“For every dollar spent on software, six are spent on services.” – Julien Bek, Sequoia Capital
❓ What You’ll Learn
- How does the “copilot to autopilot” shift change who makes money in every vertical?
- What is the 6:1 ratio that makes service-as-software businesses six times larger than SaaS?
- Where can a solo founder launch a vertical service-as-software business this month?
- How does “liability as a service” create a moat that pure automation won’t easily replicate?
- Why will freelance marketplaces lose their commodity service tiers to agent operators?
- Why does picking the vertical matter more than picking the model?
- Why do AI-first companies run at 50-60% gross margins instead of the 80-90% SaaS enjoys?
- Why is “no switching cost” the strongest critique of selling outcomes instead of tools?
💎 Why It Matters
The services market dwarfs the software market.
AI agents collapse the cost of delivering those services to near zero.
🔍 Problem
SaaS automates the interface.
The labor behind it stays manual and expensive.
💡 Solution
Sell the outcome directly.
AI agents deliver the closed books, the launched campaign, the filed claim.
The software is invisible to the buyer.
🏁 Players
AI-Native Service Companies
- Sierra • AI customer service agents that handle conversations end to end
- Harvey • AI legal platform expanding from copilot to full autopilot for law firms
- Mercor • AI recruiting that matches, vets and places candidates autonomously
- Cognition AI • AI software engineering agent (Devin) that ships code from spec to PR
- Basis • AI-native accounting platform that closes books autonomously
- Lawhive • AI-powered legal services firm pairing human lawyers with AI agents
- WithCoverage • Insurance procurement autopilot that sells to the CFO, not the broker
- Anterior • Healthcare revenue cycle automation replacing offshore billing departments
Agencies Pivoting to AI Delivery
- Jellyfish • Global digital agency using AI agents in media buying and campaign execution
- Manicule • AI-native documentation agency with agents handling code verification at scale
Agent Builder Platforms
- Lindy.ai • No-code agent builder with deep CRM, email and scheduling integrations
- Gumloop • Visual agent builder for designing multi-step AI workflows
- Zapier Agents • AI agent layer on 6,000+ app integrations
Infrastructure and Orchestration
- LangChain / LangGraph • Graph-based agent orchestration framework with broad ecosystem adoption
- CrewAI • Role-based collaborative agents, natural fit for agencies transitioning to AI
Billing and Monetization
- Metronome • Usage-based billing for outcome and usage pricing
- Flexprice • Open-source billing for AI-native companies with token, credit and outcome models
Enterprise AI Service Delivery
- Edra • Automating IT processes with outcome-based products for managed services
🔮 Predictions
- Freelance marketplaces will lose their commodity service tiers to agent operators.
- A “trust premium” will emerge for human-delivered services in judgment-heavy verticals.
- Management consulting ($300-400B market) is mostly judgment work that resists automation.
- The work that resists automation becomes more valuable because everything around it can be.
- Legal strategy, M&A advisory and executive recruiting involve relationship dynamics agents won’t easily replicate.
- Most agentic AI projects will fail and the survivors will specialize in verticals.
- Gartner predicts 40% of agentic AI projects canceled by the end of 2027.
- Agent reliability remains below 55% for complex tasks. 85% of AI projects fail before production.
- Vertically specialized companies tune agents to bounded problem spaces, achieving reliability faster.
☁️ Opportunities
- Own a vertical and sell the outcome. Pick a service where the work is intelligence-heavy, outsourced and has clear deliverables. Wire agents. Charge per result.
- Sell compliance and certification for agent-delivered services. Build the SOC 2 equivalent for AI-delivered outcomes.
- NIST launched the AI Agent Standards Initiative.
- EU AI Act mandates traceability for high-risk systems.
- PwC launched North America’s first ISO 42001 certification for AI Trust. Nobody’s built the self-serve version.
- Run an education program for “agent operators.” Certify the new job category.
- Upwork reported 27% increase in demand for AI-skilled freelancers.
- Offer cohort-based courses, job board, community membership.
- The fastest-growing categories are “build production AI systems” and “deliver measurable business outcomes.”
- Offer “human-verified” premium services. The counter-position to full automation.
- The trust premium is highest in verticals where errors carry liability.
- Charge 2-3x the AI-delivered price. Guarantee human judgment for M&A advisory, executive search, legal strategy.
