Anthony Eden

A mentor in Hawaii gave the rule early on: do not raise money unless one dollar can turn into five. Other people's money is not for building a product or buying equipment. Capital is for making more money.

A channel that returns five bucks for a dollar in is worth spending on until the channel is used up. Profit from customers is itself an investment in the work, and many cases call for a little money or no money at all.

DNSimple customers could not put a CNAME on the apex of a domain such as dnsimple.com. A Heroku customer wanted the naked domain to point at a Heroku site. The alias record started as a Perl script running inside PowerDNS, with a MySQL backend, and it worked until growth made it stop scaling.

Lua and another attempt did not fit. Anthony Eden learned Erlang to solve the scaling problem, using a library that already parsed packets, and built the name server DNSimple still runs in production. The language was learned to fix that problem inside DNSimple, not to start the company.

The niche was engineers and developers building web apps. They wanted to set domain records and leave, without hosting and without an email provider bundled in. The same approach was still the plan 10 years later.

GoDaddy was already a giant. Amazon Route 53, then Google, then Microsoft Azure each took plenty of business. The bet was that a developer niche still had room, because developers do not want to think about domain name system details.

The opening stretch was independent contracting, about a year and a half of software work to pay the bills. Living Social came next. The registrar was carved out in the contract as pre-existing work so it would not conflict with the day job. Nights and weekends went into growing it, and the first three years paid no salary at all.

A Florida limited liability company in 2010 was the legal start, using an entity that already existed. A paycheck began near the end of 2012, or maybe early 2013, once revenue could pay the other people on it. The person taking that salary was the third employee.

GoDaddy was holding a personal domain in 2010, and the domain name system screen was horrible. Constant upsells made the setup feel pointless for someone who had already worked in the domain space since the late 1990s.

The response was to stop using GoDaddy for that job and start a registrar instead. Years of working for other people had supplied the experience, and 2010 felt like the moment to take the work into his own hands.

Anthony Eden is the founder of DNSimple. Geek at Sea introduced him as the person behind a simple place to buy a domain and manage domain name system records in one spot, already about 10 years into the work.

DNSimple was started to fix a problem Anthony Eden had himself. The team stayed small, profitable, and independent, and the conversation was about building that kind of life while raising a family.

On the show, DNSimple changed billing so larger customers could pay by ACH. A credit card was the only payment option in the early days. Bigger companies refused credit cards because procurement and billing follow a structured process, and ACH became a feature only at higher tiers.

In the early days, the only way to pay DNSimple was a credit card. As the years went on, larger companies said a credit card would not work and they needed to pay by ACH. A $5 a month plan paid by ACH every month does not work, because the bookkeeping alone is a nightmare. DNSimple limits ACH to yearly plans that exceed some amount in a year. ACH stays off a yearly professional plan and stays on a yearly business

On the show, DNSimple never had a free tier. Anybody who used DNSimple had to pay, and fifteen years later a new user still cannot start without committing to putting a card on file.

The decision followed a startup already run on a free model for a long time. On the show, a heavy discount for certain people might be offered, or credit for an extended tryout, but everybody had to put a credit card in. Running the business meant investing time, with infrastructure that had to stay in place. When the product brings value and the customer sees the value, the customer will commit to putting a card on file. A free product only asks for time. An hour on the weekend to play is a different leap from

On the show, scalability is kind of irrelevant in the beginning. The point is to get somebody and form a kernel of a user base. Expanding the kernel by ten percent is enough, because what reaches one stage is different from what reaches the next stage. One approach will not run from zero all the way to a hundred million dollars.

A completely different strategy has to take over every couple of months or every year. What fits a move from a hundred customers to a thousand does not fit a move from zero customers to ten, or from ten customers to a hundred. Every added zero, whether customers, revenue, or traffic, brings very different requirements for the next step and for what comes after getti

A sale into an organization depends on a champion inside the business, even when the champion will not write the check. The champion has to be clearly identified, and the seller has to build a relationship with the champion. Champions are the ones who get the seller in with the gatekeepers and the buyers. Without champions, the seller pretty much has to have a much bigger marketing budget, because reaching the buyers and the gatekeepers takes a massive amount of marketing and sales.

The path is to find the champion, know who the champion is, and treat the champion well. Champions then take the seller into the businesses where the champions work and stay with the seller for a long time, even

On the show, DNSimple has used the Shape Up process from Basecamp for a couple of years. Framing, added to Shape Up after the original book, means writing the problem in a document and iterating until the understanding feels solid enough to design a feature. The team lacked that step at first, and adding it made the workflow make sense.

DNSimple is staffed almost entirely by engineers working at the product level, so the group had to retrain. People stopped themselves from writing solutions and focused on the problem, the people impacted, and the world people imagine after the problem is fixed. Only then did shaping begin, with the feature designed from that understanding.

