I run Echelon Advising LLC. When an established company decides it wants AI running inside its operations, there are four ways to get it built. I have built these systems inside a multibillion-dollar Los Angeles venture firm, and I now run a firm that installs them for other companies. Here is how I would choose, including the cases where I would not choose us.
What are the four ways to get AI built?
Hire in-house. Bring in a big firm. Buy tools and wire them together yourself. Or work with an implementation partner who maps the operation, builds into your stack, launches with approvals in place, and stays on to run it. Each one is right for a specific company at a specific moment. Most of the failures I see come from choosing the wrong route for the moment, not from the technology.
When does hiring in-house win?
Hiring wins when AI is the product, or when the volume of systems you will build is large enough to keep a senior engineer busy for years. If you are a software company, build the team and do not look back.
If you run an established company where the systems serve operations, one hire means one person carries mapping, building, connecting, approval design, and running it, with nobody to review the design and nobody to cover when they leave. The fully loaded cost is the smaller issue. The single point of failure is the larger one, and it shows up the week that person resigns and the system nobody else understands keeps sending. I wrote the full comparison at Echelon versus hiring in-house.
When does a big firm win?
A big firm wins when the engagement is mostly about governance: a regulated rollout across many business units, a board that needs a named brand on the deck, a procurement process that requires one. Those are real needs, and a big firm is built to meet them.
The trade is speed and proximity. The people who scope the work are rarely the people who build it, the price is built up from hours, and the result tends to arrive as a program rather than as something running in your inbox on a Tuesday. If you need the program, take the program. If you need the Tuesday, do not.
When do the tools win?
Tools win when one small tool already solves it. A scheduling link, a form that writes to the CRM, an auto-reply with the right hours in it. If that is your whole problem, buy the tool and stop. On the mapping call, when a tool solves it, we tell you, and we build only what sits between the tools.
Tools stop winning at the seams. The lead arrives by phone and by email and by text, and each one lands in a different place. The follow-up depends on what the quote said. The one thing nobody owns is the handoff between the two people who each think the other has it. That gap is not a tool. It is a layer: one record per customer, shared context, and an approval above each action.
What does an implementation partner do differently?
- 01Maps before buildingWe sit with the people doing the work and follow leads, quotes, money, and messages through the tools that carry them today. Scope and price come out of the map, in writing.
- 02Builds into the stack you runNothing ripped out, nothing migrated. Read and write both ways, every connection verified before anything depends on it.
- 03Launches with approvals in placeReview mode first, so your team approves every result before it goes out. Narrow authority is earned one action and one channel at a time, and never for moving money.
- 04Stays on and runs itA named owner on your side, a weekly report, and the next system built on the same layer. Day 90 is when the operating responsibility begins, not when it ends.
The honest trade: you are not building an internal capability, and you are not buying a brand. You are buying a system in production, fast, with one accountable person and one set price in writing. The partner is the right answer when the system matters more than the org chart.
How do I choose in one pass?
Three questions, in order. Most companies are done by the second one.
- 01Is AI your product or your operations?Product: hire, and build the team. Operations: keep reading.
- 02Does one tool solve the whole problem, seams included?Yes: buy it and stop. No: the gap between the tools is the work, and it needs a layer, not another tool.
- 03Do you need a program or a system?A program across many business units with governance at the centre: a big firm. A system running in your inbox next month, with one accountable person: a partner.
Where we start instead
Some companies are not ready for any of the four routes on the first call. That is a starting point, not a verdict, and we point the same candour at what we can build.
- If your process still changes every week, we start by making it repeat.
- If the bottleneck is getting customers, we start with the system that answers and follows up with every lead.
- If nobody inside owns the work yet, we name the owner in the map.
- If one small tool already solves it, we tell you, and build what sits between the tools.
What we cannot yet claim
Echelon Advising LLC is a young company, founded in February 2026. The number of installs under our own name is small, and the production numbers on this site come from systems this team shipped and runs, some of them inside enterprise careers before Echelon existed. I am not going to claim a win rate against the other three routes, because I do not have the sample to support one.
What I can claim is the shape of the engagement, because it is written into every agreement: a map first, one set price, approvals before autonomy, and a weekly report from launch on. If that is the shape you need, bring one process to the mapping call and we will map it together. If another route fits better, we tell you on the call.