Why Fyxer Refused to Launch Until Their AI Could Beat 10 Human Assistants
Most AI products launch when the technology is "good enough." Fyxer AI launched when their AI could beat 10 human executive assistants at inbox organization in
Agency to AI SaaS: Scaling from $1M to $18M ARR in 9 Months
Introduction and the "Unreasonable Effort" mantra
0:00Loading…
Like this episode?
Get real founder strategies for the AI era. Delivered weekly.
Free weekly newsletter · No spam
Richard Hollingsworth spent six years building the UK's largest executive assistant agency—then used every log, workflow, and customer insight to launch an AI SaaS that went from $1M to $18M ARR in nine months.
In this episode, Richard reveals the unfair advantage that service businesses have when building AI SaaS products, how a single Facebook ad signup turned into a $1.2M enterprise deal closed in 7 days, and why targeting professional services (not tech bros) unlocked explosive growth.
Richard Hollingsworth is the Co-founder and CEO of Fyxer, an AI-powered email assistant that predicts and drafts emails for busy professionals.
Richard and his brother Archie grew up on a farm, but they knew the slow pace of agricultural life wasn't for them. They saw tech as the opposite environment—fast feedback loops, results within your control.
They started by building the UK's largest executive assistant agency, bootstrapping it to $5M in revenue. But from day one, they had a bigger vision: turning the service into software. For years, they tried to build "tech-enabled" solutions, but nothing worked to pull the price down enough for the mass market.
Then GPT-3 launched. It was the breakthrough they'd been waiting for.
Unlike other AI SaaS startups starting from scratch, Fyxer had a secret weapon: six years of detailed logs from human assistants. They knew exactly how an EA organizes an inbox because they had thousands of hours of data on it. They used this proprietary data to train their AI models, ensuring their product was more accurate than a generic LLM wrapper.
The growth was explosive. They started the year with $1M ARR and a team of four. Within 9 months, they hit $18M ARR. They moved to San Francisco, joined an AI residency, and shifted their focus from "Tech Bros" to "Professional Services"—real estate brokers, consultants, recruiters—people who actually drown in email.
One of their biggest wins came from a single signup via a Facebook ad. That user turned out to be the CEO of a massive real estate brokerage. Within 7 days, Richard's brother Archie flew to Seattle, met the CEO at his lake house, and closed a $1.2M deal to roll Fyxer out to 5,000 employees.
This episode is part of our First Customers series.
Key Insight
Fyxer grew from $1M to $18M ARR in 9 months by using 6 years of executive assistant agency data to train AI models that outperformed generic LLM wrappers—then combining PLG self-serve with aggressive founder-led sales to convert individual signups into enterprise deals.
Key Ideas
- Use service business data to train AI: 6 years of EA logs gave Fyxer ground truth that competitors lacked
- Test AI against humans before launch: they only shipped when the AI beat 10 assistants in accuracy
- Target industries that drown in email: professional services (real estate, consulting) pay more than tech workers
- Turn PLG signups into enterprise deals: rank users by draft volume, then call companies with 5+ employees
- Close deals in person: Archie flew to Seattle same-day and closed $1.2M in 7 days
Key Lessons
- 🎯 Service data is an AI SaaS moat: Fyxer's 6 years of EA logs gave them training data that AI-first startups couldn't replicate, making their product more accurate than generic LLM wrappers from day one.
- 🧪 Test AI against humans before shipping: They pitted 10 human assistants against their AI and only launched when the AI won on accuracy—giving them confidence the product would fly off the shelves.
- 💰 Target pain, not tech adoption: Professional services workers (real estate, consulting) drown in email and pay willingly. Tech workers tolerate email tools but don't have the same urgency or budget.
- 🚀 PLG + founder-led sales compounds growth: Individual signups via ads created a footprint in companies. Ranking users by draft volume and calling companies with 5+ employees turned $30/month users into $1.2M enterprise deals.
- ✈️ Show up in person to close big deals: When the CEO of a real estate brokerage wanted to negotiate over months, Archie flew to Seattle the next morning and closed a 5,000-seat deal in 7 days.
- 📉 Plan for success, not just failure: Growing 5x in 3 months broke customer support—response times went from 5 minutes to 5 hours. They hadn't planned for what happens when growth actually works.
