Two-Person Teams Are Rewriting the $74 Billion AI Consulting Market

The Playbook: Deploy, Don't Advise

flowscope, a two-person team from Y Combinator's P26 batch, ships working automation in days — not the nine to twenty-four months a traditional engagement takes. Its agents sit on top of existing tools, learn who touches what and where things stall, then produce a process map, a redesign, and production automation on the systems the company already owns. No pilot. No deck. Working software.

Enterprises have spent years watching AI pilots stall in the gap between a model that works in a sandbox and the messy reality of how their business actually runs. The consultants they hire leave decks. The tools they buy sit in proof-of-concept purgatory. The work — invoices re-keyed between systems, approvals chased over email, the same spreadsheet rebuilt every week — keeps happening the old way.

AI-native consulting firms don't sell advice or software licenses. They deploy autonomous agents that shadow employees, baseline the process, redesign the work, and write back into legacy systems of record. That distinction — deployment over advice, autonomy over augmentation — separates them from both traditional consultancies and pure AI tool vendors. McKinsey, BCG, and Deloitte now all market AI frameworks and proprietary platforms, but their model still rests on leverage staffing: junior analysts doing research, partners selling board alignment, billing that rewards process over production. Their timelines to production remain measured in quarters. Pure tool vendors hand customers a toolkit and wish them luck. They don't own the outcome. AI-native firms own the outcome. They carry the integration risk, the data plumbing, the change management. They get paid when the automation runs.

The founders embody the split DNA this category requires. Javier Leguina was founding engineer at Model ML (YC W24) and a two-time founder before that. Samuel Mirpuri came from QuantumBlack, McKinsey's AI arm, and served as a platoon commander in the Singapore Armed Forces. Their pitch is explicit: they do what consulting firms charge millions for at a fraction of the cost and time. The Y Combinator launch post framed the problem plainly: the hard part of enterprise AI isn't the models anymore, it's deploying them against the messy reality of how businesses actually run. Integration is painful. Internal teams lack bandwidth. Traditional consultants leave a deck and a bill.

The technical stack signals the ambition. flowscope's founding engineer role lists Microsoft Azure, Rust, and TypeScript, a combination built for low-latency agent orchestration on enterprise infrastructure, not demo-ware. The company is actively seeking intros to CFOs and COOs at mid-market and enterprise companies with manual finance, admin, or operations processes, plus private-equity operating partners driving portfolio efficiency. That target list tells you where the first revenue lands: repetitive, high-volume workflows where the ROI math is unforgiving and the incumbent consulting bill is hardest to justify.

Forty-five upvotes on the YC launch post suggest the signal reached the right audience. David Lieb, the primary YC partner, backed a team of two with a thesis that the consulting industry's structural limits — leverage staffing, vendor alliances, strategy split from delivery, billing models that reward process — are finally exploitable.

The Market at Risk: A $74 Billion Repricing

That $74 billion figure is the headline, but it only tells half the story. The AI-native segment is growing six times faster than the traditional market it is beginning to cannibalize.

The traditional consulting pyramid solved three simultaneous problems: it packaged junior labor into leverage, it bundled strategy with implementation to justify high margins, and it sold certainty through branded frameworks. Research from FourWeekMBA notes that AI disrupts all three at once. Junior analysts who once spent weeks mapping processes can now be replaced by agents that shadow employees, baseline workflows, and write back into them in days. The leverage model — where a partner sells one hour and bills ten — collapses when the ten hours disappear.

Structural limits that incumbents have carried for decades are now exposed. An insider comparison of McKinsey, BCG, and Deloitte identified four shared constraints: the first three, and billing models that reward process over production. AI-native firms like flowscope bypass each constraint.

Talent economics accelerate the pressure. A shortfall of over 200,000 qualified AI practitioners in North America alone has pushed consulting fees upward by an estimated 30 percent relative to five-year averages. The Bureau of Labor Statistics put the median annual wage for management analysts at $101,190 in May 2024, but AI-specialized roles command far more. Traditional firms must pay escalating premiums for scarce talent while AI-native startups build the automation that reduces their own headcount needs.

Clients are responding. Industry surveys indicate corporations are adopting multi-vendor strategies, engaging multiple smaller firms rather than relying on a single large partner. The flexibility and cost effectiveness of independent consultants appeal to businesses minimizing overhead. Meanwhile, AI tools and no-code platforms have dropped 70 to 95 percent in price compared to previous years, lowering the barrier for boutiques to deliver sophisticated work.

The pricing model itself is shifting. Traditional time-and-materials contracts are giving way to outcome-based models that tie fees to tangible performance indicators: cost-to-serve reduction, revenue uplift, time-to-market acceleration. Early adopters report that aligning fees with outcome metrics improves cash flow predictability and drives internal accountability. Firms that cannot instrument their engagements with real-time analytics dashboards lose the ability to prove ROI, and without proven ROI, the premium brand justification erodes.

