The questions people actually ask us

At the start, you don’t, and anybody who hands you a percentage before looking at your business is making it up.

What we can tell you is how you’ll find out.

The diagnostic work happens before any build, it’s priced small, and the fee credits toward the build if you go ahead. So you learn what’s actually wrong and what fixing it would take before you commit to the fix.

Then the foundation stage gives you something most companies never have: a written picture of how the work happens today. That’s your before. Whatever paying off means for you, whether that’s inquiries answered the same day, quotes going out without waiting on one person, or hours coming back to somebody who’s drowning, you’ll have a real baseline to measure against instead of an impression.

And we agree on what we’re solving before anything gets built. Not "implement AI," but the specific outcome you’re after. That’s what the work gets held against.

That’s the deal we’re actually offering: not a promise about the return, but a way to see it.

Maybe not, and we’d rather establish that on a call than in a proposal.

The test is simple. Is there work that repeats, that people are doing by hand, that’s capping how much you can take on? If yes, there’s something here for you. If your volume is low, or your work is genuinely bespoke every time, or the constraint in your business is somewhere else entirely, then this isn’t your problem to solve and we’ll say so.

You probably should, and most of our clients have them. This isn’t an either-or, and we’d think less of a firm that told you otherwise.

They solve a different problem, though. A subscription gives each person a tool they have to drive. Nothing happens unless somebody is in the chair, and how much anybody gets out of it depends on how good they are at asking. It doesn’t know your pricing, your policies, or the four exceptions your business runs on, because all it knows is whatever got pasted into it that morning. And the work doesn’t complete. Somebody still has to take the answer and do something with it.

An AI employee is built on your written process, it works while nobody is watching, and it finishes the job: the inquiry answered, the quote prepared, the follow-up sent.

There’s a second difference, and it’s the reason this company exists. Every one of those chat windows is a place where a client’s personal information goes somewhere you can’t account for. Fiorella hit exactly that wall building systems for this business, couldn’t answer where the data had gone, and built the platform partly to fix it.

More than you’d expect in some places and less than you might want in one, so here’s both.

You control the decisions. You approve the written process before anything gets built. You approve what each AI employee can say, offer, and refuse before it speaks to anyone. You decide which actions wait for a person, and the AI employee operates inside those limits rather than around them. When we improve something, it reaches you as a proposal you look at, not as a change you notice later.

What you don’t do is edit the skills yourself. Changes come through us. That’s deliberate, because a workforce that anyone can adjust drifts, and drift is how these systems start saying things nobody approved. If what you want is a platform you configure yourself, we’re the wrong firm and there are good ones that do that.

The protection is built into how the system works, not into a policy document. Personal details are stripped out and replaced with placeholders before a message reaches an outside model, and the real values go back in inside your protected environment. The outside model works on the de-identified version.

How far that goes is tuned to your business. Off, when there’s nothing sensitive in play. Graduated as the volume and sensitivity rise. At the strictest setting, our own local models do the stripping, so sensitive values stay inside the protected environment.

This isn’t a roadmap item. It’s enforced in production right now for an insurance and investments brokerage, which is about as unforgiving a setting as this work gets.

And you don’t have to take our word for the path. Ask on the call and we’ll walk it with you, step by step, including where your own team still sees what it needs to see.

No. Your information is used to do your work, and that’s all. We don’t train or fine-tune models on it, and we don’t pool it with other clients’ data to make anything smarter.

The models underneath are provided by outside vendors under commercial terms that don’t permit training on customer data. And at the strictest tier of protection, sensitive values never reach those vendors at all, because our own local models strip them out before anything leaves your protected environment.

It’s your data and your business, and the engagement is built so that leaving isn’t a hostage situation.

Each phase produces something you own. The process documents from the foundation stage are yours whether you build anything or not. Backups exist because a system running a real business has to survive a bad day, which was one of the design requirements before the platform had a client on it. And the architecture isn’t welded to any single model or vendor, so you’re not tied to whichever tool happened to be best when we started.

When an engagement ends, what your business produced goes with you. The process documents, your own information, and the work your AI employees generated: the client insights, the analysis, the assets and design systems built for you. Then we remove your data from our systems.

The exact list and the mechanics of it belong in your agreement rather than on a web page, and they’ll be in writing before you sign anything.

Mostly you’ll use it to see. What happened, what’s waiting on you, and what got handed to a person. Each department gets its own view you can walk into.

Some departments are more hands-on than others. With marketing and brand, you’re working with the AI employees directly, asking for things and reacting to what comes back. Others you’ll mostly watch and approve. What you’re not doing in there is building or editing skills, which stays with us for the reasons in the control question above.

And if what you’d rather do is work with them inside the tools your team already lives in instead of in a portal, that’s possible. It’s scoped as its own piece of work rather than assumed, so raise it early if it matters to you.

If it needed a manual, we’d have built the wrong thing.

No, and we don’t build it to. The AI employees take volume, which is the repeated questions, the chasing, the after-hours. Your people keep the judgment, the relationships, and the exceptions.

The version of this we believe in is a team of three operating with the reach of a team of ten. Companies come to us to grow without growing payroll, not to shrink.

It will, sometimes. AI isn’t deterministic and we’re not going to pretend otherwise, which is exactly why the engagement is built the way it is.

You approve what it can say before launch. Builds and changes run in a sandbox before they reach production, ours included. Consequential actions wait for a person. Every action lands in the record, so a mistake is visible rather than buried. And where AI handles something badly, we tell you up front and route that work to people from day one instead of discovering it with a client.

Scoped per engagement, after we’ve looked at your business, because a published price list would be us guessing.

The shape: a small paid diagnostic that credits toward the build, a one-time build engagement, then a monthly fee to operate and improve the workforce. Nothing is billed hourly. What moves the number is how much information there is and what state it’s in, how many functions you start with, and how much volume runs through them.

Every figure lands in a written proposal before you commit, along with what isn’t included. Only the proposal binds either of us.

Depends on scope, and you’ll get a real timeline in the proposal rather than a slogan here.

What we’ll commit to about the process: you review a working version and test it yourself before anything reaches a client, and the first round of testing exists to find what we missed rather than to confirm we got it right.

No. We build it, we run it, and your clients keep reaching you on the channels they already use. Which channels those are gets defined with you during the engagement rather than handed to you as a fixed set.

That describes most of the businesses we talk to, and it’s usually the same reason. The tool didn’t know the business, and somebody still had to run it.

Off-the-shelf tools break on your exceptions, because your exceptions are where your business actually lives. We start by writing your process down and correcting it with you, and we’re the ones operating it after launch. That’s the difference between something you bought and something that works.

Yes, natively, in both directions. We run engagements in English and Spanish across the United States and Latin America, and the AI employees work in whichever language your client writes in.