- 🔥 Maintain intensity with a weekly question: "What will you put unreasonable effort into this week?" keeps the team focused on the most important goal, even as headcount grows from 4 to 40.
Playbooks from This Episode
Why Running a Service Business First Gave Fyxer Instant Product-Market Fit
Build fast. Ship fast.
Fyxer 5x'd Revenue in 3 Months and Nearly Lost Their Customers
Founders obsess over what happens if things don't work. What if the campaign flops?
How Fyxer Turned One Self-Serve Signup Into a $1.2M Enterprise Deal
Most SaaS founders treat PLG and enterprise sales as separate motions. You either have self-serve or you have a sales team.
Watch the Episode
Chapters
00:00Introduction and the "Unreasonable Effort" mantra
01:16What Fyxer does: AI email assistant
02:12Revenue and growth: $1M to $18M ARR in 9 months
03:39From farming to tech: Richard's background
04:29Building the UK's largest EA agency
06:34Day one vision: agency as AI training ground
07:36Logging every task to build the AI roadmap
09:12GPT-3 moment: the breakthrough they waited for
10:43Testing AI against 10 human assistants
11:21Recruiting a technical co-founder without being technical
14:12First million: team of 4 doing nothing but building and selling
15:59LinkedIn influencer strategy for early growth
18:43Why they left the UK for San Francisco
20:22HF0 accelerator: 8x revenue in 4 months
22:49Starting marketing after $1M ARR
24:23Targeting professional services, not tech
27:00The $1.2M deal closed in 7 days
32:35Land and expand: PLG to enterprise sales
34:46Support collapse: growing from $1M to $5M in 3 months
36:33Hiring for intensity: maintaining startup culture at 40 people
38:24Customer story: Fyxer saved his marriage
40:33Lightning round
Episode Q&A
How did Fyxer use 6 years of agency data to build their AI SaaS product?
From day one of the EA agency, Richard required assistants to log and describe every task they did. This created a complete map of what people actually use assistants for, which they used to train their AI models with higher accuracy than generic LLM wrappers.
Why did Fyxer wait for GPT-3 before launching their AI SaaS?
They spent four years trying to build "tech-enabled" service products, but the technology couldn't lower costs enough for the mass market. GPT-3 finally made it possible to automate core workflows at 99% lower cost than human assistants.
How did Richard Hollingsworth pitch a technical co-founder without being technical himself?
Richard showed Matt (their CTO) three things: six years of proprietary training data, an existing customer base willing to pay $60/hour, and a proven sales motion. This made Fyxer a "go-to-market bet" rather than just an idea.
How did Fyxer close a $1.2M deal in just 7 days?
The CEO of EXP Realty (80,000 agents) signed up via Facebook ad and had 40 team members using the product within two weeks. Archie flew to Seattle the next morning, met the CEO at his lake house, and closed a 5,000-seat deal that same week.
Why does Fyxer target professional services instead of tech companies?
Real estate brokers, recruiters, and consultants drown in email—more meetings equals more money for them. Tech workers tolerate email but don't have the same pain point or willingness to pay.
What is Fyxer's "land and expand" strategy for AI SaaS growth?
95% of revenue comes from individual signups using work emails. They rank users by draft email volume, identify companies with 5+ employees using the product, then call to convert individual users into enterprise licenses.
What broke when Fyxer grew from $1M to $5M ARR in 3 months?
Customer support collapsed. Response times went from 5 minutes to 5 hours with only two support staff. They pulled everyone (including founders) into support and aggressively hired to fix the damage.
How did joining the HF0 accelerator change Fyxer's trajectory?
Living in San Francisco with other intense founders shifted their mindset from "What if Google builds this?" to "If this works, it will be huge." During their 4-month stay, they 8x'd revenue from ~$250K to $1M ARR.
What is the "unreasonable effort" mantra Richard uses to maintain intensity while scaling?
Every team member must answer: "What is the one thing you will put unreasonable effort into this week to move our most important goal?" This keeps the startup hustle alive even at 40+ employees.
