Intelligence Brief
Accounting, Tax, and Advisory Agentic AI Report
Audit, Tax, Advisory, Operating Model, and Market Signals
Overview
Executive Summary
Strategic Narrative
Across audit, tax, advisory, and firm operations, agentic AI has crossed from experimentation into regulated production. The dominant near-term value is real, but the dominant near-term risk is governance failure under existing professional standards. Firms that act this quarter to harden controls, constrain usage to reviewable workflows, and repackage AI into managed services will gain margin and scale advantages, while those that delay risk inspection findings, liability exposure, and erosion of trust.
#1
Agentic AI governance is now an inspection risk, not a future policy issue
Act Now
Intelligence Context
SAS, EY, and Microsoft updates show agent approval workflows, immutable logs, replayable audit trails, HITL gates, and continuous monitoring are becoming baseline expectations. PCAOB commentary and CPA-focused guidance emphasize that existing standards already require explainability, reproducibility, and documented supervision. Compliance analysts warn agentic AI is outpacing firm controls, creating immediate regulatory and professional liability exposure.
Recommended Action
Managing Partner and CIO to mandate a firmwide Agentic AI Control Baseline this quarter: (1) require all AI agents used in audit, tax, advisory, and ops to run through a centralized orchestration/control plane with immutable activity logs; (2) define agent risk tiers with required human approval checkpoints; (3) assign named partner accountability per agent class; and (4) block any autonomous agent usage outside this framework.
Business Impact
Reduces near-term PCAOB, peer review, IRS, and legal risk; preserves inspection defensibility; avoids retroactive remediation costs if controls are challenged.
AuditTaxAdvisoryFirm OperationsGovernance
#2
Audit AI must be constrained to reviewable workflows or it risks non-compliance today
Act Now
Intelligence Context
Audit trends show agentic AI scaling fastest in documentation drafting, evidence ingestion, and controls testing—areas with well-defined standards and high reviewability. Analysts and PCAOB signals warn that agentic AI lacking reproducibility or documentation already violates existing audit standards. Big Four usage embeds agents but retains human sign-off, while risks include weak workpapers and over-reliance.
Recommended Action
Audit Leader to formally restrict agentic AI usage to approved audit phases this quarter (documentation, PBC management, routine SOX testing), explicitly prohibit autonomous use in risk assessment and final conclusions, and update audit methodology to require agent-generated workpapers to include source linkage, logic trace, and reviewer sign-off evidence.
Business Impact
Improves documentation quality and reviewer leverage while protecting against inspection findings related to insufficient skepticism or undocumented methodology deviations.
AuditAssurance
#3
Indirect tax is the fastest monetizable agentic AI opportunity—if packaged as managed service
Plan This Quarter
Intelligence Context
Zamp’s $30M raise and reported ~99.97% accuracy, Avalara’s agent-executed compliance, and consolidation signals show agentic AI is mature in multi-jurisdiction indirect tax. Commercial trends indicate margins improve only when AI is bundled into fixed-fee or managed offerings, not hourly efficiency.
Recommended Action
Tax Leader and COO to launch or repackage indirect/sales tax compliance as an AI-enabled managed service this quarter with exception-only human review, fixed or credit-based pricing, and clear liability boundaries, rather than embedding AI savings into hourly work.
Business Impact
Creates immediate capacity without headcount growth, protects margins, and generates recurring revenue while meeting client expectations for AI-run compliance.
Tax ComplianceManaged Services
#4
Client Accounting Services economics are shifting—review capacity is the new bottleneck
Plan This Quarter
Intelligence Context
KPMG, Safebooks, and Basis signals show AI-orchestrated close and CAS delivery compressing cycles from days to hours and replacing manual labor with agent execution. Trends warn that underscaled human review and economic pressure to cut reviewers can weaken quality and trust.
Recommended Action
CAS Leader and COO to redesign CAS staffing this quarter by explicitly sizing reviewer-to-agent ratios, defining exception SLAs, and reallocating senior capacity to review and client-facing insight, not data prep.
Business Impact
Unlocks margin expansion and scalable growth in CAS while avoiding quality failures that could damage client retention and audit defensibility.
Client Accounting ServicesOutsourced Accounting
#5
Firm knowledge must become an enforced AI constraint or agent outputs will drift
Monitor
Intelligence Context
Knowledge trends show SOPs, methodologies, and memos evolving from passive reference into active constraints via permission-aware RAG. Risks include outdated SOPs propagating errors and confidentiality breaches if retrieval is misconfigured.
Recommended Action
CIO and Practice Leaders to prioritize a permission-aware RAG foundation this quarter that grounds all agents in versioned firm SOPs, standards, and approved interpretations, with ownership assigned for content currency and access controls.
Business Impact
Improves consistency, training speed, and defensibility of AI-assisted work while reducing rework and professional risk from drift or misapplication of guidance.
AuditTaxAdvisoryKnowledge Management
News
Latest Updates
SAS Introduces Governance Controls for Autonomous AI Agents
CIO
Governance, Risk, and Controls
GovernanceRisk containmentRegulatory readiness
SAS expanded its Viya platform with governance tooling designed for agentic AI, including approval workflows, audit trails, and continuous monitoring. For accounting firms, this directly addresses control, explainability, and oversight gaps that have slowed deployment of autonomous agents in audit and tax workflows.
Agent Approval and Monitoring Emerge as Baseline AI Controls
COO
Platforms, Tooling, and Architecture
GovernanceQualityOperational control
Industry coverage of the SAS release emphasized that agent approval workflows and continuous monitoring are becoming baseline expectations for enterprise AI. This signals that firms will be expected to demonstrate active oversight of autonomous systems, not just post-hoc review.
Compliance Analysts Warn Agentic AI Is Outpacing Firm Controls
Managing Partner
Risk, Compliance, and Forensics
GovernanceProfessional liabilityRegulatory alignment
Compliance-focused commentary highlighted that agentic AI systems are executing multi-step workflows with minimal human checkpoints. Firms adopting these tools without formal accountability and escalation controls risk breaching existing professional standards.
Professional Standards Seen as the Near-Term AI Constraint
Audit Leader
Audit and Assurance
QualityGovernanceInspection readiness
Analysts noted that the biggest risk is not future AI regulation but current audit and tax standards. Agentic AI that cannot demonstrate reproducibility and reviewability may already be non-compliant under existing rules.
