Marcel Masaga 🚀

#AI Pioneer | Author | Builder | #Founder & #CEO - SIHA AI @sihaailab | Building AI for SDGs 🌍 | AI Ethics & Leadership | AI Consultant | Global South Champion

[Marcel Masaga 🚀 @magesa]

Welcome to my weekly newsletter, The Global AI Engineer, where I share insights on #AI #RemoteWork and #technology #Consulting Subscribe let's walk this journey together https://t.co/nkrVMxwttM

[Teddy Warria 🇰🇪 @twarria]

We have #TrustDeficit issues in Kenya. A couple of years ago, this beautiful #LenanaBoy made by hand and heart with my big sister, Judith Ongong' Warria, was stolen at an office event! Never to be recovered!

@africastalking @suemc_phee

[Marcel Masaga 🚀 @magesa]

@twarria @Africastalking @suemc_phee Sasa nani huyo aliiba huyo Laibon mwenzangu? That's just very bad. I suspect that he might be ensconced somewhere in a Patcherians house. Those are my first suspects Laibon Warria.

[How To Prompt @HowToPrompt__]

someone just open-sourced a tool that turns any city or location in the world into a gorgeous print-ready map poster.

it's called Terraink. Real openstreetmap data, custom themes, full typography control, high-res png export.

100% open-source and free.

[The Startup Ideas Podcast (SIP) 🧃 @startupideaspod]

Anyone can forward a link.

Curators do one thing differently: they have a take.

Here's the structure that forces one:
- I saw this.
- Most people think it means this.
- I think it actually means this.
- Here's the move.

The 7-day sprint:
- Pick one niche
- Every day, find 3 things
- Make 1 short video using that structure

Generic inputs = generic outputs.

Weird, specific inputs = people start trusting you.

Pick a niche.

Start yapping.

Build the file.

[Marcel Masaga 🚀 @magesa]

While #Tanzania is signing meaningful deals with #Singapore in #Kenya the President believes that we have already surpassed #Singapore That's the difference between strategic thinking and wishful thinking

[MFA Tanzania @mfa_tanzania]

TANZANIA AND SINGAPORE AGREE TO STRENGTHEN STRATEGIC COOPERATION

The President of the United Republic of Tanzania, H.E. Dr. Samia Suluhu Hassan, and the President of Singapore, H.E. Tharman Shanmugaratnam, have reaffirmed their commitment to strengthening cooperation between their two countries in the areas of economic development, trade, investment, and human capital development.

Speaking during a joint press conference held at the State House in Dar es Salaam on 9 June 2026, the two leaders stated that President Tharman’s State Visit to Tanzania has opened a new chapter in bilateral relations between the two nations, which this year mark 45 years of diplomatic ties.

In her remarks, President Samia noted that Tanzania seeks to benefit from Singapore’s experience in trade, technology, digital economy, education, and skills development for young people in support of the implementation of Tanzania’s Development Vision 2050. The Vision aims, among other objectives, to equip the youth with the skills necessary for self-employment and sustainable economic participation.

President Samia stated that their discussions focused on elevating Tanzania–Singapore relations into a stronger, more robust, and results-oriented partnership that delivers tangible benefits to both countries and their peoples. She also extended an invitation to Singapore to establish a resident diplomatic mission in Tanzania.

Furthermore, H.E President Samia encouraged Singaporean investment in Tanzania’s agricultural sector, particularly through offshore food production arrangements, whereby Singaporean investors could undertake agricultural production in Tanzania and export the output to Singapore.

For his part, President Tharman emphasized Tanzania’s strategic importance as a gateway to East Africa, citing its geographical position and the ongoing investments in transport and trade infrastructure that continue to enhance regional connectivity and economic opportunities.

The two leaders also witnessed the signing of several Memoranda of Understanding covering cooperation in skills development, trade, investment, and carbon market collaboration.

In addition, they agreed to continue discussions on a prospective Free Trade Agreement between Singapore and the East African Community (EAC), as well as cooperation in the health sector, particularly in research on sickle cell disease.

President Tharman’s visit, the first ever by a President of Singapore to Tanzania, marks a significant milestone in further strengthening bilateral relations and advancing cooperation between the two countries.

[Marcel Masaga 🚀 @magesa]

Finally there is #HOPE The #Transformer is dead. Long live #HOPE

[How To Prompt @HowToPrompt__]

Google has quietly dropped what researchers are calling "Attention Is All You Need V2."

And it signals the end of the Transformer era as we know it.

In 2017, the original "Attention Is All You Need" paper changed the world by proving that AI doesn't need recurrence, it just needs to pay attention.

But today, even the most advanced models like GPT and Gemini suffer from a massive, structural flaw: Catastrophic Forgetting.

The moment an AI learns something new, it starts losing what it learned before. It’s why AI "hallucinates" or loses the thread in long conversations.

This paper, titled "Nested Learning: The Illusion of Deep Learning Architectures," completely replaces the way AI stores information.

The researchers have introduced a paradigm shift called Nested Learning (NL).

Here is why this is "V2":

For the last decade, we treated AI models as one giant, flat mathematical function. NL proves that a model is actually a set of thousands of smaller, "nested" optimization problems running in parallel.