Book Recommendations
More on First Customers
Nate Baker, Qualia
How to Find Your First Customers by Living in Their Basement
Nate Baker is the co-founder and CEO of Qualia, a software platform for title companies that helps coordinate the complex process of buying a home. Today, Qualia generates over $100 million in ARR with a team of 600 and has raised more than $200 million. In 2015, Nate was 21 years old and decided to build software for the real estate industry. He had no experience in that space. He didn't talk to any customers. He just did some research and decided that was the thing he was going to do. Then he started building. Still without talking to anyone. Nate admits this was a mistake. He and his co-founders got key things wrong about how the business would work. They wasted months building things they eventually threw away. It wasn't until they found their first customer that they started making real progress. Their first customer was Barry Feingold, a state senator in Massachusetts who also ran a real estate law firm. Barry believed in the vision, taught them the industry, made introductions, and helped them understand what actually mattered. The relationship was unconventional: Nate and the first 25 employees rotated through living in Barry's basement. New hires would get a call Sunday night: "Your onboarding is in Andover. You're going to live in Barry's basement for two weeks. He's going to teach you title. You have to tutor his kids in math." But then Barry's existing software vendor found out he was working with Qualia and shut off his access overnight. Nate and his team didn't even have the core features built yet. They had to figure it out fast. It became the most productive month in company history. Barry didn't just become a customer—he introduced Qualia to his competitors. Those network-based relationships became the foundation for the first 10 customers. Nate learned that your first customers must come from your network, not cold outreach. In this episode, you'll learn: 1. How to find your first customers through network-based selling instead of cold outreach 2. The multi-year upfront contract strategy that brings cash forward and locks in commitment 3. Why geographic focus beats national expansion in the early days 4. How to embed yourself with customers to truly understand the problem 5. When to hire your first sales leader and how fast to scale
Stefan Bader, Cello
First Customers From a Notion Page and Python Scripts
Stefan Bader is the co-founder and CEO of Cello, an all-in-one referral platform that helps SaaS companies reward their users for bringing in new customers. Back when Stefan was Chief Revenue Officer at a payment processing company, he noticed something odd. His users were bringing in tons of new customers, but there was no way to reward them. Every tool out there was for affiliates and influencers. Stefan saw the opportunity. And in 2022, he quit his job and started Cello. But building it turned out to be way harder than he'd imagined. Paying individual users meant navigating compliance laws in dozens of countries, international banking regulations, and tax requirements that no one had mapped out before. And his MVP was embarrassingly basic. Customers got their analytics through shared Notion pages. No login portal. No dashboard. Just Stefan's team running Python scripts and configuring everything manually behind the scenes. "This is not a product," one early customer told him. But Stefan stayed focused on what mattered - could it actually generate referrals? The answer was yes. He also made pricing dead simple - pay nothing until you make money from referrals. However, most SaaS companies didn't even know they needed this product. Stefan had to educate every prospect about why user referrals were different from affiliate programs, and why they were leaving money on the table. As early customers started seeing results, word spread. But the real breakthrough came from something tiny - a "powered by Cello" link in their widget. As customers grew, millions of their users saw it. That little link became their biggest growth driver. Today, Cello generates $2.5M in ARR, powers referral programs for companies like Miro and Typeform, and reaches over 7 million users each month.
Jonathan Festejo, Salesbricks
First Customers: Why Selling to Enterprise Almost Killed This SaaS
Jonathan Festejo is the CEO and co-founder of Salesbricks, a deal-closing platform that handles quoting, billing, and subscriptions for SaaS companies. Before Salesbricks, Jonathan ran RevOps for multiple unicorns. He knew the pain of "Quote-to-Cash" intimately. But when he started his own company, he made a classic mistake: he built for the Enterprise first. He spent nearly two years over-engineering a robust platform designed to replace Salesforce CPQ. But when he tried to sell it, he hit a wall. Enterprise buyers had 3-month sales cycles, complex approval chains, and often decided to "do nothing" because switching was too risky. The breakthrough came when he pivoted down-market. He realized that early-stage founders ($500K-$2M ARR) were the ones with the most urgent pain. They were hacking together Stripe and spreadsheets late at night and were desperate for a solution. Jonathan used these 3 strategies to find his first customers and reach $1M ARR: 1. Targeted founders who were personally doing billing at night - they felt the pain and could make instant buying decisions 2. Built a "Powered By Salesbricks" button into every contract sent, turning each transaction into a high-intent lead channel 3. Leveraged founder-to-founder referrals from his network, converting warm intros into paying customers By shifting his focus to founders, sales cycles dropped to 5 days. The product roadmap started writing itself based on real user feedback, and Salesbricks hit $1M ARR with over 100 customers.