abstracción

Your company pushes hard, but the forces don't pull together

Maybe your version of this already has AI in it: you adopted the tools, your teams build faster than ever, and the value doesn't show up. Capabilities grow as separate features instead of one connected experience. Generic outputs dilute a brand you spent years building. And underneath it all, the data can't quite carry the automation being stacked on top of it.

Or maybe it looks more familiar: capable teams each pulling in their own direction, tools that don't talk to each other, every department with its own spreadsheet and its own version of the truth. It's the point every mid-sized or large company reaches when it grows faster than its coordination. Projects move forward, but with no shared direction, and the complete picture lives only in your head.

The underlying diagnosis is always the same: your business knowledge lives scattered, spread across people's heads, emails, spreadsheets, and chats. And scattered knowledge carries two costs that weigh more today than ever: artificial intelligence can't use it (it keeps answering in generalities, because it doesn't know your business) and strategies remain wishes: slide decks nobody opens again. It's not a lack of technology. You already have the technology. It's that nothing coordinates it toward the same place.

“A PMO, a website, a CRM, an ERP… and a QMS”

Five acronyms that sum up seven years of work, in the order they were born: first the method (a PMO with PMI discipline) and its engine (AI); with them, the strategy became three tools: the public website that attracts and quotes, the CRM that walks every prospect to the sale, and the ERP that governs production, purchasing, inventory and dispatches. And in 2026 the same method stopped governing only software and started governing the company: its management system, the QMS.

None of these figures is a projection: both columns come out of the same repository.

And the three tools operate as a chain, set off by the client with a single click:

Compressing seven years of reality into a five-acronym sentence, and being able to unfold it again, down to the last resistor. That is abstraction.

I build the office where your entire company can be seen whole, and governed

I'm not here to sell you yet another piece of software. I start by understanding how your company actually runs, the process that today lives in Excel, WhatsApp and people's heads. From there I structure that knowledge (the purpose, the strategy, the departments, the projects, the processes) into a single model ordered in layers: what we project managers call a PMO, the office where your whole strategy lives, gets decided, and gets delivered: your company's brain. With one difference that changes everything: artificial intelligence carries the weight of keeping it alive: documenting, measuring, tracing, correcting.

Out of that office come the strategies that actually get done. And along the way, whatever systems are needed get built: custom software, a sales channel, an integration. Software is a result of the journey, not the offer. The offer is that all of your efforts finally pull in the same direction.

This is how far it goes: the company itself, written down and running. Every position with its backup, every process with its owner, employment contracts and the internal regulation generated from the platform, training with records, and internal audits run as guided projects. A management system people actually use, not a binder for the auditor. In 2026 it is taking a manufacturer to ISO 9001 certification, with the audit in December.

Structure and artificial intelligence make each other grow

It's a back-and-forth: your business knowledge, simply structured, feeds artificial intelligence. And AI, armed with that knowledge, returns far more valuable guidance and strategy, which in turn enriches the model. Every turn of the circle leaves your company more ordered and your people freer for what matters: creating value and well-being.

The hardest system to upgrade is never the software

AI is only as good as the data and the systems behind it, and "good" doesn't mean "plenty". A PMO is not a warehouse of documents: piling up information only gives the AI more room to get lost. What makes knowledge usable is its shape.

That's where the discipline of the PMI (the Project Management Institute, the body behind the PMBOK Guide, the world's standard for running projects; PMP is its certification, one of the most demanding) turns out to be wiser than it looks: scope, risks, stakeholders, decisions, lessons learned. And the other half of the shape comes from being a full-stack developer: knowledge structured not only the way a manager reads it, but the way software consumes it: single sources of truth, clean layers, each piece of data meaning exactly one thing. Method gives the knowledge its order; engineering gives it its architecture.

Knowledge structured that way is something an AI can actually stand on: it doesn't guess the context, and it doesn't swallow it whole either: it knows where each piece lives, finds it, and reads it only when it needs it. AI is enormously powerful, but its attention has limits; structure is how you spend it where it matters. I learned this the unglamorous way: years of trial and error on genuinely complex projects. On loose information, AI burns through its tokens before it ever produces a real result; on structure, it goes straight to it.

And there's a second foundation the AI conversation keeps skipping: people. Resistance to change is often more complex than the technology itself, and it never announces itself. It looks like the tool everyone praised and quietly stopped using. The workflow that shines in the demo and dies on the floor. The report nobody trusts enough to decide with. When adoption is designed in from day one, the opposite happens: your people embrace the system, give it their full strength, and the investment pays off.

I also learned, by failing, the part no name tells you. I once built the "correct" project office: repository, templates, reports. Nobody used it. It died on its own, the way almost every system dies: not from bad design, but because it changed nobody's way of thinking. And when AI arrived I made the opposite mistake: speed without structure; I delivered faster, not better. That's where I understood the difference between installing and planting: what's installed gets uninstalled; what's planted (a way of seeing your own business, of deciding with data) keeps growing on its own, because it no longer lives in the software: it lives in the people.

So before building anything, I look for the risks you may be ignoring:

• Is your data actually ready to carry automation?
• Who validates what the AI produces, and against what acceptance criteria?
• Is your team using the tool, or quietly working around it?
• Which decisions still live in one person's head?

Naming these early is what turns AI from a bet into a plan.

From the uncoordinated company to the aligned one

1. Structure your entire company's knowledgeDepartments, projects, and processes distilled into a single model: your whole strategy, ordered with professional rigor.
2. Make it operable by artificial intelligenceThe strategy doesn't end up in a report: it lives in a working system that AI reads, maintains, and integrates.
3. Plant the way of thinkingYou see your whole company for the first time and put things in order step by step; the model isn't handed to you: it's planted in you.
4. Integrate what you already useYour systems, files, and routines connect to the model. I know how to connect systems because I've built them.
5. Water every departmentEach department learns to plan, coordinates with the others, and enriches the model on its own.
6. The resultStrategies that stop being wishes, and the kind of impact that only arrives when every piece pulls in the same direction.

