Rails

ThinkingNovember 2026

Rails sells a managed agents department first

Rails starts as a fractional managed agents department: a free consult, a custom build, then a retainer tied to goals.

Most companies want agents doing real work, and almost none have anyone to run them. So Rails starts by selling the people and the platform together, as a fractional managed agents department. A company gets a free consult, then an implementation that builds its agents and the custom software around them, then a retainer tied to goals we agree on up front.

That is the first step on a longer path, from the tools people use today to agents for everyone.

Toolschat apps andcopilots. Aperson does thework with help.Agentsone agent doesone job.A company'sagentdepartmentmany agents, runfor the companyin one place.Marketplaceproven agents getlisted, andanyone can hireone.Agents foreveryoneeach persondirects agents oftheir own.Rails starts herepeopleproductpeople now, product later

Most people will hire an agent instead of building one

The buying decision has moved again. Companies used to hire a person for a job. Then software made them ask whether a tool could do it. Now the choice is a person, a tool or an agent, and most people will hire an agent that already does the job instead of making their own.

That changes what a company wants to buy. Few teams want to pick model vendors and keep them working while the parts underneath change every few weeks. They want one place that handles it and a result they can count.

Companies already spend most of that money on services. Sequoia counts about $6 spent on services for every $1 spent on software (Sequoia, March 2026). Rails sells into the services side first.

Software: $1Services: $6Rails sells here firstFor every $1 spent on software, about $6 is spent on services.Sequoia, “Services: The New Software,” Julien Bek, March 5, 2026.

The service runs in three steps: consult, build, retainer

The consult is free. We sit with a team, find the problems it most wants solved and pick the jobs worth handing to agents first, starting with the lowest hanging fruit.

Then comes the implementation, paid once. We build the agents and the software each job needs, interface included, on the best models, vendors and open-source agent tools available. Each build starts from parts Rails already runs, so it takes weeks.

Then a fractional retainer. A head of agents stays at the company's side, the way companies already hire a fractional CFO or CTO for work they can't staff full time. Rails runs the agents, swaps in better parts as they ship and adds the next agent when the team is ready. The retainer is tied to goals we set together at the start, like time to hire or pipeline.

Buyers already know this shape. Compliance software sells the same way: an implementation up front, then an ongoing platform fee.

  1. 01Free consultFind the problems a team most wants solved and the jobs worth handing over.
  2. 02Implementation, paid onceAgents plus the custom software and interface each job needs, built in weeks.
  3. 03Fractional retainer, tied to goalsA head of agents at the company's side; Rails runs the agents and swaps in better parts. Example goals: time to hire, pipeline.the next agent

Your own model keys and subscriptions, or lab runtimes at cost, under a cap you set.

Rails is the IT department for agents

Underneath, Rails is an AI transformation business, packaged as a managed platform that Rails operates. The closest frame is the IT department. IT never built its own computers. It picked the vendors and handed out the machines, and each person made theirs their own.

Rails does that job for agents. It provides the agents and software a team works in, the way IT hands out laptops, and swaps the parts underneath when better ones ship. A company's memory, playbooks and pages stay its own, and they work in whichever app its people already use, whether that's Claude, ChatGPT or Grok.

Companies bring their own models

Rails never requires its own models. A company brings its own model keys and connects the subscriptions it already pays for, and much of Rails is built around that. Rails also runs agents on the labs' hosted runtimes and passes that runtime through at cost, under a monthly cap the company sets.

That is why the retainer follows goals instead of credits or tokens. Per-token billing is hard to budget, and a company pays it whether or not the job gets done.

Rails takes the past year's headaches off the table

Rails gets a team the most out of AI without the problems the last year of adoption made familiar.

Newly funded, non-technical teams buy first

The first buyers are teams that recently raised or expanded and aren't technical: the sales, operations or recruiting side of a company. They have new budget, pressure to show results fast and no engineers to build with, so a managed department is easier to buy than a platform to learn. Even at tech companies, there's a wide gap in how to give people this kind of power.

The smaller the company, the less budget it has to try something before it sees proof. New funding closes that gap, and the first clients come through my own connections.

We start with go-to-market and recruiting because their outcomes are the easiest to measure: time to hire or pipeline, before and after. That number is what renews a retainer. My own recruiting agent is the first proof, measured by the outcome a recruiting team already counts. Each deployment refines the building blocks for its vertical, and we add verticals as deployments teach us.

People now, product later

A deployment company adds people for every new client. Rails adds clients for every person, because each build becomes a building block the next one starts from. The team starts at about three engineers for each seller, since clients need a lot of build up front, and leans toward the platform as the building blocks converge.

For a while, technical people will set up the workspace for everyone else. Managed comes first because most people won't build agents themselves. Over time the platform carries more of the work, until most people can shape their own agents, and self-serve is how it scales.

new for this joba part Rails already runsclient 1client 2client 3

Further out: a marketplace, then agents for everyone

Two businesses grow out of the departments once they run. One is a marketplace where proven agents get listed and anyone can hire one. The departments supply both sides from the start: their companies buy, and the agents Rails runs for them are the first listings. The other is a roll-up that buys small agencies already doing this work, to learn their businesses from the inside.

Past that is the reason I'm building Rails. Coding agents closed the engineering gap I had early on and let me build things I couldn't have built alone. Rails gives everyone else that same lift in their own work, and every time a model gets better, every agent on Rails gets better with it.