What's new:

Perplexity has integrated with financial data network Plaid, expanding its AI platform into a full personal finance hub, letting users connect bank accounts, credit cards and loans directly alongside existing brokerage account access. More than 75% of Perplexity users already ask financial questions monthly, and the platform’s annual recurring revenue surpassed $450M in March 2026. That’s a 50% jump in a single month.

How it works:

Through Plaid’s infrastructure, which connects to thousands of financial institutions, Perplexity’s AI can now analyze spending behavior, track liabilities and calculate net worth across multiple accounts in real time. Plaid provides read-only access, and user data never touches Perplexity’s servers. Pro and Max subscribers can use the platform’s Computer agent to build custom net worth calculators, debt payoff planners and cash flow forecasts — grounded in both personal account data and institutional-grade sources like FactSet, S&P Global and SEC filings.

Behind the news:

This launch is a consumer-facing signal of a broader architectural shift in AI — one that researchers call neuro-symbolic AI. The core idea combines neural networks’ capacity for flexible reasoning and natural language with structured, verifiable data. LLMs hallucinate because they generate answers from pattern-matching over training data rather than verified ground truth. Grounding them in hard symbolic inputs — real account data, SEC filings, live market feeds — preserves the reasoning capability while removing the invention risk. Perplexity’s own positioning captures the intent precisely: “This isn’t an AI hallucinating stock prices — it’s pulling verified, auditable financial data.” Plaid is the symbolic layer. Perplexity is the reasoning engine. That combination is where the field is heading.

Why it matters:

The Financial Industry Regulatory Authority (FINRA) flagged in December that member firms’ use of generative AI was outpacing their internal controls, and the SEC made third-party AI data handling an examination priority for 2025. Both concerns trace directly to AI systems reasoning over general knowledge rather than verified data. The grounded architecture here addresses that gap. For accounting and finance professionals, this establishes the coming baseline: AI that works from a client’s actual financial picture, not a probabilistic approximation of it. The platform that answers every financial question, grounded in live account data, is moving to own the client relationship — and it’s not coming from a traditional financial services firm.

Our thinking:

Hallucination won’t be solved by building larger models. Grounding AI in permissioned, structured data is a more tractable path – and Perplexity Finance is an early, high-visibility proof point. For finance professionals, the practical implication is worth mapping now: Which parts of your workflow require AI reasoning over a client’s actual data, and which can run on general financial knowledge? Those are different tools with different reliability profiles, and knowing the distinction will matter increasingly as grounded AI systems move from novelty to expectation.

What's new:

PwC has launched PwC One, positioning it as a fundamental reimagining of how professional services are delivered — not another productivity tool, but an “AI-first” operating environment that combines the firm’s proprietary methodologies and compliance frameworks with autonomous AI capabilities. Crucially, the announcement comes with an explicit pricing model overhaul: Rather than paying for time, clients will increasingly pay for outcomes — or for access to digital platforms that deliver those outcomes at scale, through subscription or consumption-based pricing models. Tax analysis, financial reporting, sustainability assurance and deal diligence are all named as initial deployment areas.

How it works:

PwC One embeds AI into PwC’s existing methodologies and professional workflows, enabling what the firm calls “continuous insight” rather than episodic project-based engagements. The firm plans to convert some of its tax and consulting offerings into automated tools that clients can access independently, in some cases without a PwC person in the loop — effectively converting professional expertise into self-service products. Human judgment remains in the loop for complex decisions, but the analytical throughput underneath it is increasingly automated.

Behind the news:

The Big Four have historically hired thousands of junior staff to perform routine tasks, billing clients by the hour. AI automation now threatens both the workforce model and the revenue model simultaneously. The subscription pivot is the more consequential signal. Hourly billing has been the structural backbone of professional services for over a century. And abandoning it isn’t an operational tweak, it’s a business model reinvention. Senior staff bonuses will track revenue and margin per professional rather than hours billed, while leaders will be measured on progress in moving services to AI, which tells you exactly where PwC’s internal incentives are now pointed.

