logical or physical.
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.
Schellman & Company examined Expert Sierra, vi by Aderant and iTimekeep for SOC 2 Type 2 covering October 2024 through September 2025, and a separate Onyx examination covers November 2025 through January 2026. Both announcements identify the security, availability and confidentiality criteria. Aderant's trust center lists SOC 2 Type 2 and GDPR, names its subprocessors openly, and offers the reports, security whitepapers, penetration test reports, network diagrams, a SIG Lite questionnaire and a cyber insurance item through an access process. It also posts security advisories, such as an October 2026 notice on Node.js vulnerabilities. The underlying report contents are not openly available, and the examination scopes do not separately identify the suite wide AI layer.
Intapp names ISO 27001, ISO 27017, ISO 27018 and ISO 27701, SOC 1 and SOC 2, CSA STAR registration and a Global Privacy Recognition for Processors certification, with Schellman as auditor. The ISO and privacy processor certificates link to Schellman's certificate service, and the CSA STAR entry links to the Cloud Security Alliance registry for Integration Appliance, Inc., so a reader can confirm the claims at source without a request. A Shared Assessments SIG 2024 Lite questionnaire, a critical subcontractors page for DORA, a public cloud status site and individual cloud policies are published, and the Celeste pages describe the platform as CREST approved. Intapp states its GovRAMP position as membership while it pursues verified status. The compliance page gives no audit period or scope, so those come from the linked certificates, and SOC reports are not downloadable.
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.
MADDI combines Aderant's domain specific technology with enterprise model platforms, and a 2024 article from Aderant's chief executive describes it as powered by multiple language models. No model family, version or provider is named. askMADDI and the operational agents use the firm's data from licensed Aderant products. The trust center's open subprocessor list names Coforge, Braze, Auth0, Atlassian and AWS, none of them a model provider, and AWS hosts Expert Sierra. Harvey is a separately named integration partner that exchanges data with iTimekeep, not an identified foundation model provider for MADDI. No open commitment states when customers would be notified about a change of model or AI provider.
Administrators can choose to power Celeste with Anthropic Claude or OpenAI GPT models, and the subprocessor list, dated 11 August 2026, is organized by product. For Celeste it names Amazon Bedrock hosting Anthropic models across six regions, Microsoft Azure AI Services for text generation and extraction, and Exa Labs for web grounding. For Intapp Assist it names Microsoft Azure AI Services, and Anthropic PBC is named as a generative AI provider for other products. Change notice works through an effective date, a subscription form and kept prior versions, and the AI Addendum says Intapp will not introduce a new generative AI tool without the customer's knowledge or approval. Model versions within each family are not named, and the list predates the September 2026 plug-ins for ChatGPT Enterprise and ChatGPT for Financial Services.
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.
Expert Sierra is sold through a custom subscription proposal. Aderant's August 2026 cost explanation describes Sierra's price as a single fixed number for a three year subscription, with hosting and infrastructure maintenance included, separate from one time migration, implementation and training. Its product matrix distinguishes self hosted Expert from Sierra's newer financial applications, Stridyn platform and AI capabilities. iTimekeep is sold in a Premium plan with a free trial. Neither the cost explanation nor the Agent Center material gives a starting figure, user allowance, usage charge or surcharge for particular agents, and agent eligibility depends on the products already licensed.
Intapp sells by quote, and its product terms show how the AI is packaged. The Intapp Assist product terms, effective 21 October 2026, list 46 AI SKUs by product, such as AI Features for Time, Intapp Time with Celeste for Legal and Intapp Walls for AI, and the August 2026 launch of Time with Celeste and Compliance with Celeste describes them as a new entitlement. A buyer can therefore see which AI capabilities are sold separately for each product, so the shape of the price is visible. No price or unit of charge is published, and no minimum or term is stated, with every product page ending at a demo or contact request. Intapp's quarterly reports give company revenue, which does not translate into a price for an individual firm.
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.
