The case

Why Building SaaS Changed After AI

Most "fractional CTO for SaaS" advice still describes the 2021 playbook. The model has shifted under everyone: the architecture, the pricing, and the moat are all different problems than they were two years ago. I've been doing exactly this strategy work, most recently redesigning a platform's business model around usage-based pricing at MateBio.

The Seat Is Dying

Per-seat pricing assumes a human logs in and clicks. Now an agent (yours or your customer's) does the work, and one API or MCP connection fronts unlimited users with no seat to meter. In Kyle Poyar's 2025 State of B2B Monetization survey, only 15% of companies still called seats their primary pricing model, down from 21% a year earlier. I build metering for consumption rather than headcount, so the seat model won't cap what you can charge.

Hybrid Pricing, and Margin Is Now an AI Problem

The default model is now hybrid: platform fee plus usage plus overage. And every AI feature carries a real per-call cost, so margin is now an architecture decision. I wire the metering and per-account cost controls so an AI feature can't turn an 80% margin into 40%.

The Moat Moved

AI commoditizes what used to be defensible: integrations, data curation, even your schema or ontology. What's left as a real moat is proprietary data with network effects, workflow lock-in, and eventually outcomes. I help you spend engineering on the new moat instead of on the parts AI just made cheap.

One caution: outcome-based pricing is everywhere right now, and it only fits some businesses. It works when the outcome is fast and clearly attributable to your product, the way Salesforce can charge per closed deal. It's a trap when outcomes are slow or binary. I saw this directly at MateBio, where the "outcome" is a 10-year, billion-dollar drug approval you can't run a SaaS P&L against. Knowing which kind of business you are is the call I help founders make before they copy someone else's pricing.

What it takes

The Four Things SaaS Founders Actually Hand a Fractional CTO

Strip away the titles and most B2B SaaS engagements come down to these four. Each one is cheap to get right early and brutally expensive to retrofit.

Multi-Tenant Architecture

Tenant isolation, a data model that won't need a rewrite at customer 50, and a noisy-neighbor strategy so one account can't degrade the rest. The choice between shared, siloed, or pooled tenancy is made once. I make it deliberately.

Metering, Billing & Hybrid Pricing

Trustworthy metering and entitlements wired to your plans, on a billing stack (Stripe, Orb, or similar) built for hybrid pricing (platform fee plus usage plus overage) and for agent or API consumption as well as seats. This is what keeps your pricing model from being frozen by your codebase when the market moves off seats.

Security & Compliance for Enterprise Sales

RBAC, audit logs, SSO, encryption, and a SOC 2 path scoped to what actually unblocks deals, so compliance doesn't stall the roadmap. The goal is to answer the security questionnaire in days.

AI Features That Don't Leak Data or Money

Most SaaS roadmaps now have an AI line item. I ship RAG and agentic features with tenant-aware data boundaries, cost controls per account, and a multi-provider fallback, so an AI feature improves retention instead of destroying gross margin. And when customers reach you through their own agents, I expose curated, metered surfaces. A generic endpoint would let them take down your database or rebuild your margin layer on top of you.

Comparison

B2B and B2C SaaS Are Not the Same Build

The wrong reference architecture wastes months, so it pays to separate the two. B2C SaaS is a scale-and-latency game: millions of light sessions, viral signup flows, and cost-per-user measured in cents. B2B SaaS is a trust-and-control game: fewer accounts, each worth far more, each demanding isolation, permissions, an audit trail, and an admin who wants to see everything their team does.

I focus on B2B. That means I optimize for the things that close and renew enterprise contracts (provisioning, roles, security posture, and uptime that survives a procurement review) rather than for consumer-grade virality. If you're selling to other companies, those priorities are the difference between a product that demos well and one that passes legal.

This is also why a generic "CTO as a service" arrangement needs more for SaaS: the deliverable is the specific B2B SaaS architecture that makes you sellable to a 500-person company.

