Build an ugly MVP and watch your startup win

6 min. read
Smiling man in a black shirt with looping black ribbon displaying "DELIVERED" in white text encircling him. Gray background.
Podcast

Taylor Halliday, Co-Founder & CEO at Ravenna.

Taylor spent his early career as what he calls a professional nerd, then started several companies with his co-founder Kevin Coleman between 2015 and 2020, including a stint at Y Combinator and a cloud consultancy that grew to around 20 people. He spent about four and a half years at Zapier running new products and AI engineering before co-founding Ravenna, where he is CEO.

Today he runs Ravenna, an AI-enabled help desk for internal support that works natively inside Slack and Microsoft Teams. Its customers are IT, HR and revenue operations teams at B2B SaaS companies with 100 to 1,500 employees. Ravenna has raised $15 million since launching in July 2024.

We got into why he ships the ugliest MVP he can get away with, how a LinkedIn post landed his first customer in Sydney, why better design alone won't make anyone switch software, and how a flat 20-person team keeps its speed.

Key takeaways#

Why should startups ship an ugly MVP instead of building in stealth?#

Because the longer you build without customers, the longer you can fool yourself. Taylor's rule is to get the ugliest, smallest thing out the door that gives a customer a modicum of value, then tighten the feedback loop from there. He knows friends who have built stealth startups for a long time, and he doesn't know how to build that way. His reasoning is blunt: none of us are that smart, and the market is the shrewdest mechanism on the planet. It doesn't care about you. It only cares whether you provide value. He knows Steve Jobs disappeared for years and came back with the iPhone. "Good for him. I don't know how to do that."

Try this: write down the smallest version of your product that one real customer would use this month. Ship that first.

How do you know when an MVP is ready for paying customers?#

When the people giving you feedback start running parts of their business on it. Ravenna started on 15 July 2024. The rest of that year was three jobs: build, talk to customers, recruit. Taylor's goal was to onboard users in January 2025. It crashed and burned. Design partners like Zapier and Homebase kept poking at what Taylor calls "the toy" and sending the team back with notes. Around March and April, some of those partners were using it for real. By May and June the team felt ready to sell to people they had never met. Taylor remembers the first demo where the buyer said "cool, we'll take it", and his reply: are you sure?

Try this: track how many design partners use your product without being asked. When that number moves, start selling to strangers.

How did Ravenna win its first customer?#

Through a LinkedIn post. While the team was deep in work with large design partners, Josh, the IT manager at AirTree, a venture capital firm in Sydney, reached out. Taylor assumed it was another pitch aimed at a founder. Josh had seen the vision, said the software in the space kind of sucks, and admitted Ravenna wasn't there yet. He became Ravenna's first customer anyway, because he wanted a breath of fresh air in the market.

Try this: post about the problem you're solving, in plain words, before the product is finished. The first buyer might be someone you've never heard of.

Is better design enough to make customers switch software?#

Taylor doesn't let himself believe it is. Incumbent help desk software is ugly, slow, and treated as a necessary evil. There is no Figma of this category. The builder in him would love people to rip out their tools because he built something more beautiful. He thinks you have to find something deeper. Ravenna doesn't lead with looks at all. It leads with pain points around automation, insights and analytics, and where the work happens. He does give Linear credit for doing well with prettier project management software.

Try this: write down why a customer would switch to you. If "it looks better" is the first reason, keep digging.

Why should internal support tools live inside Slack and Teams?#

Because that's where work happens now. Before writing any software, Taylor asked potential customers what would make them switch. The answer kept coming back to team chat. Incumbents do have Slack apps, but they're built to pull you out into their own application, with OAuth screens and paper cuts along the way. So people give up and DM the person they need. Taylor doesn't care much for the return-to-office debate. What he thinks has changed for good since the pandemic is that companies run on team chat, wherever people sit. Ravenna delivers the whole help desk inside Slack and Teams, so people get help where they already work.

Try this: count how many clicks it takes a colleague to get help from your internal tools. Then count how often they skip it and message someone directly.

When does a company need an internal help desk?#

Somewhere around 100 employees. Taylor describes a help desk as customer service where the customer is a colleague. Below about 100 people you just talk to each other. Above that, triage and coordination get hard. IT usually needs it first, then HR, and at scale engineering, marketing and revenue operations. Ravenna's sweet spot is B2B SaaS companies with 100 to 1,500 employees. The ceiling comes from enterprise systems like Workday and ADP, which need deeper integrations than the team has built so far. What fascinates Taylor is that the need tracks headcount and nothing else. It doesn't matter what industry you're in.

Try this: if you're past 100 people, ask your IT and HR leads how many requests arrive by DM. That's your hidden help desk.

What is a startup's real advantage over big competitors?#

Speed. A startup has every disadvantage: less money, fewer customers, no go-to-market machine. Taylor has been on the other side at Zapier and has seen that no amount of money makes a big company move fast. It isn't in its DNA. So he sees his job as building an environment where Ravenna can go as fast as possible. At 20 people the company is fully flat. No VPs. People have functional titles, but everyone locks arms, and he wants a strong sense of ownership, agency and urgency. He depends on the team as much as they depend on him. He still writes code, which his co-founder sometimes yells at him for. You have to keep it fun.

Try this: remove one layer of sign-off from your team's next decision. Watch how much faster it moves.

Should AI products use deterministic or non-deterministic automation?#

Both, and the mix is where the value sits. Taylor led AI engineering at Zapier, so he knows the current excitement around LLMs. His view is that many AI startups are throwing the baby out with the bathwater by betting everything on non-determinism. Deterministic workflows still do a lot of the work. At Ravenna, the moment that felt most meaningful was when customers could see, and measure, that the software made them more efficient. A larger help desk and prettier software are fine. Proof of efficiency is what he wants.

Try this: map one AI workflow in your product and mark which steps must give the same answer every time. Keep those deterministic.

What can founders learn from running a services business?#

That survival teaches range. Between 2015 and 2020 Taylor and his co-founder Kevin Coleman started several companies and went through Y Combinator together. One consultancy grew out of the ashes of a failed product company, when their backs were against the wall and money was running out. It focused on Kubernetes and cloud consulting, but they took on almost any technical problem, in some pretty interesting languages and code bases. It grew to about 20 people. Lots of lessons, lots of tears, lots of happiness.

Try this: if a product bet fails, ask what your team can sell tomorrow. Services can buy you the time to try again.

Listen & watch#

Learn More

Contact Us