Law firms get practice finance, billing, collections, timekeeping, docketing and people management from Aderant. Expert Sierra includes client and matter records, conflict checks, trust accounting, tax, disbursements and financial reporting, and CompuLaw supplies more than 3,000 court rule sets. Aderant says it serves more than 2,500 clients in 26 countries, including 98 percent of the Am Law 200 and 86 percent of the Global 100, and its customer examples include Chartwell Law, Holland & Knight, Michael Best & Friedrich, Buchanan Ingersoll & Rooney and Kostelanetz. That breadth concerns running a firm across practice groups rather than supplying jurisdiction specific legal analysis. The corporate legal department paying an outside firm is not the main buyer. No minimum firm size or matrix maps each agent to a license and customer segment.
Seven industries have their own solution sets: legal, accounting, consulting, corporate, investment banking and advisory, private capital and real assets. Legal has more than a dozen solutions, covering client intelligence, Microsoft Copilot enablement, Microsoft 365 collaboration, timekeeping, prebilling and billing, intake, conflicts, outside counsel guidelines, AML and know your client, ethical walls, lateral hire onboarding and partner attestation, and in house legal departments get matter and relationship management. Intapp says 97 of the Am Law 100 use its software and that more than 70 Am Law 100 firms rely on Intapp Walls. Offices span the United States, United Kingdom and Asia Pacific, with seven named hosting regions. No firm size, practice area or jurisdiction is named as out of scope, and no matrix shows which products carry AI features.
The 12 legal signals, side by side
Recorded rather than graded. These are the questions a practitioner has to answer before a tool touches a client matter, and the answers are taken from public material only.
Client Data in Training
Can material a lawyer puts into this product be used to train a model?
The API terms grant a perpetual license to submitted content to provide, secure and improve the APIs and related services. The clause reserves the right to "provide, secure, and improve the APIs" without expressly excluding model training, which supports contractual permission for that API content. Aderant separately says confidential, intellectual property, personal and sensitive information is not used for training and describes aggregated, synthetic or customer-partnered inputs.
The API permission and that policy statement have different scope. The terms recognize separate product licenses, so this is not evidence that the entire hosted suite trains on client material.
Intapp's Artificial Intelligence Addendum, effective 13 April 2026 and published with its product terms, says Intapp will not use customer data to train or develop any AI models unless the customer agrees otherwise in writing. The Celeste governance page adds that prompts and outputs are never used for model training, and the addendum rules out eyes on human review of customer data for abuse monitoring. The AI transparency statement separately says Intapp uses data about the administration, configuration, support, use and performance of its products to operate and improve them, so usage data still feeds improvement while customer content is excluded.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
Agent Center keeps a record of the information used and the work prepared, without stating how long those records or AI prompts remain available. The privacy notice, last updated 8 March 2024, gives general retention language for personal data Aderant controls and excludes information processed for customers through the products. The API content license survives termination, but a license duration is not a deletion schedule.
The open material gives no terms for customer data return, prompt removal or backup retention in the hosted AI suite. Security whitepapers are available through the trust center's access process.
Retention is addressed, with no period attached after the agreement ends. The Celeste governance page says client data is stored for the duration of the client agreement, and the AI transparency statement lists data minimization and retention limits among its safeguards. No retention period is given for prompts, uploaded content or generated outputs once the agreement ends, and no customer setting or deletion control is described. The material involved is a firm's timekeeping, intake, conflicts and client obligation records, not a chat history.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Agents use authorized information from the firm's licensed Aderant products and follow product permissions and firm controls. Access is therefore inherited from the host applications, rather than unrestricted across all firm data. The open product descriptions do not specify role setup, client or matter walls, denial behavior, or how a change in membership affects saved agent context. This matters for associate evaluations as well as financial records, since the Talent Agent combines performance history and feedback. Detailed security material requires trust center access.
For Celeste, Intapp documents how walls are enforced. Its governance page says reads run within each user's own permissions, enforced at the connector level and carried through every step, and that Celeste enforces ethical walls, insider lists and information barriers through Intapp Walls for AI each time a playbook runs, including playbooks offered to other AI tools through MCP. The Walls page extends enforcement to all people, agents and AI tools, naming Harvey, Microsoft Copilot and Celeste, and Legora joined Walls for AI in August 2026.