🏔️ Risks
- Agent Reliability Gap • AI agents fail at 91%+ rates for complex tasks. Silent quality degradation goes undetected by traditional monitoring. Vertically specialized agents perform better, but the ceiling is real.
- Inference Cost Squeeze • AI-first companies run at 50-60% gross margins vs 80-90% for SaaS. Variability compresses margins when verticals get competitive.
- Foundation Model Dependency • Most service-as-software businesses run on 2-3 API providers they don’t control. One pricing change can break the economics overnight.
🔑 Key Lessons
- Each decision should answer: “What outcome does the customer walk away with?” Companies that sell access will lose to companies that sell the results.
- Pick the vertical before you pick the model. Domain knowledge is the moat. The foundation models are commoditizing. Knowing what a correct monthly close looks like matters more than which AI you use.
🔥 Hot Takes
- The real product is liability, not labor. People could close their own books. They hire a service provider so someone is on the hook when the books are wrong. The first company to offer an SLA with financial penalties for AI-delivered work could own its vertical. “Liability as a service” is a moat pure automation won’t easily replicate.
- The copilot companies are building their own replacement. Harvey sells to law firms while learning to do the work law firms charge clients for. It’s one of the rare business models where improving your product threatens your customer.
😠 Haters
“‘Service as software’ is what agencies already do. You’re selling outcomes with better tools. It’s not a new business category.”
Agencies scale by hiring. More clients means more headcount and quality variance. Service-as-software scales by deploying another agent instance. That’s a margin structure distinction.
“There’s no switching cost when you sell outcomes. A client buying ‘books closed by the 5th’ can switch providers overnight.”
Every engagement generates proprietary training signals. An operator with 200 clients has tuned agents to edge cases a new entrant hasn’t seen. The lock-in is delivery quality.
“Foundation model providers will vertically integrate and crush the operator layer. Why would OpenAI leave that margin on the table?”
Foundation model providers want to be the platform under every vertical instead of competing in each one. Running a bookkeeping service means understanding GAAP, managing clients and carrying liability. This is operational depth most platform companies don’t want.
“Vertical expertise will erode as models improve. Today’s domain knowledge moat disappears when GPT-6 can close books out of the box.”
A model that can generate a monthly close doesn’t know if the close is right. The moat is to know what “done correctly” looks like.
🔗 Links
- Service as Software: A New Economic Model for the Age of AI Agents • Thoughtworks breaks down how AI agents shift the value chain from selling tools to delivering outcomes. Pairs well with the Sequoia thesis.
- The SaaSpocalypse: AI Agents Disrupting the Software Industry • How AI agents triggered a $2T market cap wipeout in early 2026 and what it means for the SaaS-to-services transition.
📈 Want the full picture?
Why will outcome-based pricing at $0.99 per ticket replace $50,000/year SaaS seats as the default for AI-native companies?
How does the 6:1 ratio between services and software spend turn a $10M company into a $60M company serving the same buyers?
Where does the “Datadog for agents” opportunity sit and why hasn’t anyone built it yet?
What does the “Shopify for service-as-software” look like and why hasn’t anyone built it yet?
How does a model abstraction layer protect your margins when the API provider changes pricing overnight?
Why will the next wave of million-dollar one-person businesses be service-as-software operators, not SaaS founders?
What happens when the 12-18 month window for vertical operators closes and agent platforms ship one-click templates?
Trends Pro has the answers. Plus 39 players, 7 predictions, 11 opportunities and 12 links.
Get Weekly Reports
Join 54,000+ founders and investors
📈 Unlock Pro Reports, 1:1 Intros and Masterminds
Become a Trends Pro Member and join 1,200+ founders enjoying…
🧠 Founder Mastermind Groups • To share goals, progress and solve problems together, each group is made up of 6 members who meet for 1 hour each Monday.
📖 160+ Trends Pro Reports • To make sense of new markets, ideas and business models, check out our research reports.
💬 1:1 Founder Intros • Make new friends, share lessons and find ways to help each other. Keep life interesting by meeting new founders each week.
💲 100k+ Startup Discounts • Get access to $100k+ in startup discounts on AWS, Twilio, Webflow, ClickUp and more.
Brought to you by the team behind HeadsUp
