“If you sell the tools, the models are getting better… you’re at risk. If you sell services, you’re delivering outcomes.” , Julien Bek, partner at Sequoia Capital, on etn.show

The 6:1 ratio

For every dollar a company spends on software, it spends six on services. That’s the figure Julien Bek put on the table in his interview with etn.show. And his conclusion is direct: the next trillion-dollar company won’t sell tools, it’ll deliver the work done.

This isn’t a minor opinion. Bek is a partner at Sequoia Capital, the firm that backed Apple, Google, Stripe, and NVIDIA before they became what they are today. When Sequoia says services are the next massive market, it’s worth paying attention.

The logic is straightforward. If you sell software, you’re competing with AI models that improve every quarter. Your product today is worth less tomorrow because the underlying model does more for less. But if you sell the outcome (the closed books, the resolved ticket, the processed purchase) model improvement works in your favor, not against you.

From copilot to autopilot

Bek uses an example that makes this clear: QuickBooks costs about $10,000 per year. The accountant who closes the books using QuickBooks costs $120,000. The trillion-dollar company won’t sell better accounting software, it’ll close the books directly.

This shift is already happening across multiple verticals:

• Legal: A firm no longer needs contract review software, it hires a service that reviews contracts, compares clauses against precedent, and negotiates directly with counterparties
• Procurement: Instead of a purchasing platform, agents handle the full cycle: receive the request, compare three vendors, generate the purchase order, and route it to approval. The platform becomes a system of record, not the active tool
• Customer support: The ticket arrives, gets diagnosed, answered, and closed, all by an agent team. A human only steps in when the case falls outside parameters
• Compliance: Rather than running manual audits, a service continuously cross-references operations against current regulations, flags deviations, and produces the remediation plan

Agent frameworks from Anthropic, OpenAI, and Google already make this possible at scale. The infrastructure to build autonomous services exists, what’s missing is who connects it to each company’s real processes.

Why Latin America has a disproportionate advantage

The underlying technology is the same for everyone. A company in San José, Costa Rica, accesses the same Claude, GPT, and Gemini models as a company in San Francisco. The APIs don’t distinguish geography, agent frameworks are open source and the documentation is identical regardless of where you are.

What changes is who implements and at what cost.

A company in Bogotá, San José, or Mexico City can deliver AI implementation services at the same technical level as a consultancy in New York, because the technology is identical and technical talent in the region has grown strongly over the past decade. The difference: the cost of that implementation is significantly more accessible.

This isn’t about cheap labor. That cliché doesn’t apply here. It’s about teams with real enterprise implementation experience who operate at global quality standards, but in markets where operating costs are lower. The result: the client’s ROI arrives faster.

The arbitrage is clear:

Factor	NYC / SF	Latin America
AI model	Claude, GPT, Gemini	Claude, GPT, Gemini
Implementation quality	High	High
Implementation cost	$$$$	$$
Time to results	Similar	Similar
Client ROI	Positive in 6-9 months	Positive in 3-5 months

In our experience at IQ Source, clients working with teams in the region get the same operational results, in some cases faster, because local teams have the advantage of understanding Latin America’s regulatory and cultural context without needing adaptation.

What this means for B2B companies right now

If your company is buying software licenses hoping that built-in AI features will solve your operational problems, you’re betting on the $1 market. That market compresses every time a model improves, because your software vendor has to compete with the same AI everyone else uses.

The companies gaining real advantage are those investing in the $6, the services layer that turns AI capabilities into concrete operational results.

A practical example: an AI agent handling 200 support tickets per day, at 12 minutes each, that’s 40 hours of human work. At 85% automation with the rest escalated to people, ~34 hours are freed daily. But that agent doesn’t deploy itself. Someone has to design the workflow, set the escalation rules, define quality criteria, and monitor results. That “someone” is the services layer, and that’s where the real value sits.

This is exactly the model we operate under, we wrote about it in detail in The AI Agent Maestro: a Role, Not a Job Title.

For the full economics behind these implementations, our analysis of Enterprise AI Economics in 2026 breaks down the numbers with real cases.

The window is open

Sequoia isn’t betting on SaaS companies. It’s betting on services companies that use AI to deliver results, faster, cheaper, and at higher quality than the traditional model of “buy the software and figure it out.”

Most mid-market companies in Latin America still haven’t made the jump from software to services. The ones that do it first will build a hard-to-catch advantage, not because they’ll have better technology, but because they’ll have processes that work while their competitors are still configuring dashboards.

If you want to see how we build this in practice, our AI agents for enterprise operations guide details the full process.

Map your first software-to-service process

Do you have a process that today depends on expensive software and a team operating it manually? We can analyze it together in 30 minutes and show you whether there’s a concrete opportunity to turn it into an AI-delivered service, with real numbers, not theory.

Schedule a free 30-minute session →