Wall enforcement requires the Walls for AI product, and Intapp Assist, the assistant built into Time, Terms and DealCloud, has no matching statement.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
The privacy notice permits disclosures to law enforcement and regulators for data Aderant controls, without promising advance notice. It excludes personal information processed on customers' behalf, so that clause does not govern a firm's hosted matter and billing records. The API confidentiality clause requires the customer to notify Aderant before a compelled disclosure of Aderant's confidential information where permitted; it is not a reciprocal promise about customer data. The open terms contain no notice or challenge commitment for government requests affecting the hosted suite.
No compelled disclosure position, notice commitment or promise to seek protective relief appears in the AI transparency statement, the trust and compliance pages or the subprocessor list, and Intapp publishes no transparency report or government request policy. The master agreement is not public, so a firm settles notice of a compelled demand in its own contract and data processing addendum.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
A firm's own financial records, time entries, rates, performance information and client guidelines feed Aderant's operational AI. The Compliance Agent cites the guideline language behind a billing requirement. These inputs are customer business records rather than a licensed collection of statutes or judgments. A June 2025 Aderant article describes aggregated, synthetic and customer partnered training material without identifying a dataset or its license.
The underlying model platforms are also unnamed. Legal work completed in Harvey can enter iTimekeep as a draft time record, but that connection does not turn MADDI into a legal research corpus.
The AI answers from the firm's own records: Ask Intapp answers only from the firm's Intapp Terms data, and Celeste works across the firm's connected systems. No body of law assembled by Intapp sits behind the outputs. Third party content is named where it enters: the subprocessor list identifies Dun & Bradstreet, S&P Global Market Intelligence and Grata as content providers and ESRI and Mapbox for mapping, and Moody's data joined Celeste in July 2026. The license basis on which that content reaches a firm is not stated.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
MADDI's answers concern firm finances and operations, and its agents work on billing, guidelines, time, rates and evaluations. A client's outside counsel guideline can be cited as the source of a requirement, but that is a contractual source rather than a legal authority validated for later treatment. The suite also contains docketing products, which are distinct from the financial AI described here. No citator or good-law validation path accompanies askMADDI or Agent Center's output, and grounding in business records does not supply one.
Nothing the platform produces cites legal authority. The AI features answer questions about a firm's own client obligations, time records, conflicts data and relationships, and their outputs cite the firm's own records rather than decisions, statutes or secondary sources, so there is nothing for a citator to cover. The nearest equivalent is whether an outside counsel guideline held in Intapp Terms is the version in force, which the firm manages.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
The GL Forecasting Agent supplies confidence ranges and variance analysis alongside its projections. Those ranges express forecast variation, rather than a stated warning that a generated answer lacks reliable support. Other agents prepare billing appeals, time narratives, codes and rate explanations for review, but their descriptions do not define a low-confidence indicator, clarification request or refusal path. The human reviewer can change or reject a draft.
That review process supplies oversight without establishing how the system behaves when source data is incomplete or contradictory.
The AI transparency statement says outputs may be inaccurate, incomplete or context dependent, that performance depends on input data, and that no accuracy level is guaranteed unless stated. Those describe the risk without saying what the product does about it. Nothing says Assist or Celeste declines a question it cannot ground in the firm's data, reports that no responsive record exists, shows a confidence signal or flags incomplete records.
The statement's answer to uncertainty is to assign it: users must check all outputs and report unexpected or incorrect behavior to support.
Fabricated Citation Record
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
Damien Charlotin's AI Hallucination Cases database has no tool entry naming Aderant, MADDI, askMADDI, iTimekeep or Onyx. A textual occurrence of Onyx concerns a purported case title in Vision Management Group v. Constant Aviation, not this vendor or its software. Aderant's AI mainly produces financial explanations, billing narratives and operational drafts. That scope differs from generating case-law authorities for filing, and the court database does not measure errors in invoice calculations, codes or internal financial answers.
Damien Charlotin's AI Hallucination Cases database, which names legal products as well as general assistants among its 2,262 cases, has no entry naming Intapp, Integration Appliance, Inc., Assist or Celeste. Intapp's AI features answer questions about a firm's own obligations, time records and conflicts data rather than producing citations for filing, and the transparency statement warns that outputs may be inaccurate or incomplete.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Aderant's AI principles and API terms address responsible development, lawful use and respect for third-party rights. They do not map its billing or guideline features to a named bar opinion or professional conduct rule. The Time Agent rewrites narratives and suggests billing codes, and the Harvey integration turns completed work into draft time entries for review. Those workflows preserve the firm's normal approval process but supply no separate rule for charging for AI-assisted work or informing a client. Responsibility for the bill remains with the firm.