The $74 billion AI consulting forecast is not a new market growing beside the old one. It is the old market being repriced.

The Counterattack: MBB Buys Infrastructure, Hires Armies

The incumbents are not waiting. McKinsey, BCG, Deloitte, and the rest of the Big Four have spent the last two years converting their balance sheets into AI infrastructure: proprietary platforms, hyperscaler alliances, and hiring sprees that dwarf most tech companies' entire headcounts. The playbook is consistent: acquire or build a branded AI engine, retrain the workforce at scale, and lock in cloud partnerships that make switching costs painful for clients.

McKinsey's anchor is QuantumBlack, acquired in 2015 and now the firm's dedicated AI and data science arm. As of 2022 the unit was publicly described at "1,000+ practitioners"; more recent figures put it at roughly 1,700 people across 40-plus offices, billing senior work at $500–700 an hour. The firm-wide lever is Lilli, a GenAI assistant built by QuantumBlack and deployed in 2023 across 100,000-plus proprietary documents. McKinsey's AI transformation practice generated an estimated $4.2 billion in global revenue in fiscal 2025, and AI initiatives now account for roughly 40 percent of the firm's total work. Its methodology centers on six "Rewired" capabilities (strategy, talent, data, technology, governance, and adoption), and the annual State of AI survey (1,993 participants across 105 nations, fielded June–July 2025) serves as its primary market signal.

BCG took a different route. BCG X, formed in 2022 by merging BCG GAMMA, BCG Platinion, and BCG Digital Ventures, now claims 3,000-plus technologists, data scientists, engineers, and designers across 80-plus cities, including 200-plus PhDs. The unit's formal partnership with Anthropic and inclusion in OpenAI's Frontier Alliances give it direct access to frontier models, a differentiator it leans on heavily. The firm publishes a branded "10-20-70" methodology (10 percent algorithms, 20 percent technology, 70 percent people and process) and its AI at Scale survey (2,847 executives and practitioners across 84 countries) breaks adoption curves into 14 industry verticals and three company-size cohorts, yielding 42 distinct data cells.

Deloitte's AI practice sits inside its broader technology consulting and implementation business. Its Trustworthy AI framework, launched August 2020, defines seven dimensions (transparent and explainable, fair and impartial, robust and reliable, respectful of privacy, safe and secure, responsible, and accountable) and remains one of only three numbered methodologies any major firm publishes verbatim. Internally, PairD, Deloitte's GenAI assistant, rolled out from December 2023 to roughly 75,000 EMEA employees with a plan to reach 100,000 within six months. The firm's quarterly State of GenAI in the Enterprise survey (3,235 leaders across 24 countries, fielded August–September 2025) found that only 34 percent of organizations are truly reimagining their business with AI, while worker access to AI rose 50 percent in 2025. Deloitte is also scrapping titles like "analyst" and "consultant" in the U.S., replacing them with more specific job titles, a structural signal that the leverage model is bending.

Accenture, already the largest technology consultancy, disclosed roughly 77,000 AI and data professionals in its FY25 update, with 30,000 trained under the NVIDIA Business Group. The firm committed $3 billion over three years to Data & AI in June 2023 and added nearly 40,000 AI and data professionals in the last two years alone. Its formal Accenture–NVIDIA Business Group is one of only two genuinely differentiated hyperscaler alliances across the industry (the other is Bain–OpenAI). EY launched EY.ai in September 2023 alongside a $1.4 billion AI investment and has added 61,000 technologists since 2023. PwC committed $1 billion over three years (U.S., announced April 2023), deployed ChatPwC to roughly 200,000 users globally, upskilled 65,000 U.S. employees on GenAI, and introduced an engineering career track, the first new track in its 170-year history. KPMG launched Workbench in June 2025, backs a multibillion-dollar Microsoft cloud and AI commitment over five years (projecting $12 billion in incremental growth), and runs KPMG Clara for 95,000 auditors across 140 countries. Capgemini pledged €2 billion over three years (July 2023) and upskilled 150,000-plus team members via its Data & AI Campus. IBM Consulting Advantage, launched January 2024, gives all 160,000 consulting employees platform access. Bain's Vector digital delivery arm is anchored by a branded OpenAI services alliance (first announced February 2023, expanded October 2024 with a dedicated OpenAI Center of Excellence and co-design for retail and healthcare/life sciences).

