We have more data than ever. Why are decisions slower?

I spent the weekend researching a company called Palantir.

Beyond its dazzling AI technology and that American obsession with technological innovation as a universal solution, I tried to see the essence...

And here's what I can share right now: what we need isn't technology, but a change in how we think.

Let me be honest: at first, I didn't understand what made Palantir special.

$175B valuation? Hundreds of clients?

"Isn't it just another expensive data platform?"

But as I reviewed their documents, conferences, and success stories... I realized there was something different.

26+2 years at the same company, the same problem

Let me start with my story.

At Samsung Electronics Chile, leading a joint management innovation project in the SCM area for Latin America, data flowed in real-time between supply plants in different countries, headquarters, and client stores and warehouses in Santiago, São Paulo, Bogotá, Lima, Miami...

The G-SCM and ERP systems ran smoothly, and the dashboards were magnificent tools that, for those who knew how to use them, allowed reviewing everything in real-time.

But every Monday morning, in the meeting, the same question kept repeating:

"So... what do we do now?"

We knew inventory was low, but who acts? When? How?

Exchange rates spiked, but do we raise prices or reduce volume? What do we do about sales to distributors? And inventory?

We had mountains of data, but decisions still depended on our heads and Excel.

28 years have passed. The essence of the problem hasn't changed, and perhaps even now, or in the future, it will remain an even more complex dilemma.

What makes Palantir different?

Let me share three things I discovered.

First: they design 'decisions', not 'data'

What Palantir calls 'Ontology' seemed complicated at first.

But the essence is simple:

"What decision do we need to make?" → Define this first"What data do we need?" → Then

connect them

Traditional BI tools show you "what happened?"

Palantir tells you "what should you do now?"

For example, when there's a supply chain problem:

• Traditional way: review 20 dashboards → call/email 5 people → 2-hour meeting → execute in 1-4 weeks

• Palantir: "Contact supplier B (reason: ...)" execution in 30 seconds

Second: they don't sell AI, they sell 'trust'

When ChatGPT came out, I got excited like everyone else.

But when you told the CFO "the AI says this," the same question always came back:

"What's the rationale? What data did it review?"

The reason Palantir's AIP (AI Platform) grew 71% in the U.S. commercial segment after its 2023 launch is simple:

Every AI reasoning comes with 'traceable evidence'. With one click you can see the entire flow of "why did it reach this conclusion?"

In the business world, what we need isn't 'intelligent AI' but 'explainable AI'.

Third: they don't 'sell', they 'co-create'

Palantir's sales approach really impressed me.

Instead of presenting a 6-month contract, they propose:

"Let's work together for 5 days"

·         They bring the client's actual data

·         They solve a real problem together

·         In 5 days they show a working prototype

·         And they ask: "Should we continue?"

98% client retention. Not by chance.

So... what do we do?

At first I thought:

"Do we need to build a system like Palantir?"

No.

What we need isn't the platform, but to 'think like Ontology' based on essential existence.

Let me share 3 things I tried:

Step 1: Define decisions first

In your next strategic meeting, start like this:

"What are the 10 most important decisions we make?"

Example from my team:

·         Which product to increase? (every Friday)

·         Which client gets a discount? (beginning of quarter)

·         Which market to invest in? (end of year)

And for each decision:

·         What data do we need?

·         Who decides?

·         How often?

Just by making this list, you've solved 80%.

Step 2: Connect data with actions directly

We all have many dashboards.

The problem is what comes next:

"Low inventory" → So what?"Satisfaction dropped" → Who does what?"Low sales by store/product at distributors" → "Why should we care about inventory we already sold?"

What I did:

·         Low inventory alert → [Urgent order] button right next to it

·         Complaint report → [Schedule interview] automatically generated

·         Client inventory/sales → result of our business

Reduce from 3 clicks to 1. Small but big change.

Step 3: Start small (not total transformation)

Digital transformation of the entire company? 70% chance of failure.

Instead:

"One team, one problem, 30 days"

Example: SCM and logistics delay problem

·         Week 1: define 10 decisions

·         Week 2: connect data (Excel is perfectly fine)

·         Week 3: simple automation

·         Week 4: validate and expand to other teams

Palantir also started with a single CIA team.

What I recently learned in Latin America

Working in Latin America, I discovered something:

"Technology is universal, but problems are contextual"

Same G-SCM, ERP, RPA, but each country with its reality. Simple examples:

• Mexico: monopolistic distributor abuse, the hidden pitfalls of reshoring

• Chile: brand warfare in the trenches with all-powerful distributors

• Brazil: import structure where you can't profit without local assembly

That is, each 'Ontology' must be based on a customized decision-making structure to maximize execution and organizational operation.

Palantir's genius is right there. One platform, but allowing each client to define their problem in a completely different way.

A question for you

I'm still learning, still in trial and error.

That's why I'm interested in your experience:

1. What are the 3 most important decisions in your organization?

2. How much do data actually contribute to those decisions?

3. How much time passes from decision to execution?

If you share in the comments, I'd love to have a conversation and exchange experiences.

After studying Palantir, I have one conviction:

"The organizations that will survive until 2030won't be those that 'report' data,but those that 'decide' with data and execute like tachyons"

Which side is your organization on?