Written by James BrittTechnically reviewed by Jim Freeze · Reviewed
A Rails 5.2 app comes up for review and somebody suggests a rewrite. Before I’d sign off on that, I’d pay for a day with the old code. Can the new team get it running? Can they run a checkout and tell you which dependency would break on the way to Rails 6? A day spent finding out is cheap compared with rediscovering your billing rules six months into a new build.
I’d call different firms depending on what that day turns up. LITSLINK is worth a call when the Rails work is one piece of a mobile launch. If the job is getting a neglected app onto a supported release, I’d start with FastRuby.io. For someone to look after it month to month, compare Planet Argon and reinteractive. The rest of the shortlist gives you more room on design, training and the hours your teams can work together.
What Rails 8 changes for a buyer
Rails 8.0 shipped on November 7, 2024, and the release notes on rubyonrails.org describe a framework that now deploys itself: Kamal 2 turns a blank Linux box into a production host, Thruster sits in front of Puma so Nginx is optional, and the Solid trio (Solid Queue, Solid Cache, Solid Cable) replaces Redis for most apps.
That last part is worth discussing in a proposal. A Rails app with a small job queue can run happily without another service to patch. A busy app with a queue that already works might sensibly keep it. Ask the agency to draw the setup it would actually put into production and point to where the backups live.
Per the Rails 8.1 announcement, more than 500 contributors landed 2,500 commits, Shopify and HEY ran it in production for months before release, and HEY’s 30,000-assertion suite dropped from over ten minutes in cloud CI to 1 minute 23 seconds on a Linux desktop with the new local CI.
Shopify’s compiler team reported in January 2025 that YJIT 3.4 runs about 92% faster than the CRuby interpreter on its headline benchmarks, uses slightly less memory than YJIT 3.3, and carried Shopify’s storefront through Black Friday 2024 at more than 80 million requests per minute on a prerelease build.
Ruby 3.4 also made Prism the default parser. For a proposal built around speed, bring an example from your app. A dashboard that spends most of its time waiting for PostgreSQL is a good place to start. Get a before reading for that request so you can tell whether a Ruby change helps.
Find the upgrade debt first
Say a billing gem hasn’t had a release in four years. You can replace the gem in an afternoon and still spend a fortnight checking old invoices. There’s a similar problem with assets. When the only person who knew the Webpacker build left two years ago, the first task can be getting a clean build at all.
The Rails upgrade guide recommends stepping through minor releases and dealing with the deprecations along the way. For an old app I’d ask the agency to draw that route with Ruby versions pencilled in next to Rails versions. The first thing I want to see is where that chart gets messy. If the interpreter needs a move of its own, use the Ruby 3.4 to 4.0 migration checklist to scope that piece separately.
How I compared the companies
A recent case study is handy, especially if it tells you how the system was built. I also looked for things a buyer can pick up and inspect before a call. FastRuby.io has public upgrade guides. thoughtbot has testing gems. Visuality has workshops where you can see how the team approaches Hotwire. Those clues have more weight here than the size of a client logo wall.
Seven Ruby on Rails development companies
1. LITSLINK
LITSLINK’s Ruby on Rails development services cover startup MVPs, Rails backends and APIs for iOS and Android apps, ecommerce builds, and consulting, with US-hours project management from Palo Alto and senior engineering teams in Europe.
Take MySquire, one of its published projects. A traveller needed a mobile app, while the guides needed a web platform. LITSLINK built both pieces. In a different case study, for Lifepoint Clinic, the company describes setting up a deployment pipeline for an existing Ruby app across development, staging and production. These are useful examples if you’re buying more than development hours.
On the call I’d ask who owns the API when the mobile and web release dates start to diverge. That answer matters after the launch, when an old app version can still be sitting on someone’s phone.
2. thoughtbot
If you’ve used FactoryBot to set up test records or Shoulda Matchers to check a Rails association, you’ve seen some of thoughtbot’s work. Its Rails maintenance service is available part time. That gives a smaller team a way to get help with an upgrade and keep its own people building features.
