One employee, and an operating layer that runs the rest.
Everlight is my own products company. I am the only person in it. Rather than hire, or accept that admin would eat the week, I built the company an operating system and then ran the business on top of it.
The problem I set myself.
A one-person company fails in a predictable way. Not from a shortage of ideas. From the hour a day that goes to remembering what needs attention, chasing what got dropped, and rebuilding the same document for the fourth time.
So I stopped treating that hour as the cost of doing business. The obvious fix is to hire it away, and I could not afford to, so I engineered it away instead. Everything below came out of that decision, and I use all of it every week.
The morning briefing.
Before I open anything else, I get a briefing. Not a checklist I wrote once and stopped reading. It reads the current state of the business and tells me what needs attention today, in the order it matters.
The hard part was never generating text. It was grounding. A briefing that invents a priority is worse than no briefing, because you stop trusting the next one.
Ordered by what changed, not by what is oldest.
Every line traces back to a record, so nothing is inferred.
It arrives before I ask. A tool you have to remember to open is a tool you stop using.
The advisory panel.
When I have a decision I would normally take to a board I do not have, I put it in front of a panel instead. Several perspectives, argued separately, before I commit.
The value is not the answer. It is being made to state the decision clearly enough to be argued with. Half the time I change my mind while writing the question.
Each perspective answers independently, so they disagree in useful ways.
It reads the same business context the briefing does, so the advice is grounded rather than generic.
The operations dashboard.
The health of the business on one screen, updating without me. What is moving, what has stalled, and what is quietly overdue.
Most dashboards fail by showing everything. This one is built around a smaller question: if I only looked at this once a week, what would I need to see?
Built around the weekly glance, not the daily stare.
Stalled items surface themselves rather than waiting to be searched for.
The client database.
Contacts, the work in flight, and what happens next, in one place, shaped like a small business actually works.
Every small firm I have watched buy an enterprise tool ends up using a sliver of it and maintaining the rest. I would rather build the sliver and have it fit.
One record holds the person, the work, and the next step.
Nothing in it exists because a bigger tool had the field.
The document system.
Proposals, reports and client-facing documents generated from a template that holds the brand, in seconds, identical every time.
This is the unglamorous end of the work, and it is where most of the quiet damage happens. Documents are where a small company's brand falls apart: the logo drifts, the spacing changes, someone reuses last quarter's file and forgets to change the name.
The brand lives in the template, not in whoever is typing.
Seconds rather than hours, and identical every time.
What this proves
- I do not just specify systems. I build them and then live on them.
- The interesting problem in AI work is grounding, not generation. Everything here reports what it can see.
- A tool nobody opens is a tool that failed. Each of these arrives or updates without being asked.
- One person can carry the workload of a team if the systems underneath are right, and building those systems costs less than hiring for them.
Tell me what you are trying to finish.
I would rather build the thing than write a deck about the thing.
Send me the project that has been sitting there. I will tell you what I would do with it and whether I am the right person to do it.