Why it matters:

For accounting and finance professionals, PwC One signals where the competitive bar is moving — and fast. Tax analysis and financial reporting, long the defensible core of CPA value, are explicitly in scope. But the pricing shift may be the sharper threat: When a Big Four firm starts selling tax and advisory access on a subscription basis, it reframes the competitive landscape from “who has the best people” to “who has the best platform.” PwC’s own research, published in January, found that more than half of businesses using AI saw little or no benefit, meaning clients will scrutinize results, not just the model. That’s an opening, but only for firms who can demonstrate clear outcomes.

Our thinking:

The hourly billing model didn’t just determine how firms got paid — it shaped how they were organized, how they hired, how they trained, and how they measured performance. Subscription pricing unravels all of that at once. PwC is already recruiting fewer traditional accountants and consultants in favor of engineers and data specialists. The profession has seen technology compress timelines before (spreadsheets, tax software, cloud audit tools) and survived each time by moving up the judgment stack. The question now is whether that stack is tall enough. If PwC One and its equivalents at Deloitte, EY and KPMG commoditize the analytical layer and wrap it into a subscription, the remaining value proposition for independent practitioners narrows to relationship depth, local knowledge and fiduciary trust. Those are real advantages, but only if you’re actively building them now and not after the platform is ubiquitous.

What's new:

International Data Corporation’s (IDC) Feb. 2026 InfoBrief surveyed 1,005 audit and assurance decision-makers and found that most firms feel only moderately prepared to reskill staff for effective AI use. In the same survey, 76% believe AI will fundamentally transform audit within ten years, and 88% agree it could risk undermining professional judgment. The profession is staring at a transformation it knows is coming, with a workforce it has not equipped to meet it.

How it works:

The report breaks the readiness gap into three concrete barriers: cost, talent shortage and the absence of a clear AI roadmap inside the firm. Vendor support isn’t filling the gap. Only 28% of audit leaders are very or extremely satisfied with the training their software vendors provide, and a majority say vendor tools lack depth and don’t reflect the latest advancements. Meanwhile, 79% rate algorithmic bias in critical AI systems as moderately, very or extremely significant, and 64% say auditors should always validate AI outputs before they reach a client.

Behind the news:

The reskilling number is the loudest finding, but the structural finding underneath it is more important: 66% of respondents say there is an urgent need for a globally harmonized AI framework for audit and assurance, and 55% are willing to trade some level of AI performance for stronger security or safety. Audit leaders are not asking for more capability. They’re asking for guardrails, standards and an execution model that lets them deploy AI without forfeiting the assurance the profession is built on. The vendors selling capability are answering a question the buyers have already moved past.

Why it matters:

The gap between “AI will transform audit” and “we have not trained anyone to do that work” is where firm-level risk lives. It surfaces in three places:

1. Talent - where the firms that don’t build AI fluency this year will lose the people who already have it.
2. Client engagements - where AI-generated work product will appear in deliverables whether the firm has approved it or not.
3. Liability - where unvalidated AI outputs become evidence in disputes that didn’t exist three years ago.

The 64% validation requirement is not a best practice. It is a defensibility standard, and it has to be built into the workflow before the workflow runs.

Our thinking:

The honest read of this data is that the profession agrees on the destination and disagrees on the path. Every firm leader knows that AI changes audit engagements. Almost none has a written, signed-off plan for who validates what, where the training budget sits and what the firm will and will not let AI do on its behalf. That document (boring, internal, unglamorous) is the one piece of work that separates the firms that come through this transition with their judgment intact from the ones that quietly outsource it.

What's new:

The World Economic Forum’s (WEF) “Four Futures for Jobs in the New Economy” white paper models what AI does to labor markets through 2030 across four scenarios. Here’s the headline for any firm leader making compensation decisions: In scenarios where AI advances quickly and the workforce is ready, the gap between AI-ready and non-ready workers widens significantly. Across the broader survey, 54% of executives expect AI to displace existing jobs, 24% expect new job creation, and only 12% expect a positive impact on wages overall.

How it works:

The WEF builds its four futures from two variables: the pace of AI advancement and the depth of workforce readiness. In Supercharged Progress (fast AI, ready workforce), productivity climbs and the AI-ready earn dramatically more than everyone else. In the Co-Pilot Economy (gradual AI, ready workforce), labor productivity growth crosses 1.5%, fluidity in the labor market rises and wage gaps narrow modestly. In the Age of Displacement (fast AI, unready workforce), unemployment spikes and consumer confidence falls below the historical low. The scenario that defines a given firm or country is not decided by AI; it is decided by what was done with talent in the prior 24 months.

