Bullet Thesis: 18 Services Giants Waiting to Be Disrupted by AI Startups

Hi, it's Michael from firstminute capital. We are a $500m seed fund backed by 130 unicorn founders. Along with my brilliant colleagues Sam, Lina, Lorcan, and Adriana, I invest in European pre-seed and seed B2B software companies like Vocca, Scalera, and ai.work. Here's our latest thesis on where the real AI opportunity lies.

—

Like most early stage VC funds, we have now seen hundreds of "AI for outbound marketing" startups. And "AI customer support" startups. And "AI for market research". The pitches all make sense. These are all categories clearly ripe for AI disruption. But the sheer number of them is exhausting — and says a lot about the brutal competition ahead.

Here's what's not obvious though: there are dozens of service-heavy sectors that tech investors and founders are completely ignoring right now.

Our basic thesis at firstminute is simple: while cloud fundamentally changed the $650bn software market by digitising how companies store and access information, AI is going to fundamentally disrupt the $10 trillion services market by automating the actual work humans do. We're moving beyond "systems of record" that just track what happened, towards "systems of action" where AI agents can analyse, decide, and execute tasks that previously required armies of consultants, analysts, and specialists.

This shift makes finding services markets where companies employ thousands of people doing now-automatable tasks (which are non-obvious) a crucial task for early stage investors. So here are 19 companies you've probably never heard of. They're massive though. They do services you've maybe never thought of. And they represent exactly where founders should be building AI systems that can replace human-heavy workflows with intelligent automation.

1. Genpact, GEP (Procurement Services)

Genpact employs 125,000+ people generating $4bn+ revenue in the global procurement services industry. Put simply, procurement services companies help big corporations buy things — from office supplies to raw materials — managing trillions in global corporate spend but still relying heavily on Excel spreadsheets and email.

A core challenge is that procurement teams are made up of individuals each managing 1,000+ suppliers that have to tracking every contract clause, every shipment, and every invoice discrepancy. This is hard for mere humans! There is inevitable "leakage", or stuff that slips through the net, which costs large enterprises approximately 2% of total spend — equivalent to $200m annually for every $10bn of cost of goods sold. It's not incompetence by humans, just maths: you can't hire 1,000 people to babysit 1,000 suppliers effectively.

Key workflows involve manually scoring suppliers across dozens of criteria (pricing, quality, delivery, financial stability), contract analysis requiring legal teams to review hundreds of pages for unfavourable terms, and spend analytics teams pulling data from multiple systems to generate insights that are outdated by delivery. The industry also handles supplier onboarding, performance monitoring, and invoice reconciliation — all dependent on human coordination across fragmented systems.

AI is a big opportunity here. The technology can simultaneously read 10,000 contracts and cross-reference them against millions of invoices to flag every discrepancy in real-time — something you simply can't hire 1,000 people to do effectively. AI can automate vendor selection across entire supplier ecosystems, optimise contracts by identifying cost-saving opportunities across massive portfolios, and provide real-time spend analytics rather than periodic manual reports. Act I will see companies automating existing procurement jobs, but Act II companies will be able to do new tasks — like optimising supplier relationships across thousands of variables in real-time.

2. ZS Associates, L.E.K. Consulting (Pharma Market Access & Commercial Strategy)

ZS Associates and L.E.K. Consulting are two of the big consultancies that advise drug companies on what to build, how to price it, and how to launch it across dozens of reimbursement regimes. The work spans market landscaping and patient segmentation, value proposition development, price corridor setting, tender strategy, launch sequencing by country and key opinion leader mapping. Under the hood it's armies of analysts pulling from trial readouts, registries, stitching together payer interview notes, building spreadsheet-based pricing simulations, and hand-assembling 100- to 300-page value dossiers with meticulously sourced claims and country-specific labels.

These are big companies, with ZS employing 13,000 people and doing $4.2bn in revenue while L.E.K employs 2,100 people and does $800m in revenue. AI in theory should be able to augment, and possibly revolutionise, how these companies operate. Agentic workflows can: continuously ingest clinical data and guidelines to auto-draft country-specific HTA dossiers with citations; simulate price/volume trade-offs and reference-pricing across markets; generate and stress-test payer objection handling based on prior HTA precedents; run indication-sequencing and launch-wave optimisers constrained by supply, label timing, and comparator dynamics; auto-map key opinion leader networks and site feasibility from publications and trials; and keep living value stories aligned with label updates and post-marketing evidence. Act I tools could shrink weeks of desk research and slide-building to hours; Act II platforms could operate as "virtual market access teams" that plan, execute, and adapt launch strategy in real time — threatening the slide-based, bill-by-the-hour core of this sector.

3. Veritext (Court Reporting & Legal Support)

Veritext generates $750m revenue with over 10,000 court reporters and legal support specialists. It’s one of the many companies that provide the stenographers who create written transcripts of everything said in depositions, hearings, and trials — an essential but fragmented industry supporting the entire legal system.