Information Governance Identified as a Hidden Agentic AI Risk
CFO
Governance, Risk, and Controls
Risk managementClient trustGovernance
Legal and information-governance experts warned that autonomous agents can generate decisions and workpapers outside traditional engagement controls. This creates discovery, retention, and evidentiary risks for accounting firms.
Agent-Generated Workpapers Raise Inspection Defensibility Concerns
Audit Leader
Audit and Assurance
Inspection readinessQualityGovernance
Commentary emphasized that agentic AI may produce undocumented or weakly governed artifacts. For audit firms, this raises concerns about documentation sufficiency and defensibility during PCAOB or peer inspections.
CPA-Focused AI Governance Guidance Gains Momentum
Managing Partner
Audit and Assurance
GovernanceOperating model designRegulatory defensibility
New CPA-oriented guidance stressed that regulators already expect AI outputs to be explainable and reproducible. Firms waiting for explicit agentic AI rules may fall behind peers building compliant operating models now.
Sources
Agent-in-the-Loop Models Positioned as Near-Term Best Practice
Advisory Leader
Practice Management and Internal Operations
GovernanceClient deliveryRisk containment
Governance guidance highlighted agent-in-the-loop designs as a practical bridge between innovation and compliance. These models allow firms to scale AI assistance while retaining human accountability.
Sources
Lack of Accounting-Specific Agentic AI Launches Signals Caution
COO
Commercial Impact and Adoption Signals
Governance over speedDeployment pacingMargin protection
Despite heightened attention on agentic AI, no major accounting- or tax-specific product launches occurred this week. This suggests firms and vendors are prioritizing governance and inspection readiness over rapid deployment.
Audit
Audit and Assurance
#1
Big Four audit firms (EY, Deloitte, PwC, KPMG)
Workpaper Drafting and Documentation
Firmwide RolloutAI-generated workpapers with automated cross-referencing, evidence linkage logs, and reviewer gap flags.
What Changed
Expanded live usage of agentic AI embedded within core audit platforms, moving from controlled pilots to broader engagement-level deployment without new product launches.
Workflow Shift
Agents now draft and cross-reference workpapers, flag documentation gaps, and prepare reviewer-ready files, while auditors retain final sign-off and judgment.
Quality Implication
Improved documentation consistency and reviewer leverage, reducing rework and late-stage review issues.
Key Risk: Over-reliance on AI-drafted documentation without sufficient partner and manager scrutiny could weaken professional skepticism.
#2
Audit-specific AI vendors (e.g., Workpapr, Fieldguide ecosystem)
Evidence Collection and PBC Management
Limited RolloutSystem-generated evidence completeness reports and assertion-mapped evidence repositories.
What Changed
Late-April expansion of early-access programs validating AI-driven full-population evidence ingestion and assertion mapping, particularly for SOX engagements.
Workflow Shift
Evidence requests, ingestion, and mapping are increasingly automated, shifting auditors toward exception handling and evaluation rather than manual collection.
Quality Implication
Higher assurance over completeness and traceability of audit evidence, supporting more defensible audit files.
Key Risk: Data integrity and access controls over client-provided evidence streams remain critical to prevent audit file contamination.
#3
SOX and controls-testing AI platforms
Controls Testing and SOX Support
Commercial ProductAutomated control test logs, exception summaries, and AI-generated audit trails suitable for SOX 302/404 sign-off.
What Changed
Scaling use of AI to execute consistency testing across large control populations, highlighted in late-April practitioner and vendor publications.
Workflow Shift
Routine control testing is increasingly AI-executed, with auditors focusing on outliers, remediation evaluation, and management discussions.
Quality Implication
Greater coverage and consistency in controls testing, improving confidence in SOX conclusions.
Key Risk: Misalignment between AI test logic and firm methodology could create undocumented deviations from approved testing approaches.
#4
PCAOB
Quality Review and Engagement Supervision
ResearchStandardized audit documentation structures and AI governance policies referenced in inspection preparation.
What Changed
Renewed circulation and citation of prior PCAOB guidance emphasizing standardized documentation, technology-assisted analysis, and AI governance within audit firms.
Workflow Shift
Firms are reinforcing internal review protocols and AI governance frameworks to align agentic AI usage with regulator expectations.
Quality Implication
Clearer regulatory expectations reduce inspection risk and drive more disciplined AI deployment.
Key Risk: Inconsistent interpretation of regulatory signaling could lead to uneven application across firms and engagements.
#5
Audit analytics and continuous assurance platforms
Continuous Assurance and Monitoring
Limited RolloutControl health dashboards, anomaly alerts, and streaming analytics reports.
What Changed
Late-April acceleration of practitioner adoption and marketing around near-continuous analytics and anomaly detection, without accompanying standards changes.
Workflow Shift
Continuous monitoring outputs are increasingly used to inform audit planning and interim procedures, supplementing periodic audit work.
Quality Implication
Earlier risk identification enhances audit responsiveness, though outputs remain supplemental to statutory audit evidence.
Key Risk: Expectation gaps may arise if stakeholders overinterpret continuous monitoring as providing full audit assurance.
Trend Insight
Agentic AI is becoming operational primarily in documentation, evidence handling, and repetitive controls testing, where standards are well-defined and outputs are highly reviewable. Audit planning, risk assessment, and final conclusions still require heavy human review due to judgment, skepticism, and regulatory expectations. The most important structural shift this period is the transition from isolated pilots to embedded, workflow-native AI judged on defensibility and inspection readiness rather than novelty.
Tax
Tax Compliance
#1
Zamp
Indirect Tax and Sales Tax Workflows; Multi-Jurisdiction Filing Orchestration
U.S. multi-jurisdiction indirect and sales tax filings.Commercial Product
What Changed
Raised $30M to scale an agent-based, end-to-end sales tax compliance platform covering 12,000+ U.S. jurisdictions, with AI agents executing filings and offering a penalty guarantee.
Productivity Impact
Material reduction in cycle time (automated end-to-end filings), high review leverage via exception-only human review, and step-change improvement in filing accuracy (reported ~99.97%).
Review and Control Model
Human-in-the-loop with agent-led execution; humans intervene on exceptions and escalations only.
Key Risk: Regulatory scrutiny and scalability of liability guarantees as volumes and jurisdictional complexity increase.