I would put thoughtbot on a shortlist when the handoff is important. Pick a task your team finds hard today and ask how the firm’s developers would work through it with yours. A pairing session or a real pull request gives you something concrete to judge.
3. FastRuby.io (OmbuLabs)
Start here when the Rails upgrade is the job. FastRuby.io calls its code and dependency assessment The Roadmap. It publishes a guide for the jump from Rails 8.0 to 8.1, alongside tooling such as next_rails. You can look through some of its process before asking for an estimate.
Bring the lockfile. Ask which dependency they’d try first, where they might need a fork and how the upgrade branch stays in step with regular feature work. If you’ve ever spent a Monday untangling the results of two branches changing the same models, you’ll know why that last part belongs in the estimate.
4. Planet Argon
The Portland firm has a Rails upgrade assessment and ongoing support. It also runs the Ruby on Rails Community Survey, which published its 2026 results. The survey asks about the actual setups people run, including their Rails versions and deployment choices.
Planet Argon is a natural call when an app already pays its way and nobody on staff has enough time to look after it. Use a real ticket to test the service. A background job has failed for the third time this month. Who opens it, how quickly, and where will the eventual fix go in the queue?
5. Hashrocket
Hashrocket has long worked in pairs. Its Today I Learned archive is full of small Ruby and PostgreSQL discoveries from the work itself. The company also describes helping Regions Bank refresh a legacy application and training people on the client’s team.
That’s the kind of engagement I’d want if my own engineers were staying with the code. Ask how the pairs are staffed. An outside pair can ship a lot of work, but the plan is different if one seat needs to belong to a member of your team.
6. Visuality
Visuality is based in Warsaw. One of its workshops builds a Kanban board with Rails and Hotwire, including live updates with Turbo Streams. The firm also teaches Kamal deployment. The workshop material makes a good starting point for a discussion about an app with a tangled front end or a deploy nobody likes touching.
For a long-running engagement, get the calendar out early. A UK team and a US West Coast team will have very different windows for pairing with Poland. Work out who will handle a production incident near the end of your own day.
7. reinteractive
Mikel Lindsaar, who founded reinteractive, wrote the Mail gem that Action Mailer uses. The Australian firm splits the work into services. App Review examines the application, CodeCare handles ongoing maintenance, and OpsCare looks after the server layer. You can ask for the piece you need.
This belongs on the shortlist if an app needs both regular changes and someone to look after production. Ask for the response hours in your time zone. I’d also want to know when someone last tested a restore and who would run one for my app.
How the companies compare
| Company | Start here when | First thing to ask |
|---|---|---|
| LITSLINK | The same release needs a Rails API and a mobile app | Who maintains compatibility with older mobile clients? |
| thoughtbot | Your developers will keep working on the product | What work will the teams do together? |
| FastRuby.io | Several Rails releases stand between you and your target | Which gem makes the first step risky? |
| Planet Argon | An existing app needs regular Rails support | How does the monthly plan handle an urgent ticket? |
| Hashrocket | You want hands-on help and training | Can my engineers take part in the pairing? |
| Visuality | You want a Rails team in Europe | Which hours are shared by both teams? |
| reinteractive | You need maintenance and production operations | Who is on call when our business day starts? |
Five questions before signing
- What versions have you shipped lately? A redacted Gemfile.lock or a public repo is enough to get the conversation going. Ask why those versions were chosen and how the team tracks security releases.
- Where would you run our app? Compare Kamal on a VM with a managed platform. Price the work involved in backups, patching and a rollback alongside the hosting bill.
- What does the first upgrade step cost? Hand over the current Rails and Ruby versions and the awkward parts of the Gemfile. Ask for a range and for the test or deployment that would tell you the step is done.
- How would you build this screen? Put a real screen in front of the team. Talk through Hotwire and React in that context. On a form with a few live fields, the answer can be quite different from one for an offline editor.
- Who answers after launch? Get the retainer and escalation terms in writing. Ask about the last security patch they handled for a live app and who made the call to deploy it.
Given two plausible firms, I’d buy the same small discovery job from each before committing to a long engagement. Give them the same failing build or the same sketch of a new feature. A week later you’ll have something to compare beyond the hourly rate.
