Every sustainable business will run its own AI brain.
That's my vision: every product, engineering, B2C, B2B and investment decision becomes context for the next.
Kortex
The brain of
your business.
Every conversation. Every decision.
One memory to build everything on.
Concept, in development. Pieces already run in production.
Your history goes in. Your next move comes out.
The vision
My AI agents
write AI agents.A goal.
Not a task list.A brain that grows
with your business.
I live and breathe AI agents in my workflow. This is where I think it all goes.
KORTEX
Why every business needs a brain
Like having a product online in the internet age.
I believe today's AI models, configured and fine-tuned, can grow that foundation with very little infrastructure.
Agents won't gain consciousness.
They don't need to.
My take: not in our lifetimes. Give them a goal, feedback and memory to learn, acquire tools and expand their skills.
Start collecting now.
Everything.
A brain that learns from your history, proposes the next move and lets you choose how far it goes.
Start before you need it
The memory you save.
The insight you keep.
The data you don't capture today is the insight you can't have next year.
It's a long road. Wiring this into how a company actually works is not easy. Better early than late.
Months you didn't captureMemory you can build on
Hire me or not:
start collecting
your data now.
Transcribe client calls. Remove personal information, but keep the useful context. Save emails, photos, videos, work-session caches. Everything that explains how your business works.
Collect it religiously. One day soon, you feed that history to your new brain and get a suggestion that makes you think: “Wow. How did I not think of that?”
Useful from the start
Insights.
Straight away.
Feed it the history you already have. My aim: answers in week one to questions nobody had time to ask.
Sales · Example
Which pitches became customers?
Match the conversations to the deals that followed. Look for what actually worked.
Pipeline · Example
Where do our deals stall?
Follow the gaps between calls, replies and decisions. Find the step that keeps waiting.
Product · Example
What do clients want that we don't sell?
Bring the recurring requests out of calls, chats and tickets. See the pattern.
Trust · Example
What did we promise and never deliver?
Read the commitment in the transcript. Check it against what happened next.
Every morning after
An analyst that
never clocks out.
What changed. What's at risk. What to do next.
The business stays in the picture.
Your morning brief.
What changed
A request from a sales call has appeared again in support. There may be something worth building.
At risk
A commitment from a client call still has no matching delivery. Check it before the next conversation.
Do next
Draft the follow-up. Pull in the original promise. Ask you to review it before it goes out.
Illustrative brief. Real answers would come from your company's history.
Building automation is already in the past.
The choice is yours.
Control on every automation.
Without micromanaging every move.
It writes. You send.
The agent prepares the work. You decide what leaves the building.
Example: a follow-up after a sales call, ready in your voice.
It checks with you first.
The agent proposes the next move and waits for your approval.
Example: a reply that makes a commitment, held for your say-so.
It just does it.
For the work you've chosen to hand over, the agent gets on with it.
Example: file the meeting notes and update the CRM after a call.
The foundation
Every agent.
Every hire.
Already in the loop.
One source of context for whatever you build next.
No starting from a blank page.
New hire, day one · Illustrative chat
Why did we lose that client last spring?
The follow-up went quiet after the call. A promised integration never shipped. I'd check the original promise against the delivery history.
Example sources: call transcript · email thread · delivery ticket
Ask anything about the company, since day one. The sales context. The support history. The decisions behind the code. A new hire should inherit the memory, not have to reconstruct it.
Proof, not a pitch deck
It already runs a business.
Mine.
Kortex runs my own operation every day: calls, agents, reports and decisions in one memory. Your company gets the same engine, built around how you work.
1,045,056
Events captured, connected and searchable.
1,367
Calls transcribed.
50,347
WhatsApp messages.
975
Learnings agents read.
80
Projects behind those learnings.
Kilazi Live.
My live call transcription. Speech translated live. A PDF report after every call. One tap sends the call to an agent.
Call copilot.
While I'm on a call, it reads the transcript and everything Kortex knows. Takes, pushback and follow-ups, updated every minute.
AI engineer.
335 PRs merged for a client in 102 days. Work that reached the codebase.
AI employee.
3,247 messages handled on WhatsApp. Another piece of the same idea, running for a client.
Reports that show up.
Daily PPC and SEO reports, written by an agent every morning. Weekly operations reviews. Already running for a client.
All my AI products. In one.
From the idea to the work
Two ways to
work with me.
Let's build your brain.
I'm dying to build
one for a company.
I build from Buenos Aires. Bring your case. Let's brainstorm with my agentic team and grow a toolset your business can keep building on.
A few good
questions.
What is Kortex?
My idea for a business brain: one memory for your conversations, decisions and work. It feeds insights, an ongoing analyst, your agents and your people.
Is it a product I can buy today?
Not off the shelf. Kortex for business is a concept in development. I build it with you, around your company. The engine already runs my own operation, and several of its pieces run for clients today.
What data does it need?
Start with the history you already have: emails, call transcripts, chats, documents and decisions. Connect the useful channels, remove personal information where appropriate, and keep collecting.
How long does it take?
There is no honest universal timeline. The first step is useful history and a clear question. Wiring a brain into how your business actually works is a longer road. We scope it together.
Is my data private?
Access, storage, retention and model providers need to be agreed before connecting your data. I would design those boundaries with you. There is no blanket security promise for a system we have not built yet.
Does it replace people?
The point is to give your people and agents a shared memory, and let you choose what gets drafted, approved or automated. You decide the responsibility it gets.