The core challenge is that human stenographers are expensive, scarce, and create delivery delays of days while legal teams need instant access to transcripts. Scheduling court reporters across multiple jurisdictions requires extensive coordination between law firms, courts, and reporter availability.

Key workflows involve stenographers manually transcribing speech in real-time during depositions (often 6+ hours of complex legal testimony), exhibit coordinators physically managing and indexing dozens of documents during proceedings, and scheduling teams coordinating reporter assignments across geographic regions and time zones. The industry also handles video synchronisation with transcripts, delivery of certified transcripts within tight deadlines, and maintaining chain of custody for sensitive legal documents.

AI represents a shift from human-limited transcription to automated accuracy. Real-time AI transcription can now match human stenographer accuracy while delivering instant searchable transcripts instead of days-long delays. Automated exhibit indexing and management can eliminate manual coordination errors while providing real-time document search during proceedings. Intelligent scheduling optimisation can match reporter availability with case requirements across geographic regions automatically. This extends to AI-powered transcript review for accuracy, automated legal citation formatting, and eventually autonomous deposition management that handles the entire workflow from scheduling to final delivery.

4. SHI, CDW, Carahsoft (Government Technology Distribution)

Carahsoft generates $16bn+ revenue as the largest government technology distributor, while SHI processes $15bn and CDW handles $8bn in public sector sales annually. These companies are the middlemen that help government agencies buy technology — from Microsoft licences to cybersecurity software — taking a 10-15% cut on the $100bn that flows through them each year to vendors like Adobe, Cisco, and Salesforce.

The core challenge of this industry today is that government procurement workflows are heavily manual and paper-heavy, despite the digital nature of what they're selling. Each bid opportunity requires parsing unstructured RFP documents that can be hundreds of pages long with complex compliance requirements and vendor eligibility verification. Pricing involves manually emailing vendors for quotes since there's no unified API, often taking weeks to assemble responses to government bid requests. Award notifications arrive as unstructured PDFs and emails, while purchase orders require manual processing and fulfilment coordination across multiple government systems and vendor relationships.

AI represents an opportunity to automate the entire distribution workflow from bid identification to fulfilment. The technology can scrape and analyse thousands of bid opportunities automatically, parsing complex RFP requirements and routing them to appropriate vendors instantly rather than requiring human review. Pricing can be automated through AI agents that gather quotes from multiple vendors simultaneously, while bid preparation and submission can be handled automatically including all compliance documentation and certifications. This extends to automated purchase order processing, fulfilment tracking, and customer service — essentially creating an AI-native distributor that can bid on millions of opportunities annually while traditional players can only handle thousands due to labour constraints. Current distributors charging 10-15% margins for "pushing paper" create massive opportunity for AI automation to deliver the same service at near-zero marginal cost.

5. Mercer, Aon, Willis Towers Watson (Employee Benefits Administration)

Employee benefits administration involves managing 401k plans, health benefits enrolment, life insurance, disability coverage, and benefits compliance for millions of employees across large enterprises. Mercer employs 25,000+ people generating $5.2bn revenue with 45,000 employees serving 140+ countries; Willis Towers Watson serves 45,000+ employees across 140 countries with employee benefits accounting for 60% of revenues. But these massive operations still rely heavily on manual processes, outdated systems, and time-consuming tasks that burden HR teams, with manual intervention required throughout enrolment and administration. AI can optimise enrolment by analysing employee demographics and benefit utilisation patterns to recommend optimal plans, automate life event processing for marriages, births, deaths, and job changes that trigger benefit modifications across multiple carriers, generate regulatory filings automatically from transaction data with built-in compliance checking and streamline premium administration by automating the complex process of exchanging benefit enrolment data and premium files across various insurance carriers.

6. ISO Certifying Bodies, Consulting Firms (ISO Certification Services)

ISO certification involves helping companies achieve compliance with international standards like ISO 9001 (quality management), ISO 14001 (environmental), ISO 27001 (information security), and ISO 45001 (health & safety). Major certification bodies like DNV, SGS, Bureau Veritas, and TÜV generate billions in revenue collectively while thousands of consulting firms help 480,000+ European companies maintain these certifications annually. But the process remains painfully manual with companies spending 6+ months and producing 300+ pages of documentation just to achieve initial certification. AI can automate gap analysis by scanning existing company documents against ISO requirements to identify compliance gaps instantly, generate customised policies and procedures based on company-specific information collected through intelligent questionnaires, conduct automated internal audits by comparing actual practices against documented procedures through AI analysis, and maintain continuous compliance monitoring instead of periodic manual reviews.

7. Adecco, Randstad, Manpower (Staffing Agencies)

Adecco employs 34,000+ people placing temporary workers globally in the hundreds of billions staffing industry. Randstad operates in 38 countries with 36,000+ employees while ManpowerGroup employs 28,000+ people connecting job seekers with employers. Put simply, these companies are the matchmakers between people looking for work and companies that need workers — from warehouse staff to accountants — but their processes remain surprisingly manual despite dealing with digital resumes and job postings.