#2
Avalara (existing platform signal)
Multi-Jurisdiction Filing Orchestration; Indirect Tax and Sales Tax Workflows
U.S. and selected international indirect tax filings.Firmwide Rollout
What Changed
No new launch, but continued market consolidation around agent-assisted execution layers embedded in indirect tax platforms, signaling maturation rather than experimentation.
Productivity Impact
Incremental cycle time reduction through orchestration and automation; moderate review leverage as humans still approve filings.
Review and Control Model
Copilot-to-agent hybrid with mandatory human approval prior to filing.
Key Risk: Fragmented data inputs and reliance on rules engines limit full autonomy.
#3
HubSync / CPA firm ecosystem
Data Intake and Normalization; Reconciliation and Workpaper Assembly
U.S. federal and state returns across individual and entity filings.Limited Rollout
What Changed
Accelerated adoption of AI-driven document ingestion and workpaper assembly tools during busy season, though without new product launches.
Productivity Impact
High cycle time reduction in prep phases; significant leverage for reviewers by standardizing workpapers.
Review and Control Model
Human review remains mandatory; AI prepares and reconciles source data.
Key Risk: Data quality and source-document variability can propagate errors upstream.
#4
EY / IBM
Multi-Jurisdiction Filing Orchestration; Provision Support and Close
Global corporate tax compliance and provision processes.Firmwide Rollout
What Changed
No new capabilities announced, but ongoing client deployments of the global tax AI platform indicate scaling of agent-assisted workflows in large enterprises.
Productivity Impact
Moderate cycle time compression in global compliance; review leverage via standardized controls.
Review and Control Model
Strong human governance with AI-assisted execution and controls.
Key Risk: Complex change management and dependency on client data harmonization.
#5
CPA Pilot / Practice platforms
Notice Handling and Resolution
IRS and state tax notices.Limited Rollout
What Changed
Continued embedding of AI-driven notice response automation within practice platforms, without standalone agentic launches.
Productivity Impact
Localized cycle time reduction for notices; limited impact on overall filing accuracy.
Review and Control Model
AI drafts responses; humans review and submit.
Key Risk: Risk of misinterpretation of notices without sufficient human oversight.
Trend Insight
The strongest agentic adoption is in indirect and sales tax compliance, especially multi-jurisdiction filing orchestration where rules are structured and volumes are high. Corporate income tax preparation, complex provisions such as ASC 740, and notice resolution still require substantial human review because ambiguity and materiality remain judgment-heavy. The structural shift is the emergence of liability-bearing, operator-style AI in tax compliance, signaling a move from copilots to accountable agents that execute filings end to end.
Tax Advisory
Tax Research and Advisory
#1
Tax Authorities (OECD-aligned jurisdictions)
International Tax and Transfer Pricing
HighFirmwide Rollout
What Changed
Tax authorities are actively deploying AI systems to algorithmically scan, parse, and flag transfer pricing risks, shifting audits toward data-driven, consistency-based enforcement rather than narrative-heavy reviews.
Technical Significance
This forces advisors to design transfer pricing positions that are explainable to machines, with consistent data, traceable assumptions, and audit-ready logic aligned to Tax Administration 3.0.
Client Delivery Impact
Clients now expect continuous TP risk monitoring, pre-audit simulations, and AI-aligned documentation rather than annual compliance-only studies.
Key Risk: Advisory positions that are technically correct but poorly structured or inconsistent across datasets may be algorithmically flagged, increasing audit exposure.
#2
Baker Tilly + Basis
Planning and Scenario Modeling
HighLimited Rollout
What Changed
Baker Tilly announced a collaboration with Basis to scale AI-enabled accounting, tax, and finance managed services focused on decision-ready insights rather than task automation.
Technical Significance
Represents a shift from point AI tools to an agentic advisory operating model that continuously analyzes client data and supports forward-looking tax decisions.
Client Delivery Impact
Improves responsiveness through near-real-time insights and reduces turnaround time for planning discussions and scenario analyses.
Key Risk: Reliance on continuous AI outputs increases the need for strong governance to prevent drift or misinterpretation of tax positions.
#3
KPMG + Google Cloud Gemini Enterprise
Controversy and Notice Response
HighLimited Rollout
What Changed
KPMG expanded its alliance with Google Cloud to deploy agent-style AI capable of reasoning across regulations, client data, and internal controls in regulated functions, including tax.
Technical Significance
Demonstrates Big Four confidence in semi-autonomous AI agents handling complex regulatory reasoning under human oversight.
Client Delivery Impact
Enhances speed and consistency in controversy analysis, position validation, and response strategies while maintaining defensibility.
Key Risk: Model governance and explainability failures could undermine credibility in high-stakes controversy matters.
#4
Tax Research Platforms (e.g., Thomson Reuters, Blue J, CoCounsel Tax)
Technical Tax Research
Medium-HighCommercial Product
What Changed
AI tax research tools are now evaluated primarily on defensible citations, explainable reasoning, and audit-ready outputs, converging toward agent-style research-to-memo workflows.
Technical Significance
Marks a maturation from generative chat to structured legal reasoning systems that mirror traditional tax research methodology.
Client Delivery Impact
Improves quality and defensibility of advice while reducing time spent assembling first-draft research and memos.
Key Risk: Overreliance on AI-generated reasoning without senior technical review may lead to subtle authority misapplication.
#5
Advisory Firms (Internal Agentic Systems)
Memo Drafting and Client Delivery
MediumPilot
What Changed
Agentic AI is increasingly used to orchestrate multi-step memo drafting: pulling facts, mapping authorities, drafting structured analyses, and flagging controversy risks.
Technical Significance
Transforms memo preparation from manual assembly to supervised technical reasoning pipelines, increasing leverage of senior reviewers.
Client Delivery Impact
Faster delivery of higher-quality, more consistent advisory memos across jurisdictions and service lines.
Key Risk: Inconsistent factual inputs or poor document management can cascade errors across the entire agentic workflow.
Trend Insight
Agentic AI is moving beyond drafting support into real technical reasoning, particularly in transfer pricing, tax research, and controversy readiness, where explainability and defensibility are now core design requirements. The most significant shift is the growing alignment pressure between advisory AI and tax authority AI, forcing firms to redesign tax advice, documentation, and operating models so they are interpretable and defensible to both human reviewers and enforcement algorithms.