The hiring arms race is visible in the numbers. Accenture's ~that many, EY's 61,000 technologists added since 2023, PwC's 65,000 upskilled U.S. staff, Capgemini's 150,000-plus trained members, and BCG X's 3,000-plus technologists represent a collective talent absorption that rivals the entire output of top computer-science programs. Yet the structural limits persist. Research from AI Advisory Practice identifies four constraints shared by all large firms: vendor relationship conflicts are embedded and invisible; senior expertise does not deliver the work; strategy and delivery are structurally disconnected; and engagement models, still heavily time-and-materials or fixed-price, do not align with AI program success. BCG's own data shows roughly 60 percent of providers still rely on those traditional contracting structures for agentic-AI-enabled services, while enterprises increasingly prefer output- or outcome-linked models (over 70 percent of decision makers, per BCG's survey of 115 enterprise executives and 75 technology-service-provider executives).

M&A is the accelerator. Accenture's acquisition strategy targets marketing transformation, customer engagement, and AI-powered digital experiences. The broader trend: consulting M&A deals are accelerating as firms use acquisitions to acquire specialized expertise they cannot hire fast enough. Solganick's YTD 2026 AI M&A report notes that while overall corporate dealmaking has become more selective, AI-driven acquisitions have kept pace, particularly in AI consulting and integration services. The hyperscaler alliances (Accenture–NVIDIA, Bain–OpenAI, KPMG–Microsoft) function as de facto exclusive channels, locking clients into cloud ecosystems while the consulting layer manages the change.

The counterattack is real, resourced, and visible in revenue mix. But the same surveys that show AI reaching 40 percent of work at McKinsey and BCG also show that only 20 percent of organizations are achieving revenue improvement from AI (versus 40 percent seeing cost savings), and MIT Sloan's 2025 AI Execution Census found 54 percent of enterprise AI pilots fail to reach production. The incumbents have bought the infrastructure and hired the armies. Whether their engagement models can deliver outcomes at the speed of agent-native startups remains the open question.

The Talent War: Implementation Is Where the Leverage Lives

The consulting disruption isn't just about business models — it's rewiring where the best engineers choose to work. Microsoft's $2.5 billion commitment to a 6,000-person forward-deployed engineering unit, announced in July, signals that the biggest talent grab in enterprise AI now sits inside implementation, not research. Amazon followed with a $1 billion FDE initiative days later. Anthropic and OpenAI both stood up their own FDE groups in May, partnering with private equity firms, banks, and consulting shops to embed engineers directly inside client operations. The pattern is clear: the bottleneck isn't model performance anymore. It's the scarce layer of people who can wire those models into live workflows.

LinkedIn's January 2026 ranking of fastest-growing U.S. jobs puts three AI roles in the top five: AI engineers (hiring concentrated in San Francisco, New York, Dallas), AI consultants and strategists (San Francisco, New York, Boston), and AI/ML researchers (same trio of hubs). The "consultant" label here is misleading — these are builders who ship production systems, not slide decks. As of September, 121 open AI Engineer positions across 79 Bay Area startups show 41 percent sitting at Series A–B stage, where the work is hands-on and the equity is real. Y Combinator's current Bay Area AI cohort lists 363 hiring companies, from agent-infrastructure plays like Lab0 and Glen to humanoid robotics (Darwin), quantum-AI hybrids (Conductor Quantum), and materials discovery (Discovered Materials).

Compensation has decoupled from traditional software engineering bands. Recruiting from Scratch's June benchmark puts senior SWE (5–8 years) total comp at $270K–$345K at Series B, with a median offer base of $228,000 and 68 percent of candidates holding competing offers at decision time. The premium for AI fluency runs 15–25 percent above equivalent-level SWE comp at non-AI startups. At the frontier labs, the ceiling is higher: Anthropic's live board data shows Performance Engineers at $350K–$850K (Zero G Talent's data shows), Staff+ Research Engineers at $500K–$850K (Zero G Talent found), and Pre-training Distributed Systems Tech Leads at the same band (Zero G Talent's figures put). Databricks, meanwhile, lists 475 salaried roles (Zero G Talent reported) with a median of $250K (Zero G Talent's data shows) and a band of $140K–$321K (Zero G Talent's figures put), with 44 new postings in the past week alone (according to Zero G Talent).

Defense tech, which nearly doubled funding to $49 billion in 2025, added 1,000 employees at Anduril in nine months and offers 40–100 percent pay premiums over FAANG, the most aggressive hiring shift in a generation. Space, robotics, and energy startups now compete for the same systems engineers who can integrate heterogeneous hardware, manage real-time data pipelines, and satisfy regulatory or safety constraints. The YC Summer 2026 batch makes the overlap explicit: Aktoria Robotics, Osmosis (post-training platform), Theorem (program verification), and Glen (agent orchestration) all recruit from the same talent tier that AI-native consulting firms and frontier labs target.