The core challenge is that human recruiters can only process a limited number of candidates per day while quality varies dramatically between individual recruiters. Manual resume screening takes hours per position as recruiters read through hundreds of applications looking for relevant experience and skills. Phone screening compounds this bottleneck since recruiters can only interview one candidate at a time during business hours, creating delays that cause good candidates to accept other offers. Candidate-job matching relies heavily on individual recruiter experience and gut feel rather than data-driven analysis of what actually predicts success in specific roles.

AI represents a shift from human-limited processing to unlimited parallel candidate evaluation. The technology can rank and score hundreds of candidates based on job requirements in seconds rather than hours, analysing not just keywords but understanding context and transferable skills. Automated interviewing through AI chatbots can conduct initial screening conversations 24/7, asking consistent questions and evaluating responses objectively. Predictive matching algorithms can learn from successful placements to identify patterns that human recruiters miss, improving candidate-job fit over time. This extends to automated reference checking, background verification, and even predicting which candidates are likely to no-show or quit early, allowing staffing agencies to scale their operations without proportionally increasing headcount.

These others below are better known, and some startups have tackled them, but still interesting (and potential AI goldmine). Writing less about them or this piece will get too long!

8. Questel, CPA Global, Dennemeyer (Patent & IP Services) Patent groups manage IP portfolios for major corporations, handling filing deadlines and prior art research that currently takes weeks. AI should be able to automate patent searches, provide intelligent deadline management, and conduct patent landscape analysis automatically.

9. CBRE, JLL, Greystar (Property Management). Property management involves maintaining buildings, coordinating tenant services, and optimising building operations for office buildings, apartments, and commercial real estate. CBRE employs 115,000+ people generating $31bn+ revenue managing billions of square feet globally while JLL employs 103,000+ people and Greystar manages 830,000+ apartment units. Cool startups are already taking on the property space such as Dwelly and Buena.

10. Mr. Cooper, Ocwen, Cenlar (Mortgage Servicing). Mortgage servicing companies collect payments and manage loan modifications for millions of homeowners, processing hundreds of billions annually. AI should be able to automate payment allocation, screen modification eligibility, and generate regulatory reports automatically.

11. 3M Health Information Systems, Optum360, Maxim (Medical Coding & Billing) Medical coding converts physician notes into standardised billing codes, processing billions of healthcare transactions. AI can automatically assign codes from clinical notes, predict denial likelihood, and conduct automated compliance auditing.

12. C.H. Robinson, Expeditors (Freight Brokerage) Freight brokerage coordinates shipments between companies and trucking firms, handling 8-10% of world GDP through logistics. AI enables dynamic pricing optimisation, load matching to minimise empty miles, and automated communication throughout shipments. Lots trying to build an AI-native Forto.

13. Standard Chartered Trade Services, HSBC Trade Services (Trade Finance Processing) Trade finance processing involves banks handling letters of credit and international commerce financing for trillions in trade annually. AI could in theory streamline the process of automate document verification, conduct sanctions screening, and provide intelligent trade finance matching.

14. IQVIA, Labcorp Drug Development (Clinical Trials) Clinical trials recruit patients and collect data for pharmaceutical companies, with 80% failing to meet enrolment timelines. AI should be able to mine healthcare records to identify eligible patients in hours, conduct risk-based monitoring, and detect adverse events in real-time. Firstminute portfolio company LindusHealth is a full stack digital native CRO in this space.

15. NielsenIQ, Kantar, Ipsos (Market Research) Market research companies conduct surveys and consumer analysis, with survey analysis taking weeks due to manual processes. AI can analyse thousands of survey responses in hours, process social media for sentiment analysis, and provide continuous brand monitoring.

16. ADP, Paychex (Payroll & HR Services) Payroll services process payments for 39+ million workers, with ADP generating $17bn+ revenue despite significant automation gaps. AI can handle complex payroll calculations automatically, provide employee self-service through chatbots, and deliver predictive analytics for retention.

17. Lionbridge, TransPerfect (Translation Services) Translation services convert content between languages for global brands, employing 8,000+ people generating $1bn+ revenue. AI enables machine translation post-editing achieving 5x faster results, maintains consistent terminology, and provides real-time translation capabilities.

The $10 Trillion Thesis

These sectors represent the tip of an enormous iceberg. Each highlights the same fundamental opportunity: AI is going to unleash productivity gains in services that will dwarf what cloud computing did for software.

While everyone fights over the more obvious markets, the real value creation will happen in sectors most VCs have never heard of. The founders bold enough to tackle these "boring" industries with cutting-edge AI will build the next generation of unicorns.

Want to discuss building in one of these sectors? We're actively investing in European founders attacking these massive, overlooked opportunities. Reach out to us at firstminute capital at sam@firstminute.capital, michael@firstminute.capital and lorcan@firstminute.capital.