Client Accounting
Client Accounting and Close
#1
KPMG – Ignite Financial Close Companion
Month-End Close Orchestration
Implied multi-day close compression; positioned to move close cycles from 5–7 days toward 1–3 days through automated exception surfacing.Commercial ProductWorkday Financial ManagementGoogle Cloud Gemini EnterpriseEnterprise GL and subledger data
What Changed
KPMG launched an AI close companion embedded directly into Workday, shifting from assistive analytics to AI-guided close execution with exception detection and workflow orchestration.
Control Impact
High positive impact via standardized close steps, anomaly detection, and guided approvals; strengthens consistency and audit defensibility.
Key Risk: Over-reliance on vendor-specific orchestration logic and black-box AI decisions could challenge explainability during audits.
#2
Forbes Finance Council (industry-wide architecture signal)
Finance Operating Rhythm Automation
Not quantified; architectural shift enabling systemic cycle-time reduction across close and reporting.ResearchERP platformsClose management toolsFP&A systemsWorkflow orchestration layers
What Changed
Formal recognition that multi-agent orchestration layers—not single copilots—are becoming the dominant architecture for finance, treating close, reconciliation, FP&A, and controls as one continuous system.
Control Impact
Moderate to high, as orchestration enables centralized exception escalation and policy-driven workflows.
Key Risk: Conceptual adoption may outpace operational readiness, leading to fragmented implementations without true orchestration.
#3
Safebooks
Reconciliation and Exception Resolution
60–80% of reconciliations completed before day 0; week-long closes compressed to hours in advanced cases.Commercial ProductGeneral LedgerBank feedsSubledgersClose management tools
What Changed
Reframed the month-end close as a coordinated system of autonomous agents (recs, journals, variance analysis) managed by a conductor agent, emphasizing pre-close reconciliation clearance.
Control Impact
Positive if well-governed; early reconciliation reduces late-period pressure but increases need for strong agent oversight.
Key Risk: Front-loading reconciliations may surface data quality issues earlier without sufficient human review capacity.
#4
Basis
Bookkeeping and Coding
Implicit replacement of large portions of manual and offshore labor; significant capacity gains rather than explicit close-day metrics.Limited RolloutSMB ERPs (e.g., QuickBooks, Xero)Bank feedsClose and reporting layers
What Changed
Late-April demos and firm commentary emphasized agents executing end-to-end CAS delivery (bookkeeping through close) with humans positioned primarily as reviewers.
Control Impact
Mixed; improves consistency through automation but introduces review bottlenecks if human oversight is underscaled.
Key Risk: Economic pressure to reduce reviewer staffing could weaken quality and client trust if exceptions scale faster than review capacity.
Sources
#5
Mindra (and similar FP&A orchestration vendors)
Reporting Package Preparation
Eliminates post-close manual variance analysis cycles; reporting insights generated continuously rather than monthly.Commercial ProductERPData warehousesFP&A and reporting tools
What Changed
Agent orchestration narratives now explicitly link close outputs to always-on FP&A agents, dissolving the traditional boundary between close and analysis.
Control Impact
Moderate; continuous analytics improve issue detection but rely on the integrity of upstream close agents.
Key Risk: Blurring close and FP&A may complicate sign-off boundaries and formal period controls.
Trend Insight
Agentic AI is materially changing outsourced accounting economics by shifting CAS from labor-arbitrage models to AI-executed delivery with human review. That compresses cost per client, increases gross margin potential, and lets firms scale without proportional headcount growth. The structural shift is the normalization of AI-orchestrated close systems, where agents run the workflow and humans manage exceptions, moving agentic AI from an experimental efficiency tool to core CAS delivery infrastructure.
Deals
Deal, Diligence, and Valuation
#1
Shepi.ai
Quality of Earnings Support
Lower-middle-market PE or search fund uses AI QoE to validate EBITDA and cash conversion pre-LOI or between signing and close without full Big Four engagement.Commercial ProductN/A
What Changed
Repositioned QoE as a self-serve, lender-ready workflow producing traceable EBITDA normalization, proof-of-cash, and working-capital bridges in hours rather than weeks, explicitly targeting sub-$50m PE and search-fund deals.
Diligence or Valuation Impact
Material compression of QoE timelines with outputs structured as opinion-grade schedules suitable for IC and lender review, not just analyst workpapers.
Key Risk: Risk of over-reliance on AI-generated conclusions without sufficient judgment around non-recurring items or aggressive normalization assumptions.
#2
SmartRoom AI / DealRoom Diligence AI
Diligence Document Review
Buy-side diligence teams rapidly identify change-of-control, indemnity, and pricing risks without manual document review.Commercial ProductPublic product messaging updates
What Changed
Shifted positioning from static summarization to conversational interrogation of data rooms, enabling deal teams to query contracts, clauses, and pricing terms directly.
Diligence or Valuation Impact
Significant speed gains in issue identification and Q&A formulation; marginal analytical depth increase as document AI becomes table stakes.
Key Risk: False confidence if AI misses nuanced legal language or poorly scanned source documents.
#3
DiligenceSquared
Buyer and Seller Advisory Workflow Support
PE sponsors augment commercial diligence with rapid, low-cost customer interviews to validate revenue durability and go-to-market claims.Limited RolloutMedia profile of new capability
What Changed
Introduced AI voice agents to conduct primary-research interviews with customers and partners during PE diligence, extending AI beyond documents into field diligence.
Diligence or Valuation Impact
Step-change in analytical depth by scaling customer and market validation that was previously time- and cost-prohibitive.
Key Risk: Interview quality and bias risk if AI agents fail to probe effectively or misinterpret qualitative responses.
#4
Big Four Valuation Practices (PwC, EY, KPMG, Deloitte)
Valuation Research and Benchmarking
Deal teams receive faster comps, scenarios, and sensitivity analyses while maintaining audit- and litigation-defensible valuation conclusions.Firmwide RolloutIndustry commentary and conference reporting
What Changed
Reaffirmed AI usage as an internal 'junior analyst' for data normalization, guideline company screening, and scenario scaffolding, while reserving valuation judgment for human signatories.
Diligence or Valuation Impact
Improved speed and consistency of valuation support work, but governance constraints cap full automation of valuation opinions.
Key Risk: Two-tier market risk where smaller firms may over-automate without equivalent governance, increasing challenge risk.
#5
Advisory and PMI AI Platforms (e.g., AcuityAI)
Carve-Out and Integration Tracking
Corporate acquirers and PE portfolio teams use agentic AI to coordinate finance, HR, and IT integration plans immediately after close.PilotThought leadership and early platform demonstrations
What Changed
Language and tooling shifted from AI analytics to agentic PMI orchestration, with AI sequencing integration tasks, flagging conflicts, and triggering actions across functions.
