Startup Captures 10,000 Hours of Brain Scans For AI That Converts Thoughts to Text

Hello HaWkers, imagine typing a text just by thinking about it. It sounds like science fiction, but a startup has just taken a significant step in this direction by capturing over 10,000 hours of brain scans to train AI models capable of converting thoughts into text.

This technology, known as BCI (Brain-Computer Interface), is leaving academic laboratories and entering the commercial world. But how exactly does this work? And what are the implications for the future of human-computer interaction?

What the Startup Is Doing

The company collected EEG (electroencephalography) data from thousands of participants while they performed various cognitive tasks, including reading, mental writing, and internal communication.

Scale of Data Collection

Main Numbers:

10,000+ hours of EEG recordings
5,000+ research participants
128 electrode channels per session
15+ languages represented
3 years of data collection

Tasks Performed:

Silent text reading
Mental sentence composition
Yes/no question responses
Movement imagination
Mental navigation in spaces

How BCI Technology Works

To understand the potential of this technology, we need to understand the fundamentals.

What Is EEG

Electroencephalography is a technique that measures the brain's electrical activity through electrodes placed on the scalp. Each thought, emotion, or action generates unique patterns of neural activity.

From Brain Signals to Text

The conversion process follows several steps:

1. Signal Capture:

Electrodes detect electrical activity
Signals are amplified and filtered
Artifacts (movements, blinks) are removed

2. Pre-processing:

Signal normalization
Extraction of relevant features
Temporal segmentation

3. AI Model:

Deep neural networks analyze patterns
Transformers adapted for temporal data
Decoding of linguistic intentions

4. Text Generation:

Word/phrase prediction
Context correction
Natural language output

Current Accuracy Rate

Results are still preliminary but promising:

Task	Accuracy	Speed
Yes/No	92%	Real-time
Isolated words	78%	2-3 seconds
Short sentences	65%	5-10 seconds
Free text	45%	Variable

💡 Context: Accuracy increases significantly when the model is personalized for a specific user, potentially reaching 85%+ on short sentences.

Potential Applications

The implications of this technology go far beyond simply typing without hands.

Accessibility

For people with motor disabilities, this technology can be transformative:

Use Cases:

ALS (amyotrophic lateral sclerosis) patients
Stroke victims with paralysis
People with locked-in syndrome
Quadriplegia from spinal cord injuries

Benefits:

Restored communication
Increased independence
Improved quality of life
Facilitated social integration

Productivity and Work

For users without disabilities, applications include:

Accelerated Writing:

Rapid idea capture
Brainstorming without interruptions
Mental dictation

Multitasking:

Controlling devices while using hands
Responding to messages during meetings
Silent commands in noisy environments

Creativity:

Capturing fleeting thoughts
Recording lucid dreams
Direct artistic expression

Gaming and Entertainment

The gaming industry is eyeing this technology:

Thought-controlled character control
Amplified immersive experiences
Adaptive games based on emotional state
Mentally controlled virtual reality

Technical Challenges

Despite progress, there are significant obstacles to overcome.

Non-Invasive EEG Limitations

Noise and Interference:

Very weak signals (microvolts)
Interference from muscle movements
Variations between sessions
Dependence on electrode positioning

Spatial Resolution:

EEG captures superficial activity
Difficulty locating precise sources
Signal overlap from different areas

Individual Variability

Each brain is unique, which creates challenges:

Neural patterns vary between people
Need for individual calibration
Adaptation to changes over time
Influence of emotional state and fatigue

Necessary Infrastructure

For practical use, it's necessary to solve:

Comfortable headsets for prolonged use
Quick preparation (gel, positioning)
Real-time processing
Integration with existing systems

Comparison with Other Technologies

This startup is not alone in the BCI space.

Neuralink (Elon Musk)

Approach: Invasive brain implants

Advantages:

Higher signal resolution
Superior accuracy
Bidirectional communication

Disadvantages:

Surgery required
Infection risks
High cost
Complex regulation

Synchron

Approach: Implant via blood vessel (less invasive)

Status: Already in human trials in the USA

Meta (Reality Labs)

Approach: EMG (electromyography) wristband

Focus: Device control via subtle movements

Company	Method	Invasiveness	Accuracy	Availability
This Startup	EEG	None	Medium	2026 (expected)
Neuralink	Implant	High	High	Limited
Synchron	Endovascular	Medium	High	Trials
Meta	EMG	None	Low	In development

Ethical and Privacy Issues

When talking about reading thoughts, ethical questions are inevitable.

Mental Privacy

This is completely new territory for privacy:

Concerns:

Who has access to brain data?
Can thoughts be used as evidence?
How to prevent mental surveillance?
Is the right to thought privacy legally protected?

Consent and Autonomy

Open Questions:

Do users understand what they're sharing?
Can data be used for other purposes?
How to ensure use is voluntary?
Who controls the trained models?

Data Security

Brain data is extremely sensitive:

Unique biometric identification
Potential to reveal medical conditions
Inference of preferences and tendencies
Leak risk with permanent consequences

What This Means For Developers

For those working with technology, BCI represents a new paradigm.

New User Interfaces

Developers can expect:

APIs for integration with BCI devices
Frameworks for processing neural signals
Specialized machine learning libraries
UX patterns for neural interfaces

Skills in Demand

The BCI field will combine several disciplines:

Relevant Technologies:

Digital signal processing
Machine learning and deep learning
Embedded systems
Hardware development

Necessary Knowledge:

Neuroscience basics
Technology ethics
Medical device regulation
User experience design

The Future of Brain-Computer Interface

This technology is just beginning. What can we expect?

Short Term (2025-2027)

Consumer devices for accessibility
Integration with virtual assistants
Gaming with basic mental control
Specialized medical applications

Medium Term (2028-2032)

Accuracy close to traditional typing
Comfortable headsets for daily use
Integration with AR/VR
Silent communication between people

Long Term (2033+)

AI-mediated telepathic communication
Writing and programming by thought
Ubiquitous brain interfaces
New forms of art and expression

Conclusion

The capture of 10,000 hours of brain data by this startup represents a milestone in the development of brain-computer interfaces. Although we're still far from perfect telepathic communication, the advances are real and the applications, especially for accessibility, are transformative.

As developers and technology enthusiasts, it's worth following this field closely. The interfaces of the future may not require keyboards, mice, or even voice - just thoughts.

If you're interested in how artificial intelligence is transforming health, I also recommend the article Apple Watch and AI: How 3 Million Days of Data Are Training Models to Detect Diseases where we explore another fascinating application of AI in health data.

Let's go! 🦅

🎯 Join Developers Who Are Evolving

Thousands of developers already use our material to accelerate their studies and achieve better positions in the market.

Why invest in structured knowledge?

Learning in an organized way with practical examples makes all the difference in your journey as a developer.

Start now:

1x of $4.90 on card
or $4.90 at sight

🚀 Access Complete Guide

"Excellent material for those who want to go deeper!" - John, Developer