The systems come out of the same path, each one when the strategy calls for it

This is not a proposal: it's been my day-to-day for seven years

What's built and running today, inside a real manufacturing company: all different systems, all connected, all governed from the same office:

And that same office is home to the ISO 9001 quality system I lead today: job roles and their backups, the Academy's courses, the internal audit and the certification are planned and tracked there, and people carry them out on the platform: they sign in, train and report progress inside the very system they use. External audit planned for December 2026: the abstraction submitted to audit, not self-praise.

Juan Camilo Becerra

Find me on LinkedIn ↗

I don't only build technology. I use it to change how a company works: taking people and processes to a genuinely AI-native way of working, installing strategic thinking and breaking resistance to change, so the technology is actually adopted and the organization reaches goals that technology alone cannot reach.

Today, Director of Digital Transformation at a manufacturing company in Colombia, reporting to the General Manager: the platform, the PMO and the ISO 9001 program this page describes are what I operate there, from the inside, every day.

Civil engineer, full-stack developer (end-to-end software, from the server to the screen) and certified project manager (PMP): a combination that almost never lives in a single person, and that lets me understand your company whole: the people, the projects, the information, and the artificial intelligence.

The training holding all of this up: Civil Engineering at the Universidad Nacional de Colombia (the country's leading public university, famous for how hard it is to get in), a Master's degree in Project Management, and the PMP certification from the Project Management Institute (USA), one of the discipline's most demanding standards, taken and passed in English.

I've been doing this for ten years, in very different worlds

It started back in my Civil Engineering undergrad: that's where I fell in love with data and automation, and that focus has followed me through every job since. The same craft, across realities that look nothing alike:

• Rivers and rain.I started out processing hydrometeorological data across Cundinamarca (rainfall, streamflow, climate stations): cleaning it, structuring it, and on that foundation running complex statistical and even risk-management analyses, so engineering projects could make decisions on real terrain.
• The city underground.At Bogotá's water utility I turned the scattered records of a district's sewer network into a geographic system built with two everyday tools: QGIS and Excel, crossed with each other and with the real progress of the works on the ground: over 4,200 acres of invisible infrastructure, made into a living map used to illustrate, plan and guide the different crews in the field.
• The factory.Seven years inside a manufacturing company: three generations of management systems, up to the current platform, built end-to-end with AI as a daily working tool. Today it is truly one single system: the public website (catalog and quoting tool wired live to it, positioned in search engines) drops every prospect into the CRM, which walks them through quotation, sale and receivables; a sale triggers dispatches, production with its recipes and materials, purchasing and inventory; operators report progress from the plant floor; the team gets its alerts automatically; and an in-house academy trains people inside the very system they use. The latest addition: a single inbox for everything that reaches sales (WhatsApp, the sales mailbox and phone calls), where every message is kept under custody, voice notes are transcribed, and an AI reads the whole conversation with one schema for all three channels and proposes what it is: a new lead with product, quantity, city and urgency, a customer with something in progress, a complaint or warranty claim, a supplier, a payment, a job application or spam; a person accepts, discards or redirects from one inbox. Before switching the AI on we measured it dry against 30 days of real traffic: 154 unknown senders, 68 real leads, and none of those 68 existed in the CRM (the inbox and the custody are live; the AI's judgment waits for the privacy notice). And the ISO 9001 quality system is planned and tracked in that same office, with the external audit planned for December 2026. To grasp the complexity it covers: a single traffic light can require up to 58 parts crossing four different production lines, each with its own processes, and everything, from sourcing and producing every part to arriving on time in the country where it's needed, coordinated inside one single system.

  A real order moving through the system: the client orders on the website; the system checks what's in inventory, buys what's missing and coordinates four production lines that converge into a single 58-part traffic light, packed with its own label per package and dispatched on time. This image is, in itself, the craft: all of that complexity, abstracted until it can be understood at a glance.

• The store.A digital shelf for a retailer: the real gondola, with its products and its labels, turned into a model that is designed, stocked and checked from a phone. It's a case in progress, under a confidentiality agreement, so it has no figures here yet. It's the other end of the spectrum: where manufacturing thinks in months, retail thinks in hours.

Different industries, same craft: understand the reality, structure its knowledge, and leave behind systems people actually use.

The technique behind it, in case you want to see it: I started in advanced Excel with macros (VBA), moved through Microsoft's Power Platform (Power Apps, Dataverse), and today I build with end-to-end JavaScript (Node.js, React), SQL and Azure cloud: the same stack holding up everything this page shows you. Having lived at all three levels has a practical consequence for you: whatever tool your operation runs on today, I understand it, I can integrate it, and I respect it; the best move is almost never to wipe everything and start from zero.

One last confession: I love videogames. They taught me something engineering sometimes forgets: a tool shouldn't just work: it should feel good to use, intuitive, even satisfying. Otherwise people don't come back. A good part of why my systems actually get used comes from there.

Today this is a one-person studio, and I'm not shy about it: you're not buying from a corporate façade; you work with the same person who designed, built, and keeps growing everything you see here.

"I don't like simply selling a tool: I commit to solving complex problems in a way that holds up over time. And once you get used to that, you feel empowered, and you never stop dreaming about where you want to go next."

The first step is small on purpose: a 45-minute conversation about your real case. You leave with an initial map of where your business knowledge lives loose, and what to put in order first. No strings attached.

camilo@abstraccion.ai