Diligence or Valuation Impact
High deal-team leverage post-close by moving from analysis to partial execution, reducing integration leakage and accelerating value capture.
Key Risk: Governance and change-management risk if agentic systems trigger actions without clear human oversight.
Trend Insight
Agentic AI is beginning to change advisory economics by compressing time to insight and expanding the feasible scope of diligence and integration work without proportional headcount increases. While pricing power remains with branded advisors, AI is eroding labor-based moats in lower-middle-market deals and shifting value toward judgment, governance, and integration execution. The key structural shift is the move from AI-assisted analysis to AI-structured conclusions and actions, signaling that AI is no longer just accelerating work but starting to redefine where human sign-off truly adds value.
Risk
Risk, Compliance, and Forensics
#1
Microsoft (Purview + Copilot)
Regulatory Evidence Management
Financial Services, Healthcare, Public SectorCommercial ProductAutomated metadata tagging, time-stamped retention policies, and cryptographic integrity checks embedded in Purview audit logs.
What Changed
Introduction of agentic AI workflows that autonomously classify, retain, and produce regulatory evidence across M365 environments with immutable audit logs.
Control or Investigation Impact
High control impact through continuous evidence capture and reduced manual handling risk; improves defensibility during regulatory inquiries.
Key Risk: Over-reliance on automated classification leading to mis-scoped regulatory production.
#2
Relativity
Fraud Detection and Investigation
Legal, Financial ServicesCommercial ProductModel decisions logged with reviewer override tracking and defensible AI explainability reports.
What Changed
Deployment of agentic review agents that sequence document review, anomaly clustering, and investigator prompts without human task orchestration.
Control or Investigation Impact
Significant investigation speed gains by reducing first-pass review time and prioritizing high-risk evidence sets.
Key Risk: Bias or model drift affecting prioritization of evidentiary material.
#3
ServiceNow
Controls Monitoring and Testing
Financial Services, EnergyFirmwide RolloutSystem-of-record logging for control tests, agent actions, and remediation artifacts.
What Changed
Agentic AI introduced to continuously test controls, open issues, assign remediation tasks, and validate closure evidence.
Control or Investigation Impact
High control impact via near-real-time controls assurance and reduced audit cycle time.
Key Risk: Control owners deferring judgment to agents without sufficient challenge.
#4
Palantir
Case Management and Escalation
Financial Crime, National SecurityLimited RolloutProvenance tracking across integrated data sources with replayable decision graphs.
What Changed
Agentic orchestration of multi-source risk signals to automatically escalate cases, recommend investigative steps, and simulate outcomes.
Control or Investigation Impact
Improves investigation defensibility by standardizing escalation logic and documenting decision pathways.
Key Risk: Opacity of complex decision graphs to regulators.
#5
OpenText
Policy Compliance Monitoring
Life Sciences, Financial ServicesCommercial ProductTamper-evident content repositories with versioned policy mappings.
What Changed
Agentic AI monitors communications and transactions against policy, autonomously flags breaches, and drafts compliance reports.
Control or Investigation Impact
Moderate to high impact by expanding coverage and consistency of compliance monitoring.
Key Risk: False positives increasing compliance noise and investigation burden.
Trend Insight
Firms are increasingly willing to rely on agentic AI for upstream, repeatable, and high-volume regulated workflows such as evidence capture, control testing, and first-line investigation triage, while retaining human judgment for conclusions and regulatory attestations. The most important structural shift this period is the movement from AI as a decision-support tool to AI as an autonomous workflow actor with auditable action logs, forcing organizations to redesign governance, evidence chains, and model risk management to preserve defensibility.
Knowledge
Knowledge, Research, and Document Intelligence
#1
Microsoft
Document Extraction and Classification
Audit teams, accounting advisory, internal delivery centersMulti‑document reasoning grounded in enterprise document stores with validation against firm policies and standards.Commercial Product
What Changed
Azure Document Intelligence evolved from standalone extraction into agent‑orchestrated, multi‑document workflows that reason across related documents (e.g., contracts, invoices, amendments) and trigger next actions.
Document or Knowledge Impact
Transforms document intelligence from passive OCR into end‑to‑end workflow execution, improving audit prep, revenue recognition analysis, and complex accounting reviews.
Key Risk: Complex orchestration increases dependency on accurate document linking; errors in document relationships could cascade across workflows.
#2
Thomson Reuters
Memo and Deliverable Drafting
Technical accounting, tax, advisory professionalsRAG over authoritative sources (standards, regulations) combined with firm memos and policy repositories.Commercial Product
What Changed
Agentic AI capabilities now draft accounting and advisory memos using firm policy, authoritative guidance, and structured reasoning with explicit human review checkpoints.
Document or Knowledge Impact
Significantly accelerates memo production while preserving defensibility through citations and firm‑approved interpretations.
Key Risk: Over‑reliance on AI‑drafted memos without sufficient professional skepticism during review.
#3
Accounting‑focused AI vendors (various)
Methodology and SOP Navigation
Core accounting staff, shared service centers, new hiresStructured SOP repositories retrieved contextually and cited during task execution.Limited Rollout
What Changed
SOPs are now treated as structured, queryable memory that agents actively use as constraints during live work rather than static reference documents.
Document or Knowledge Impact
Improves consistency, training speed, and compliance in recurring processes such as close, onboarding, and tax preparation.
Key Risk: Outdated SOPs can propagate incorrect practices if governance and versioning are weak.
#4
Professional‑services AI platforms (e.g., Google Workspace Intelligence)
Proposal and RFP Support
Advisory partners, pursuit teams, business developmentKnowledge‑graph‑backed retrieval from proprietary proposals and engagement artifacts.Limited Rollout
What Changed
Proposal drafting agents now leverage firm‑specific knowledge graphs of prior proposals, engagement letters, and win themes to assemble proposals consistent with firm precedent.
Document or Knowledge Impact
Enhances win rates and reduces proposal cycle time while reinforcing firm voice and risk controls.
Key Risk: Potential leakage of sensitive pricing or strategy if access controls are misconfigured.
#5
Industry‑specific RAG providers (e.g., zTabs)
Internal Knowledge Grounding and Memory
Tax, audit, and advisory professionalsMulti‑source RAG combining standards, firm interpretations, and engagement history with strict citation enforcement.Commercial Product
What Changed
RAG knowledge bases for accounting firms now emphasize multi‑source grounding, jurisdiction awareness, versioned standards, and zero‑tolerance citation requirements.