Professional judgment is mentioned in general terms, and no bar or regulatory guidance on AI is named. The AI transparency statement says outputs are no substitute for professional judgment or expertise, that users must check all outputs before relying on them, and that the customer and user are responsible for decisions taken. The ethical walls page presents its product as fulfilling a firm's ethical duty to its clients and profession.
Intapp engages named regimes in detail elsewhere, describing its EU AI Act classification and publishing a DORA subcontractor page, but no rule of professional conduct, ethics opinion or court practice direction is cited.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
Aderant sells to the law firm issuing the bill. Its Harvey integration places matter context, duration and an AI-generated narrative into a draft iTimekeep entry, which a lawyer reviews, edits and submits. The release does not say that the resulting time record is marked as AI-assisted or attributed to Harvey. The integration therefore captures work without establishing an identifiable AI billing record. Aderant also promotes reduced administrative effort and faster conversion of work into revenue. Its materials do not explain how saved time changes client charges or what the firm should disclose.
Intapp's AI now writes time entries inside the product built to govern the bill. Time with Celeste, launched in August 2026, drafts complete entries from the work it already sees, Intapp Time checks entries against firm and client requirements before submission, and Billstream and Terms apply the same rules to prebills and invoices. The launch material notes that clients increasingly expect pricing that reflects those efficiencies, and Ask Intapp is said to cut the time lawyers and billing analysts spend on client obligation questions.
Nothing marks an entry as AI drafted, keeps a per matter record of AI assisted work, or gives guidance on fee or disclosure treatment.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
Aderant's trust center lists its subprocessors openly, naming Coforge, Braze, Auth0, Atlassian and AWS, which tells a client where Aderant's data sits. None of them runs the models behind MADDI and Agent Center, and no Aderant page says who does, so the list does not tell a client who sees its content through the AI. SOC reports, security whitepapers, penetration test reports and network diagrams sit behind an access process, and no client facing AI consent package is offered, so a firm still needs Aderant's answers before forwarding a complete disclosure.
Onyx's own role is different: it converts a firm's client guidelines into billing controls, which does not itself supply the information a client needs about Aderant's model and data chain.
A firm can answer a client's AI clause from published material. The subprocessor list, dated 11 August 2026, is ungated and organized by product, activity and location, with prior versions and a subscription for change notices. It names Amazon Bedrock hosting Anthropic models for Celeste, with the customer choosing Anthropic or Azure OpenAI models, Microsoft Azure AI Services for Intapp Assist, Exa Labs for web grounding and Anthropic PBC for other products.
The public AI Addendum, effective 13 April 2026, gives a firm a contract document it can pass to a client, alongside the data processing addendum, the AI transparency statement, a SIG Lite questionnaire and certificates linked to the auditor. The list predates the September 2026 ChatGPT plug-ins.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Agent Center retains a record of the information used and the work prepared, giving a reviewer some basis for understanding an AI-assisted operational task. The Compliance Agent also cites client-guideline language when preparing rules. No product description specifies an export naming the model, recording human verification and distinguishing generated from edited text for each document. The typical output concerns billing, collections or internal management rather than a court filing. The available source and draft information does not identify the model or human verification for a court submission.
Celeste keeps an action record a firm can use with clients and regulators. Intapp's general availability announcement says the firm can show clients and regulators a full record of every action, and Intapp Walls logs AI tool access and blocked attempts. The AI transparency statement says users are told when they deal with an AI system and that outputs are marked as AI generated. No per document export naming the model, the sources retrieved and the human verification is described, and Intapp Assist has no matching record.
The questions both sides leave open
Derived from the records above rather than written, so it cannot favor either vendor. Take these into both conversations and ask each side the same question.
- Third Party Request and Subpoena Notice
- Good Law Verification
- Refusal and Uncertainty Behavior
Product updates
The latest logged changes from each vendor, newest first, each linked to its source.
Intapp
-
Integration / interoperability
Intapp launched a Celeste plugin for ChatGPT as part of OpenAI's Astra for Law launch.