In practice

Case Study: Scaling an Established B2B SaaS Platform

Director of Engineering · Cascade

I led the Integrations, Metrics, and Reports teams at Cascade (the strategy-execution platform) and sat on the global Product Leadership Team, fully remote across time zones. Cascade is a mature multi-tenant B2B SaaS serving enterprise customers, so the work was scaling a product that already had demanding buyers.

Metrics Library. A front end, serverless functions, and an ingestion pipeline that let customers pull core metrics from business systems, data lakes, BI tools, and spreadsheets. The enterprise-integration surface every B2B SaaS eventually has to build.
Serverless Integration Architecture. Designed on AWS serverless so the pipeline scaled with customer data volume, without standing up dedicated infrastructure per tenant.
Balance Framework. Made engineering investment legible to the business. Resource allocation across teams became a data-informed decision.
Technical Leadership. Hired tech leads and principal engineers and served as the technical point of reference across Engineering, Product, and Customer Success: the durable team that keeps a B2B SaaS shipping as it grows.

In practice

Case Study: Scaling an AI SaaS Platform for Biotech

CTO · MateBio

I led the transformation of raw biomedical knowledge-graph data into a scalable, defensible, commercially viable SaaS platform. MateBio, a spin-off from the University of California, San Francisco, gave wet-lab researchers at biotech and pharma companies AI-powered tools to explore complex biological relationships through natural-language queries.

GraphRAG Pipeline. A multi-tool LLM agent that turns natural-language biomedical questions into validated Cypher over a Neo4j knowledge graph, with entity recognition, provenance tracking, and confidence scoring.
Entity-Resolution Engine. A hybrid engine that maps imprecise researcher terms to exact graph identifiers, fusing heuristic NER, type-scoped full-text and vector search, and small-model disambiguation.
Multi-Tenant, Multi-Cloud Platform. Stood up the SaaS platform across AWS and GCP with Terraform-driven CI/CD, automated migrations, secure multi-tenant APIs, and a tested cross-cloud database migration path.
Systems-Biology Analytics. Added a Graph Data Science centrality layer and a synthetic-EHR "signatures" service. It turns the graph from a lookup tool into a discovery engine that deepens the product moat.
Business-Model Redesign. Moved the platform off ad-hoc seat pricing toward a hybrid platform-fee-plus-usage model with metered, query-level billing, and mapped where outcome-based pricing genuinely applies and where it's a trap. Post-AI, pricing is as much a CTO problem as the architecture.

Timing

Fractional vs. Full-Time CTO for a SaaS Startup

A full-time CTO for a SaaS startup costs $250K+ in salary plus meaningful equity. Before product-market fit, that's the wrong spend: you're paying executive comp for work that is mostly architecture and hands-on building. A fractional CTO gives you the senior judgment on the four decisions above without the cap-table dilution.

Hire fractional while the questions are technical, like what to build, how to make it multi-tenant, how to charge, how to pass security. Hire full-time when the questions become organizational: managing 15+ engineers, owning a multi-year roadmap, sitting on the exec team. Part of my job is telling you when that line has been crossed and helping you make the transition cleanly.

If you're a US-based SaaS startup, the same logic applies with a time-zone advantage. See fractional CTO for US startups.

Stage

Fractional CTO for B2B SaaS Startups: Seed to Series A

A seed-stage B2B SaaS startup and a Series B platform are different engineering problems, and most SaaS architecture advice is written for the second one. At seed you have two design partners, no security questionnaire yet, and a codebase that has to stay cheap to change. By Series A there is a procurement review on the table, a customer who wants their data in their own region, and a billing system that cannot say what anyone owes this month.

Three decisions turn expensive in between: tenancy, metering, and where customer data lives. All three are nearly free at three accounts and brutal at thirty. Most of what a startup fractional CTO is worth is making those calls deliberately and early, then getting out of the way while the team ships.

Pivots are the other thing founders bring me. A SaaS startup moving from B2C to B2B, or from a tool to a platform, or from seats to usage, usually finds the pivot stuck in the data model. You can rewrite a pricing page in an afternoon and still be six months from billing for it, because the events you would charge on were never recorded.