Document or Knowledge Impact
Raises trust in AI‑assisted research and advice by ensuring answers are traceable to authoritative and firm‑approved sources.
Key Risk: High implementation effort and ongoing curation cost to maintain accurate, versioned knowledge bases.
Trend Insight
Firms are clearly moving from chat‑based AI interactions toward embedded, workflow‑executing agents integrated directly into audit, tax, advisory, and proposal processes. The most important structural shift in this period is the elevation of firm knowledge—SOPs, policies, memos, and prior work—from passive reference material into active constraints and memory that govern how AI agents plan, execute, and escalate work, with human‑in‑the‑loop controls designed explicitly for auditability and regulatory scrutiny.
Operations
Practice Management and Internal Operations
#1
Hapax
Utilization and Realization Monitoring
COO, practice leaders, engagement partners, resource managersShifts firms from retrospective management to closed-loop operational control, where agents act as always-on operations managers rather than reporting tools.Commercial Product
What Changed
Expanded from KPI dashboards to continuously running agentic monitors that observe utilization, realization, billing risk, and delivery signals and autonomously trigger staffing, billing, or scope interventions before KPIs break.
Utilization or Margin Impact
High margin impact via earlier course correction on underutilization, write-offs, and scope creep; materially improves realization discipline and reduces revenue leakage.
Key Risk: Over-automation of staffing or billing interventions without sufficient partner override could create trust or governance issues.
#2
PracticePro 365
Engagement Management
Engagement managers, partners, PMO, client service teamsMoves engagement management from checklist-based workflows to adaptive, agent-managed orchestration across the full engagement lifecycle.Commercial Product
What Changed
Introduced AI-driven engagement lifecycle orchestration that replaces static workflow rules with agents that manage follow-ups, stage progression, and pipeline forecasting across CRM-to-billing.
Utilization or Margin Impact
Moderate-to-high margin impact through reduced engagement slippage, faster cycle times, and improved forecast accuracy feeding staffing and billing decisions.
Key Risk: Forecast quality is highly dependent on data hygiene across CRM, time, and billing systems.
#3
Multiple vendors (e.g., Singoa)
Billing, Collections, and WIP
Finance leaders, billing teams, partnersReframes finance operations from back-office processing to proactive, agent-managed cash flow supervision.Commercial Product
What Changed
Acceleration of agentic AR and WIP agents that supervise the full WIP-to-invoice-to-collection loop, predicting delays, preparing invoices, and triggering collections based on client behavior patterns.
Utilization or Margin Impact
Very high margin and cash-flow impact through faster billing cycles, reduced aged WIP, and improved realization and DSO.
Key Risk: Client relationship risk if automated collection actions are poorly calibrated to client context.
#4
AgenticPractice (apurv.ai)
Operating Model Redesign
Firm leadership, operations, partnersIntroduces the concept of agents as durable organizational actors embedded in firm operations, not task-level bots.Limited Rollout
What Changed
Repositioned platform around persistent AI coworkers managing scheduling, documentation, coordination, and revenue operations as a unified agentic operating layer.
Utilization or Margin Impact
Indirect but significant impact through reduced partner and manager administrative load, increasing leverage and capacity for higher-value work.
Key Risk: Change management complexity and unclear ownership between human roles and agent responsibilities.
#5
PromptPartner (and similar onboarding platforms)
Recruiting and Onboarding
HR, operations, new hires, team leadsTransforms onboarding from a one-time HR process into an agent-supported, continuous enablement function.Commercial Product
What Changed
Refreshed AI-native onboarding positioning with agents handling document completeness, follow-ups, verification, and embedded AI coworkers to accelerate staff ramp-up.
Utilization or Margin Impact
Moderate margin impact by shortening time-to-productivity for new hires and reducing non-billable onboarding overhead.
Key Risk: Limited differentiation over time as onboarding automation commoditizes.
Trend Insight
Firms are applying agentic AI first to internal operations rather than client delivery, prioritizing utilization, billing, engagement control, and operating discipline before external-facing advisory use cases. The most important structural shift is the move from workflow automation and reporting toward autonomous, goal-directed agents that intervene in real time, effectively creating an agentic operating system for professional services firms.
Governance
Governance, Risk, and Controls
#1
EY
Auditability and Traceability
Named engagement partner retains accountability, supported by centralized AI governance and assurance teams responsible for log integrity and access.Operational Standard
What Changed
Agent-level activity logging and replayable audit trails are now being operationalized across assurance workflows following EY’s April 7 agentic AI launch, becoming the de facto control baseline during the last 14 days.
Control Implication
All agent actions (prompts, tool calls, data access, outputs) must be immutably logged and replayable, with clear separation between AI system logs and formal audit workpapers.
Risk Exposure
Regulatory inspection failure, inability to evidence audit judgments, and enforcement risk if agent decisions cannot be reconstructed.
Key Risk: Insufficient log fidelity or tamper resistance undermining regulator confidence.
#2
Microsoft (enterprise agent control plane)
Human-in-the-Loop Review Design
Task-level human owners are assigned per agent class, with escalation paths to engagement leadership for judgment-bearing outputs.Operational Standard
What Changed
Shift from binary human-vs-autonomous controls to tiered HITL models with pre-execution approval for high-risk actions and exception-only review for low-risk tasks.
Control Implication
Agent classes must be risk-rated, with explicit approval thresholds and named human approvers mapped to task criticality.
Risk Exposure
Unauthorized or inappropriate autonomous decisions in tax positions or audit conclusions.
Key Risk: Misclassification of task risk leading to under-supervision.
#3
KPMG
Client Confidentiality and Data Access Controls
Engagement partners own confidentiality risk, with firmwide data governance functions enforcing technical isolation.Operational Standard
What Changed
Active enforcement of client-level memory isolation, blocking of model retraining on client data, and permissions aligned to professional standards rather than IT roles.
Control Implication
Agent architectures must hard-separate engagement data contexts and prevent cross-client learning or inference.
Risk Exposure
Breach of independence, confidentiality violations, and professional misconduct findings.
Key Risk: Latent data leakage through shared embeddings or memory layers.