Source -
Integration / interoperability
Intapp released a DealCloud with Celeste plug in for ChatGPT for Financial Services, available to eligible Intapp clients from 10 September.
Source
In summary
Aderant
Aderant, an Atlanta company within Roper Technologies, builds the systems law firms run their business on: Expert Sierra for finance, matters and billing, iTimekeep for time, Onyx and BillBlast for guidelines and bills, CompuLaw and Milana for docketing, and the vi suite for talent. MADDI answers business questions from the firm's own data, and Agent Center prepares billing, collections, evaluation, rate and forecast work for review. According to the AI Legal Index, Aderant serves more than 2,500 clients in 26 countries, including 98 percent of the Am Law 200, and lists its subprocessors openly. A June 2025 Aderant article says confidential and sensitive information is excluded from model training. Deane Price has been acting chief executive since August 2026.
Intapp
Intapp, listed on NASDAQ as INTA, sells cloud software for law firms and other professional firms, covering intake, conflicts, AML checks, ethical walls, outside counsel guidelines, timekeeping, billing and relationship intelligence through DealCloud. Celeste, generally available since July 2026, is its agentic AI platform, sold with products such as Time with Celeste, and Intapp Assist adds AI inside Time, Terms and DealCloud. The AI Legal Index records a public AI Addendum that bars training on customer data and a subprocessor list naming model providers by product and region. Intapp says 97 of the Am Law 100 use its software, and Mintz reports 35 to 40 percent faster prebill review.
Questions buyers ask
What is the difference between Aderant and Intapp?
Aderant runs a firm's finances, billing, docketing and talent, and its agents prepare collections, appeals, rate and forecast work. Intapp runs the risk and client side, from intake and conflicts to ethical walls, and its Celeste platform applies those walls to AI tools. They overlap in timekeeping and outside counsel guideline compliance, where both now draft time entries with AI. From the AI Legal Index, based on each vendor's own published materials as of November 3, 2026. No vendor pays for placement.
How much do Aderant and Intapp cost?
Both sell by quote. Aderant describes Expert Sierra as a single fixed price for a three year subscription, with hosting and infrastructure included and migration, implementation and training priced separately. Intapp sells each product by quote, and its product terms list AI features as separate SKUs, such as Intapp Time with Celeste for Legal and Intapp Walls for AI. From the AI Legal Index, based on each vendor's own published materials as of November 3, 2026. No vendor pays for placement.
Do Aderant and Intapp train AI on law firm data?
Both say no. Intapp's AI Addendum, effective 13 April 2026, says it will not use customer data to train or develop any AI models unless the customer agrees in writing. Aderant's June 2025 article says confidential, intellectual property, personal and sensitive information is excluded from training, while its API terms give it a license to submitted content to provide, secure and improve the APIs. From the AI Legal Index, based on each vendor's own published materials as of November 3, 2026. No vendor pays for placement.
Does their AI respect ethical walls?
Intapp's Celeste runs every read within the user's permissions and enforces walls through Intapp Walls for AI each time a playbook runs, which requires that product, and Walls for AI extends to Harvey, Microsoft Copilot and Legora. Aderant's agents follow the permissions and controls in the firm's licensed Aderant products. From the AI Legal Index, based on each vendor's own published materials as of November 3, 2026. No vendor pays for placement.
Which AI models do Aderant and Intapp use?
Intapp lets administrators power Celeste with Anthropic Claude or OpenAI GPT models, and its subprocessor list names Amazon Bedrock and Microsoft Azure AI Services by product and region. Aderant says MADDI combines its own domain technology with enterprise model platforms and multiple language models, and AWS hosts Expert Sierra. From the AI Legal Index, based on each vendor's own published materials as of November 3, 2026. No vendor pays for placement.
Neither vendor paid for inclusion, placement or a grade, and neither reviewed this page before it published. Everything on it comes from public material on the dates shown. How the index grades.
Neither vendor publishes a price figure or an accuracy rate for its AI. Aderant's training exclusions sit in an article rather than its contract terms, and its API terms grant a license to submitted content without a training exclusion. Intapp's master agreement is not public, so the liability position behind its AI Addendum, whose liability clause excludes inaccuracies and errors, is a point to settle in the order.