Jesus Vargas

Founder and CEO · @LowCodeAgency and @phosailabs (400+ products shipped for Coca-Cola, Amex, Zapier, Sotheby's, Medtronic)

[Jesus Vargas @eltintero]

https://t.co/Tfl63vsWiJ

[Jesus Vargas @eltintero]

How accurate does an AI need to be? Better question: who checks the answer before it goes out?

A person reviews every answer: ship at 80% and keep learning. Nobody reviews: wait until it's close to 95%.

[Jesus Vargas @eltintero]

Nick here is the Jim Cramer of X

[Jesus Vargas @eltintero]

@erikgoinsHQ 15 is a killer price point

[Jesus Vargas @eltintero]

Before we make any recommendations, we interview every relevant employee by voice using an AI agent.

They walk through their actual day. The details that matter most come from the people doing the work.

In one audit with 34 people, we found a team running three design tools and four communication channels simultaneously. Seeing the full picture is what made it possible to fix.

That's the kind of insight that tells us where to start.

[Jesus Vargas @eltintero]

You can redesign a process for AI agents and keep every step in place.

Customer support is a good example, since an AI agent can handle it today.

The steps stay the same, and the agent carries out the ones that fit.

Most processes were designed around people before agents existed, so a weekly routine is a good place to look first. Say it has four steps, which are gathering information, checking it, deciding, and sending the result.

Gathering and sending often suit an agent, and checking and deciding often stay with your people, because they depend on judgment and context.

We map this with mid market teams at @LowCodeAgency. Sometimes a tool you can buy today covers a step, and we tell you.

Pick one weekly process and read through its steps. Which one would you hand over first?

[Jesus Vargas @eltintero]

The FDA is moving toward continuous monitoring requirements for medical devices with integrated models.

Getting the model right at launch is the beginning of the requirement.

Building the infrastructure to know whether it is still right six months later is the ongoing one.

Most teams are building the recommendation.

The audit trail for the recommendation is not on the roadmap.

The FDA is telling you the audit trail is the product.

[Jesus Vargas @eltintero]

Here's what a custom AI agent for sales research does, step by step.
We built it for an employer of record. The goal is to find US companies about to hire in Mexico and write the first email.

First, it searches job posts for US companies hiring in Mexico. Second, it removes companies that already have a Mexico office. Third, it finds the hiring manager. Fourth, it picks the closest case study. Fifth, it writes the draft email.

The full version runs about 90 steps, and the agent chooses the order based on what it finds. A person reviews each draft before it goes out.
Researching one prospect by hand takes a bunch of tabs and a lot of checking, and the agent does it for the whole list.

What goal would you hand to an agent in your company?

[Jesus Vargas @eltintero]

A library checkout ties every book to a name and a date. Nobody has to remember who borrowed what.

Equipment handoffs can work the same way. In the app we built for Rising Ground, the new employee signs digitally when they receive the laptop.

Three months later, the answer to "where is that laptop?" is already there: who has it, since when, and the signature that confirms it.

A quick signature at handoff saves a bunch of back and forth later, right?

DM me and I will walk you through how we built it.

[Jesus Vargas @eltintero]

How accurate does an AI need to be? Better question: who checks the answer before it goes out?

A person reviews every answer: ship at 80% and keep learning. Nobody reviews: wait until it's close to 95%.

[Jesus Vargas @eltintero]

Before we make any recommendations, we interview every relevant employee by voice using an AI agent.

They walk through their actual day. The details that matter most come from the people doing the work.

In one audit with 34 people, we found a team running three design tools and four communication channels simultaneously. Seeing the full picture is what made it possible to fix.

That's the kind of insight that tells us where to start.

[Jesus Vargas @eltintero]

You can redesign a process for AI agents and keep every step in place.

Customer support is a good example, since an AI agent can handle it today.

The steps stay the same, and the agent carries out the ones that fit.

Most processes were designed around people before agents existed, so a weekly routine is a good place to look first. Say it has four steps, which are gathering information, checking it, deciding, and sending the result.

Gathering and sending often suit an agent, and checking and deciding often stay with your people, because they depend on judgment and context.

We map this with mid market teams at @LowCodeAgency. Sometimes a tool you can buy today covers a step, and we tell you.

Pick one weekly process and read through its steps. Which one would you hand over first?

[Jesus Vargas @eltintero]

Since January, we've been building new projects with AI coding tools, which are programs that help developers write and test software.

We found that building this way is more efficient for our team, and the code we write belongs to you and lives in your own accounts, so you can host it wherever you choose.

AI coding tools speed up a lot of the work, and our developers still design the structure of the software, check the security, and test the results with your team, right?

Sometimes a prebuilt tool covers what you need, and I'll tell you that on the call.

At @LowCodeAgency, we figure out the right approach for each custom software project during discovery. I'm happy to talk it through when you DM me "AI build."

[Jesus Vargas @eltintero]

Healthcare compliance is three different problems depending on who you are building for.

Private clinic: flexibility above the HIPAA floor.

Government-regulated program: state-approved forms, exact language, the product conforms to the regulatory artifact.

Health tech startup: evaluated at the scale the product is designed to reach, with minimal flexibility from day one.

Most agencies apply one approach to all three.

The diagnosis changes depending on the patient.

[Jesus Vargas @eltintero]

AI agent vs automation, explained with one example.

Automation follows a rule. A new row shows up in a spreadsheet, and an email goes out. The rule runs the same way every time, and it's typically zeros or ones.

An AI agent takes a goal and picks its own next step. It fits work that changes from case to case, like researching a prospect, where each step depends on what the agent found before it.

If your steps repeat the same way every time, build an automation, because it costs less and runs more reliably.

Which process in your company has steps that change every time? Happy to help you figure out which one fits.

[Jesus Vargas @eltintero]

@martinvars eres un Crack, Martin.

[Jesus Vargas @eltintero]

@Rpaulindaigle @cap , https://t.co/1HPcZoPP4J

[Jesus Vargas @eltintero]

Picture a request for a second iPad landing in an approver's inbox. A thoughtful reply takes a few minutes to write, so the answer often waits for a quieter day.

In the issuance app we built for a nonprofit client, admins review requests item by item. Standard requests get approved in one click, and a mixed request gets a yes for the iPhone and a decline for the second iPad in the same pass.

Every decline needs a written reason, and the app emails it to the requester right away. The requester gets clarity the same day, and the approver gets a documented answer with no extra email to write.

If you review a handful of requests a month, email works fine. Once requests flow across departments, DM me and I will walk you through it.

[Jesus Vargas @eltintero]

Ask this before building a custom CRM: does it tell my team who to call first?

Ours had 1,007 companies and almost no data on them. Reps googled every account before a call, or called whoever replied fastest.

An off-the-shelf CRM only holds what people type in.

So we built our own qualification into it. An AI agent researches each company: headcount, revenue range, AI signals. One rule scores the fit. Every field shows its source.

Reps now open a deal already knowing who to call first.

[Jesus Vargas @eltintero]

Someone on your team pastes a customer's details into a free AI app. Where does it go?

Depends on the plan: does it train on your chats, how long is it kept, what contract covers it. For customer data: business plan, signed data agreement, training off.

[Jesus Vargas @eltintero]

Five minutes after a sales call ends, a draft follow-up is already in my inbox.

The agent pulled the transcript, scored the deal, built the to-do list, matched the case studies to the conversation, and wrote the email. All of it on its own.

I review it and decide what to send. That's where my judgment matters.

My team used to spend two to three hours a day on follow-up. Now they open the action center, review what the agent prepared, and send.

We built this for ourselves first. You need to live with it before you hand it to anyone else.