A frame that is

DNSimple bound each user to an account at the start. Login credentials were tied to billing and access control. Over the years the binding blocked new work, so DNSimple decoupled users from accounts into a many-to-many relationship. Many users can now belong to many accounts.

People use DNSimple for personal domains. The founder has domain names and also works on DNSimple domain names. DNSimple segments domains for set reasons: production domains sit in one highly restricted account, and other domains are experiments for fun. The founder says DNSimple might build a site or run something off an experimental domain without knowing whether it will live. A single user reaches each account under

On IndieRails, Anthony Eden says the story that got Anthony Eden into .mp began on New Year's Day in January 2004, when somebody in Hawaii reached out. Eden cannot remember whether the contact was an email or a phone call, and Eden thinks it might have been email. The caller had a job application somebody had built, then encrypted all of the code, took the key, and tried to extort money to give the code back.

The caller asked Eden to reverse engineer the application and recover the code. Eden said encrypted code like that is pretty much lost, then asked for a couple of weeks to look. Two weeks later Eden reported that the code could not be broken into, so Eden rewrote the job application. T

Anthony Eden says optionality means you prepare and you open up so you are ready. If you do too much to fill all of your time, you have no optionality. You get optionality by leaving space open so you can explore something else, whether inside the business or outside the business, at work, at home, or in hobbies.

The idea landed when he read a book and realized he likes optionality. He says you have to leave moments for inspiration to come and go, then follow an idea and see what happens. You only do that if you actually plan for the open time, or if you get really lucky.

The open time is the plan. Leaving the space is how he stays ready for a direction in the business, at home, or in a ho

Post-incident reviews were institutionalized so every incident follows one set process. Even internal incidents still get a post-incident review. The only way to get better is to record what happened in a blameless post-incident report and state that the team will take action.

The reports are never perfect. Some post-incident reports were not resolved in a satisfactory way, and the shortfall became a lesson while the process kept evolving. No company, person, or entity is ever going to be perfect. Admitting incidents happen is the first step to getting better, and a lot of companies do not want to admit them. The stigma is a fear of lawsuits. Anthony Eden does not think a suit happens as of

Anthony Eden says a domain name system business only runs at any sort of scale when the operator stays willing to continue evolving over time. The first infrastructure could not scale at all, and it did not look anything like the system in place later.

The first rollout was four Unicast servers running on VPSs. The servers sat at Linode and at a small orange, really tiny companies doing VPSs, and the setup was terrible. Because the design was Unicast, if one server went down everybody was impacted. It was a hot mess, but the team had to start somewhere.

Eden says any infrastructure company has to be ready for the long haul and willing to continuously evolve. The standard, ideally, is that

On IndieRails, a name server in Australia could take almost an hour to reload because it connected to the core database and pulled all the zone data, while the same reload in Chicago took four minutes. Replicating zone data to each edge brought a full reload down to four minutes, and the edges now finish together.

Stock tools were tried early and did not work well across a large spread of systems or at the needed speed. The design still running today keeps the core application in Rails with a Postgres database behind it. A zone change sends a notification through a small Go application that lives at each edge, and the Go application fans that notification out to the name servers, which are

Anthony Eden passed the technical accreditation test to become a domain registrar in seven days in 1999, during deregulation of the dot com namespace, and became part of one of the first seven registrars accredited to sell dot com names.

Network Solutions had managed the entirety of the dot com namespace until deregulation in 1999. Eden was in France, planning to come back stateside and looking for work, when someone who had already finished the administrative part gave him seven days to accomplish the technical parts of becoming a domain registrar. Writing in Java, he built a Swing-based Java interface that ran Jython, a version of Python that runs in Java, with a text editor inside the in

On IndieRails, DNSimple made a conscious decision to do a set of things and do them well. A few months spent building a knowledge base then showed why DNSimple stays focused.

DNSimple strayed from the plan. Its CTO and the person telling the story decided knowledge bases were poor and spent a few months building a knowledge base. The side effort was a complete distraction and a very bad idea, and the experience quickly pulled DNSimple back to the reason for staying focused.

DNSimple still releases new products. Work from last year and the year before produced the domain control plane, a concept for connecting to other DNS providers and other domain providers and seeing all of them from wit

DNSimple is small enough that the company has not applied machine learning to all the data coming through. The generative part is not so interesting. One panelist says learning from the amount of data could be interesting. The work is feasible, and the team is looking at ways to do the job better without putting things at risk. An objective for the year is to figure out the place of generative AI at DNSimple, if any place exists other than internal use. Both panelists say the answer could be no, and one adds that more capable models probably fit in somewhere.

DNSimple added an export of domain lists as a CSV file. Rule sets for filtering would otherwise need an interface for every field or