#4
Moody’s (MRM guidance adopted by professional services)
Model Risk Management
Central model risk teams own validation and monitoring, with third-party model risk explicitly assigned.Operational Standard
What Changed
MRM has shifted from point-in-time validation to continuous behavioral monitoring, including version-locking of approved models and defined fallback models for audit and tax agents.
Control Implication
Agents must operate only on approved model versions with ongoing performance, drift, and behavior monitoring.
Risk Exposure
Model drift leading to incorrect judgments or inconsistent audit outcomes.
Key Risk: Unapproved model updates propagating into regulated workflows.
#5
IMDA / NIST (referenced by accounting firms)
Policy and Governance Frameworks
Board-level and executive AI governance committees oversee alignment and risk acceptance.Policy
What Changed
No new frameworks launched, but active adoption of IMDA’s Model AI Governance Framework for Agentic AI and NIST AI RMF to bound agent autonomy and map controls to regulator expectations.
Control Implication
Agent deployments must be explicitly mapped to recognized risk taxonomies and governance principles for inspection readiness.
Risk Exposure
Regulatory misalignment and inconsistent control interpretation across engagements.
Key Risk: Framework adoption remaining theoretical without technical enforcement.
Trend Insight
This period shows a structural shift from announcing agentic AI capabilities to hardening regulator-ready evidence. Firms are operationalizing governance by embedding controls directly into agent architectures—logs, permissions, HITL gates, and continuous model monitoring—rather than relying on high-level policy. The most important shift is treating agent behavior itself as a regulated risk surface, equivalent to traditional audit judgments, requiring continuous oversight, named human accountability, and replayable evidence.
Platforms
Platforms, Tooling, and Architecture
#1
Multiple (enterprise orchestration platforms emerging across vendors)
Multi-Agent Orchestration / Control Plane
Buy orchestration platforms for governance and scale; build custom agents on top where differentiation is required.Production PatternLLM providers (OpenAI, Azure OpenAI, Anthropic)Workflow engines (Power Automate, ServiceNow)Identity and access management (Entra ID, Okta)RAG and document systems
What Changed
In the last two weeks, orchestration-first architectures have become the dominant enterprise pattern, positioning orchestration as the central control plane for coordinating multiple AI agents with built-in governance, logging, human-in-the-loop, and policy enforcement.
Architecture Implication
Accounting and advisory firms now treat orchestration platforms as core infrastructure (akin to ERP or workflow engines), enabling agent-to-agent (A2A) communication, Model Context Protocol (MCP) support, and centralized risk management across audit, tax, and advisory agents.
Key Risk: Over-centralization may slow innovation if orchestration layers become overly rigid or vendor-locked.
#2
EY
Audit Execution Platform with Embedded Agents
Buy commercial audit platforms for execution; build firm-specific methodology agents for differentiation.Production PatternMicrosoft data platformsAudit workpaper systemsIdentity and governance toolingEnterprise orchestration layers
What Changed
EY announced global deployment of enterprise-scale agentic AI embedded directly into audit execution, with agents performing testing, sampling, and evidence validation under human review gates.
Architecture Implication
Audit architectures are shifting from AI-assisted tools to AI-executed workflows, requiring deep integration between agent systems, audit methodology, data platforms, and governance controls.
Key Risk: Regulatory and inspection risk if agent decisions are not fully explainable or auditable.
#3
Avalara
Tax Compliance Agent Platform
Strong buy signal for commodity tax execution; limited build opportunities in advisory overlays.Commercial ProductERP and transaction systemsIndirect tax enginesRegulatory content feedsFirm workflow and review systems
What Changed
Avalara advanced from AI-assisted to AI-executed tax compliance, introducing autonomous agents that continuously monitor transactions and regulatory changes and execute filings end-to-end.
Architecture Implication
Tax architectures now support continuous compliance models, where agents operate persistently and humans focus on exception handling and advisory interpretation.
Key Risk: Operational dependency on vendor accuracy and regulatory update latency.
#4
Enterprise RAG tool providers (various)
Retrieval and Knowledge Grounding (Permission-Aware RAG)
Buy scalable RAG platforms; configure rather than custom-build retrieval engines.Production PatternDocument management systemsPractice management systemsIdentity and access controlAgent orchestration platforms
What Changed
Permission-aware RAG became mandatory, with new capabilities for access-controlled retrieval, massive-scale indexing, and positioning RAG as a shared knowledge fabric for agent systems.
Architecture Implication
RAG is no longer a chat add-on but a foundational grounding layer, enforcing client confidentiality, audit separation, and role-based access across all agents.
Key Risk: Misconfigured permissions can lead to data leakage or audit independence breaches.
#5
Industry-wide (architecture convergence across firms)
Multi-Agent Architecture Pattern
Build custom agents within standardized patterns; buy the underlying platforms.Design PatternCentral orchestration hubsSpecialized task agentsEnterprise RAG systemsWorkflow and review tools
What Changed
Accounting and advisory implementations converged on hub-and-spoke orchestration with specialized agents and RAG as the default grounding layer, deprioritizing peer-only agent meshes and fine-tuning-heavy approaches.
Architecture Implication
Standardized patterns improve explainability, auditability, and scalability, reinforcing orchestration-led designs over experimental autonomy.
Key Risk: Risk of homogenization if firms fail to layer proprietary methodology and insight.
Trend Insight
Firms are standardizing around centralized multi-agent orchestration, permission-aware RAG, AI-executed audit and tax workflows with human review gates, and hybrid build/buy models. They are still experimenting with firm-specific advisory agents, advanced agent-to-agent collaboration, and continuous compliance extensions beyond tax. The decisive structural shift is from selecting individual AI agents to owning the orchestration and data layers that control them, establishing orchestration platforms as the strategic system of record for agentic AI in accounting, tax, and advisory firms.
Commercial
Commercial Impact and Adoption Signals
#1
Accounting & Tax Firms (SMB-focused)
Revenue Growth from New Services
Revenue expansion through new managed AI services; margin protection by avoiding billable-hour erosion and reallocating capacity to advisory.SMB clients are buying firms to run agentic AI on their behalf, not just to advise or implement tools.Revenue Generating
What Changed
ROI measurement has shifted from internal efficiency (hours saved) to operating-model ROI, with firms packaging agentic AI as managed, outcome-oriented services for SMB clients.
Adoption Barrier
Partner hesitation around cannibalizing traditional compliance revenue.
Key Risk: Firms that fail to redesign service models risk productivity gains translating into revenue loss.