Behind the news:

LinkedIn data inside the report shows demand for AI literacy skills rose roughly 70% from 2024 to 2025, and the share of tasks performed by technology has jumped sharply since then. Capital is following the same logic: Investment is concentrating in AI and digital-asset fintechs and a small set of compelling early-stage challengers, while mid-stage firms face capital scarcity. The labor market and the capital market are running the same screen, and they are screening for the same thing — readiness, not headcount.

Why it matters:

For accounting and finance leaders, this reframes the talent conversation. The competitive pressure is not “do we hire more people?” It’s “do we have the people who can charge a premium?” The wage polarization in the WEF model is not only a forecast of inequality between firms and clients. It’s a forecast of inequality inside firms, between the partners and managers who can run AI-augmented engagements and the staff who cannot. Compensation bands, billing rates and partnership tracks are all about to be tested against that distinction, and most firm comp models were not designed to express it.

Our thinking:

The “no-regret moves” the WEF identifies — align technology and talent strategy, invest in human-AI collaboration, anticipate talent needs, future-proof the value chain — sound generic until they are translated into a firm calendar. Aligning technology and talent means the AI tooling decision and the hiring plan are made in the same conversation, by the same people, in the same quarter. The firms that close that organizational seam first will set the wage benchmark the rest of the profession ends up hiring against.

What's new:

McKinsey’s new report, “The Next Age of Fintech,” puts a number on a market the profession’s clients are already living inside: $650 billion in fintech revenue in 2025, representing 4% penetration of the $15.5 trillion financial-services market and growing at 21% year over year — roughly 3.5 times faster than incumbent banking. If that trajectory holds, the market reaches $2 trillion by 2030 and 9% penetration. Five fintechs are now approaching $100 billion valuations. McKinsey calls this the Fifth Age of fintech: the era after the 2021 hype cycle and the 2023 reset, in which scaled players are profitable, AI-native insurgents are arriving, and the regulatory environment is evolving alongside them.

How it works:

Three structural shifts give the chapter its shape:

• First, more than 50% of fintech acquisitions in 2025 were made by other fintechs — consolidation, not venture-funded experimentation, is now driving M&A.
• Second, 21 applications were filed for U.S. banking charters in 2025, more than in the previous four years combined. The regulatory perimeter is expanding.
• Third, “horizontal” fintechs — software firms selling infrastructure to incumbents rather than competing with them — represent 13% of fintech revenue and have grown roughly 25% faster than direct competitors over the past four years.

Behind the news:

The number that matters most to advisors is buried later in the report. Of the $35 trillion in stablecoin transaction volume in 2025, only about 1% relates to true payment activity. The remaining 99% is trading and crypto-native flow. Read carefully, that ratio is not a sign that stablecoins have failed as payment rails. It is a sign that the payment use case is still in front of us and that the infrastructure for it has already been built. Lending is moving in parallel: Emerging-market fintech lending is growing close to 50% annually and Latin American fintech overall is expanding at a 40% compound rate.

Why it matters:

For accounting and advisory practices, the client base is tilting. The bookkeeping client of 2020 has fintech exposure in 2025 — embedded payments, lending APIs, treasury through a fintech rather than a bank and stablecoin balances on the books. The audit, tax and advisory questions follow the money: revenue recognition for embedded financial products, controls testing against API-driven flows, treasury reporting for assets that move on chains and transfer pricing across jurisdictions where the fintech entity sits in a different tax regime than the operating entity. Firms that do not have a working point of view on these questions are about to start receiving them.

Our thinking:

The Fifth Age framing is useful because it names what came before it. The first three ages produced fintech as a category. The fourth, the reset, separated the businesses from the decks. This one is about distribution, infrastructure, and trust. In a market growing 21% and consolidating into a smaller number of larger players, trusted advisors with regulatory credibility are not displaced by the fintech wave, they are integrated into it. The opportunity is to be the firm that helps a fast-growing fintech meet its assurance, controls, and compliance burden as it scales. The risk is being the firm that finds out about its client’s fintech stack from the audit working papers.