So for a seed-stage team the scope is small: pick the tenancy model, wire metering you can trust, write the security answers down before a buyer asks. Three months of that is worth more than a year of a full-time hire you cannot afford yet.

Formats

How Engagements Are Structured

Most SaaS engagements start with a $4,000 assessment of your multi-tenancy, billing, security posture and AI plans, then move to Advisory + Build Slice at $5K–8K a month or an Embedded CTO at $10K–15K a month. Every engagement runs month to month with 30-day notice, paid in cash, with no equity. See full pricing

Next step

Let's Make Your SaaS Sellable to Bigger Customers

Whether you're choosing a tenancy model, wiring up billing, prepping for a security review, or shipping your first AI feature, let's talk through where your SaaS is and what to fix first.

Book a Free 30-min Call Connect on LinkedIn

FAQ

Frequently Asked Questions

When should a B2B SaaS startup decide on multi-tenancy?

Before the second paying customer, ideally before the first. The tenancy model (pooled, siloed, or a hybrid) shapes your data model, your security story, and your per-customer cost. Changing it later is one of the most expensive migrations a SaaS company can face, so it's the first thing I lock down.

Can a fractional CTO get us through a SOC 2 or security review?

Yes, and it's a common reason SaaS founders call. I scope the security work to what unblocks the deal in front of you (access controls, audit logs, SSO, encryption, and the SOC 2 path) rather than treating compliance as a project that swallows the roadmap. The aim is to answer the questionnaire in days.

How do you add AI features without blowing up our margins?

By treating AI cost as a first-class metric: per-tenant cost controls, caching, model routing across providers, and falling back to cheaper models where quality allows. I've built multi-provider abstractions over OpenAI, Anthropic, and Bedrock so an AI feature stays affordable as usage grows instead of eroding gross margin.

Should we still price per seat?

Probably not as your only lever. Seats still work for the surface a human clicks. But the moment agents or APIs consume your product, a seat stops being a meter. One connection can front unlimited usage. Most B2B SaaS is moving to hybrid (platform fee plus usage), and I'd build the metering for that now even if you launch on seats, so you're not forced into a migration when your customers' agents arrive.

Is outcome-based pricing the future?

In specific cases. It works when the outcome is fast, measurable, and clearly attributable to your product. Salesforce can price per closed deal. It breaks when outcomes are slow or binary; at MateBio the outcome was a 10-year, billion-dollar drug approval you can't run a SaaS P&L on. I help you tell which kind of business you are before you copy someone else's pricing.

Do you build, or just advise?

Both, depending on the engagement. I ship production code alongside the team when that's what moves the needle. Most SaaS startups before product-market fit need a builder more than a meeting. As you grow, the balance shifts toward strategy, hiring, and architecture review.

We're early and pre-revenue. Is a fractional CTO worth it?

Often more so, because the decisions that are cheapest to get right are made earliest. A short assessment plus advisory can save you from the multi-tenancy and billing rewrites that derail SaaS startups six months in, for a fraction of a full-time hire.

What does a fractional CTO do for a seed-stage B2B SaaS startup?

Three things, in order: pick the tenancy model, stand up metering you can trust, and write the security answers down before a buyer asks for them. Everything else at seed stage is product work your team can already do. I take the decisions that are cheap at three accounts and brutal at thirty, and leave the rest alone.

We're a B2B SaaS startup pivoting our pricing. Where does that usually break?

In the data model. Teams move from seats to usage and find the events they would bill on were never recorded, so there is nothing to meter. The sequence that works: instrument usage first, run it in shadow mode against the old plan for a month to see what customers would have paid, then change what they see. I ran that transition at MateBio.

How quickly can you start?

For urgent situations (a technical co-founder left, an enterprise security review landed, a launch is slipping), I can start within 48 hours. For planned engagements, typically 1–2 weeks from our first call.

Fractional CTO for B2B SaaS