Meanwhile, the layoff backdrop sharpens the asymmetry. More than 150,000 cuts across 549 companies hit in 2025; 22,000 more in early 2026, including 16,000 in February alone. Microsoft trimmed 9,000, Intel 21,000, Amazon 14,000. But those reductions concentrate in legacy product lines and middle management. The net flow moves toward roles that combine model literacy with systems integration, exactly the profile AI-native consulting firms, FDE units, and defense primes are bidding up.

The talent market has split into four visible strata: frontier labs (OpenAI, Anthropic, Google DeepMind, xAI), Series B–D AI startups, big-tech AI divisions, and early-stage/seed ventures. AI-native consulting firms (flowscope, Lab0, and their peers) occupy a new niche between the second and third tiers, offering equity upside closer to early-stage with implementation scope that rivals big-tech FDE programs. Engineers who choose this path bet that the highest-leverage years of the AI cycle will be spent not training foundation models but deploying them inside the messy, regulated, legacy-laden operations that run the physical economy. The next signal to watch: whether defense and energy primes start acquiring AI-native consulting shops outright to lock in that deployment muscle before the FDE units scale further.

Outlook: The Consulting Market of 2030 Will Not Resemble 2024

The numbers tell a story of violent expansion. The spread reflects different definitions of what counts as "AI consulting," but the vector is unanimous: the addressable market is multiplying by an order of magnitude in a decade.

Asia-Pacific leads the growth charge at 35.6 percent CAGR, with IDC projecting $200 billion in AI investment across the region by 2030. Healthcare consulting will register the highest sector CAGR, driven by generative AI's move from experimental to FDA-cleared in imaging and drug discovery. Cognitive integration services (the work of stitching models into live workflows) will grow fastest of all at 36.68 percent. The U.S. consulting market overall, at that 2025 valuation, will reach $198 billion by 2034 at a modest 5 percent clip, meaning AI-native work will increasingly represent the growth layer atop a mature base.

The business model is shifting underneath that growth. They are yielding to outcome-based pricing tied to cost-to-serve reduction or revenue uplift. Early adopters report the shift improves cash-flow predictability and forces internal accountability — project teams must continuously validate that the AI layer delivers promised efficiencies. Firms that master this transition will lock in longer-term relationships and repeatable revenue streams. Simultaneously, the engagement model is morphing from discrete projects to continuous, platform-centric partnerships, a transition expected to complete within three to five years as ongoing model management and compliance become permanent requirements.

Vertical specialization is hardening into the primary moat. Healthcare, finance, and manufacturing now demand playbooks combining regulatory insight, domain data models, and ROI calculators; consultants who curate these command premium pricing. The white-space opportunities cluster in ethical AI governance, explainability frameworks, and safety protocols, areas identified as underdeveloped high-value niches. Generative AI advisory units are already emerging as dedicated practice areas inside large firms, while AI-as-a-Service integration consulting (stitching cloud platforms to legacy ERP) promises recurring revenue from ongoing optimization.

Regulatory readiness is becoming a prerequisite for scale. The EU's AI Act and emerging U.S. state-level statutes require localized expertise; firms with version-controlled policy repositories auditable during compliance reviews gain competitive edge. Cross-border data transfer restrictions force bespoke, region-specific model deployment, fragmenting what once looked like a global playbook. Consulting engagements now allocate substantial scope to risk-assessment workshops and compliance roadmaps, a cost center that becomes a differentiator when done rigorously.

Technology convergence will rewrite the engagement stack. Multi-modal models, federated learning, and AI hardware accelerators are compressing deployment timelines. MLOps and CI/CD pipelines are turning what were month-long integrations into week-long sprints. The integration of AI with IoT, edge computing, 5G, and blockchain is expanding consulting scope into real-time, decentralized decision-making at industrial scale: predictive maintenance in manufacturing, autonomous logistics in energy, sensor-fusion autonomy in defense.

The competitive landscape is bifurcating. Traditional giants (Accenture, Deloitte, McKinsey, BCG) are consolidating through acquisitions and strategic alliances with cloud providers, using brand equity and board-level relationships. AI-native startups like flowscope, deploying autonomous agents that map and automate processes end-to-end at a fraction of traditional cost and time, are attacking the leverage model itself. The winners will own proprietary AI assets and platform ecosystems, not just advisory methodologies. Partnerships with model providers (Anthropic, OpenAI) and cloud platforms (AWS, Azure, Google Cloud) are becoming table stakes.

Three signals to watch through 2026: the pace of outcome-based contract adoption in Fortune 500 renewals; the emergence of sector-specific foundation models that collapse implementation effort; and the first wave of regulatory enforcement actions that test whether compliance frameworks hold up in court. The consulting market of 2030 will not resemble the one of 2024 — it will be smaller in headcount, larger in revenue per engineer, and defined by who controls the agent layer that actually does the work.

Working in AI? Zero G Talent tracks the openings: see every open Databricks role, browse AI jobs, openings at Anthropic, and the people building the field.