#2
Sage + AWS
Cycle-Time and Turnaround Improvement
Faster deployment improves realization rates and supports scalable fixed-fee offerings.SMBs expect AI-enabled close, reporting, and finance operations as default platform functionality.Scaled Offering
What Changed
Agentic AI embedded directly into core SMB finance workflows, lowering integration cost and accelerating time-to-value for firms serving these clients.
Adoption Barrier
Dependence on vendor ecosystems may limit customization for complex client needs.
Key Risk: Platform commoditization could pressure differentiation for firms relying solely on embedded tools.
#3
Professional Services Firms (Cross-industry)
Margin Expansion and Delivery Leverage
Margin stabilization and expansion through recurring managed-service revenue models.Clients accept recurring fees tied to continuity, assurance, and outcomes rather than hours or tools.Market Commentary
What Changed
Clear evidence that margins improve only when agentic AI is bundled into fixed-fee or managed offerings rather than sold as efficiency within hourly billing.
Adoption Barrier
Legacy pricing models and partner compensation structures.
Key Risk: Firms slow to change pricing may see margin compression despite higher productivity.
#4
Filed
Pricing and Packaging Changes
Improved margin predictability and ability to monetize AI efficiency without discounting.Firms are actively testing alternative pricing structures aligned to AI-augmented delivery.Revenue Generating
What Changed
Emergence of credit-based and per-return AI pricing models that scale with volume while preserving firm margins.
Adoption Barrier
Client education required to shift expectations away from hourly billing.
Key Risk: Mispriced credits could erode margins if AI usage outpaces assumptions.
#5
KPMG (Managed Services Perspective)
Managed Service Model Expansion
Creation of durable, recurring revenue streams with higher lifetime client value.Enterprises and SMBs prefer outsourced, always-on AI operations over internal build efforts.Market Commentary
What Changed
Managed services are now recognized as the primary engine for scaling agentic AI, bypassing talent shortages and integration complexity.
Adoption Barrier
Trust and governance concerns around autonomous systems.
Key Risk: Failure to establish strong AI governance could stall adoption despite demand.
Trend Insight
Clients are responding to agent-enabled delivery by assuming AI-augmented service as the baseline, not a premium add-on, and by favoring firms that provide continuous, reliable outcomes over tool-level innovation. ROI is showing up first in SMB segments through managed services where agentic AI reduces cycle time and enables recurring revenue without eroding fees. The most important structural shift this period is the move from AI as an internal productivity lever to AI as commercial infrastructure—forcing partner-led changes in pricing, service design, and operating models.
Market
Market Moves, Regulation, and Ecosystem Signals
#1
EY
Big Four production deployment
Agentic AI is no longer experimental at the top end of the market; governance maturity is now a core competitive differentiator.Product Launch
What Changed
EY deployed enterprise-scale agentic AI across its global assurance network, embedding multi-agent workflows directly into EY Canvas for live statutory audits.
Market Implication
This marks the first widely recognized production use of agentic AI in audit, shifting competitive advantage from AI-assisted professionals to AI-run workflows with human oversight.
Regulatory or Liability Angle
Raises immediate questions around system-level accountability, audit quality control, explainability, and non-delegable human responsibility under PCAOB and professional standards.
Key Risk: Inspection and enforcement risk if agentic systems cannot demonstrate explainability, validation, and effective human supervision.
#2
Big Four (Deloitte, PwC, EY, KPMG)
Market structure shift
Scale, capital, and platform control are becoming decisive advantages in the agentic AI era.Market Signal
What Changed
All Big Four are now positioned as running autonomous or semi-autonomous AI agents across audit, tax, and managed services, increasingly described as "digital staff."
Market Implication
Revenue models are shifting toward multi-year managed services contracts, putting pressure on traditional billable-hour economics and mid-market firms to respond.
Regulatory or Liability Angle
Firm-wide AI systems of quality control become focal points for regulators, increasing exposure if controls fail at scale.
Key Risk: Margin compression or reputational damage if autonomous execution outpaces governance and professional oversight.
#3
Instead
Mid-market platform disruption
The next adoption wave is mid-market, driven by labor shortages and pricing pressure rather than experimentation.Product Launch
What Changed
Instead launched an end-to-end agentic AI platform handling tax research, planning, workpapers, filing, and resolution for CPA-scale firms.
Market Implication
Mid-market firms can now access integrated agentic workflows previously limited to large firms, accelerating adoption and platform standardization.
Regulatory or Liability Angle
Firms remain fully responsible for positions taken by autonomous agents, increasing malpractice and preparer-penalty exposure without robust review controls.
Key Risk: Over-reliance on a single agentic platform without sufficient human review and documentation.
#4
PCAOB
Regulatory trajectory
Compliance readiness around agentic AI is becoming as important as technical capability.Commentary
What Changed
PCAOB commentary continues to emphasize that AI use does not change auditor responsibility, with inspection focus shifting to firm-level AI governance and quality control systems.
Market Implication
Firms must invest in documentation, validation, and explainability of agentic systems or face heightened inspection risk.
Regulatory or Liability Angle
Inspection findings are increasingly likely to cite weak AI governance rather than improper AI usage per se.
Key Risk: Regulatory findings driven by inadequate system documentation or unclear accountability for AI-driven judgments.
#5
IRS
Public-sector AI escalation
Tax advisors must treat AI governance and documentation as defensive infrastructure, not optional enhancement.Market Signal
What Changed
IRS leadership highlighted expanded AI use in return processing, fraud detection, and taxpayer services, enabling near-real-time cross-checking of filings.
Market Implication
Tolerance for inconsistent positions and weak workpapers is shrinking, increasing the burden on preparers and advisory firms.
Regulatory or Liability Angle
Higher likelihood of audits, adjustments, and penalties if AI-assisted preparation lacks strong documentation and review.
Key Risk: Professional liability exposure from AI-generated errors detected faster by IRS systems.
Trend Insight
The Big Four are setting the pace by deploying agentic AI at production scale and reframing AI as digital staff rather than tools. Mid-market firms are the fastest followers, enabled by vertically integrated platforms like Instead. Regulators are not slowing adoption, but they are redirecting scrutiny toward governance, explainability, and firm-level quality control; human responsibility remains non-delegable regardless of autonomy. The structural shift is the move from AI-assisted work to AI-run workflows, driving consolidation around dominant platforms and elevating governance capability as a core source of competitive advantage.