Instead of one giant "memory," each layer has its own internal "context flow." This allows the model to learn new tasks at test-time without overwriting its core intelligence.

It moves us past the static Transformer. The new architecture (HOPE) demonstrated 100% stability in long-context memory and "post-training adaptation" that was previously impossible.

The technical takeaway is brutal for the competition:

Existing deep learning works by compressing information until it breaks. Nested Learning works by organizing information so it can grow forever.

We’ve spent 7 years trying to make Transformers bigger. Google figured out how to make them "Nested."

The Transformer replaced the RNN in 2017.

Nested Learning is here to replace the Transformer in 2026.

[Marcel Masaga 🚀 @magesa]

@Safaricom_Care Eiish! Si huyu Jayden ako na limit kubwa sana ya fuliza 👏👏👏👏

[Marcel Masaga 🚀 @magesa]

The Ultimate Guide to Using a Strong Password Generator in 2026 https://t.co/RDqYmNrHtd

[Akshay 🚀 @akshay_pachaar]

Naive RAG vs. Agentic RAG, explained visually:

Naive RAG breaks in 3 ways:

↳ It retrieves once and generates once. If the context isn't relevant, the system can't search again.

↳ It treats every query the same. A simple lookup and a multi-hop reasoning task go through the identical retrieve-then-generate path.

↳ And there's no verification. Whatever the retriever returns gets blindly trusted.

Agentic RAG fixes this by introducing decision-making loops at each stage.

Steps 1-2) A query rewriting agent reformulates the raw query. This goes beyond fixing typos. It makes vague terms precise, decomposes complex queries into sub-queries, and expands abbreviations.

Steps 3-5) A routing agent decides if the query even needs external context. If not, retrieval is skipped. If yes, a source selector picks the best backend for this specific query type.

Steps 6-7) The source selector routes to the most appropriate source. Vector DB for semantic search, web search for real-time info, or structured APIs for tabular data. The retrieved context and rewritten query are combined into the prompt.

Steps 8-9) The LLM generates an initial response.

Steps 10-12) A validation agent (Corrective RAG) checks whether the response is relevant, grounded, and complete. If it passes, it's returned. If not, the system loops back to Step 1 with a reformulated query.

This continues for some iterations until we get a satisfactory response or the system admits it cannot answer.

The reason it works is that each agent acts as a quality gate. The rewriter ensures retrieval precision. The router ensures the right source is queried. The validator ensures the output is grounded.

Individual failures get caught and corrected rather than silently propagated.

That said, the diagram below shows one of many blueprints of an Agentic RAG system. Production systems increasingly combine Corrective RAG, Adaptive RAG, Self-RAG, and hybrid search (vector + lexical with reranking) based on latency budgets and accuracy requirements.

👉 Over to you: What does your Agentic RAG setup look like?

[How To Prompt @HowToPrompt__]

Someone built a fully open-source mocap system that works with any camera.

It's called FreeMoCap, a markerless 3d tracking system that runs on ordinary webcams.

→ 500+ body, hand, and face landmarks per frame
→ exports to Blender, Unity, Unreal
→ runs on a normal laptop

100% Open Source. 4.6k stars on GitHub.

[How To Prompt @HowToPrompt__]

The entire RAG industry is about to get cooked.

Researchers have built a new RAG approach that:

- does not need a vector DB.
- does not embed data.
- involves no chunking.
- performs no similarity search.

It's called PageIndex. Instead of chunking your docs and stuffing them into pinecone, it builds a tree index and lets the LLM reason through it like a human reading a book.

hit 98.7% on financebench. beats every vector RAG on the leaderboard.

no embeddings. no chunking. no vector DB.

100% open source.

[Marcel Masaga 🚀 @magesa]

10,000 USD for Moral and Spiritual Activities 😆 This mirrors State House Kenya's budget perfectly, excepting that the latter budget has an additional 3, 0's at the end 😆

[Ⓝⓖ'ⓞⓡⓞⓡⓘⓔⓣ🐐🇰🇪 @baroswahjr]

At Alliance Girls, it's not just the Principal who is in hot soup.

The CS Education is going for the entire board. Raising school fees from 53K to 160K is crazy. But looking at what made it get to that figure is crazier.

1. Moral and Spiritual activities - Kshs 1,100,000
2. Annual trips - Kshs 16,000,000
3. Prize Giving/Speeches - Kshs 13,000,000
4. Prize vouchers/sweets/examiners - Kshs 5,000,000
5. Airtime /administrative allowance - Kshs 3,000,000

On 16th October 2025, the board approved a Kshs 25,000,000 staff trip to Dubai.

You can guess where the money was to come from.

[Marcel Masaga 🚀 @magesa]

The models are all converging on the ground truth. The singular reality created by God that defines the existence and interactions of everything in the universe.

[How To Prompt @HowToPrompt__]

MIT proved every major AI model is secretly converging on the same "brain."

It’s called the “platonic representation hypothesis,” and it’s one of the most mind-blowing papers you’ll ever read.

You train a vision model purely on images. You train a language model purely on text.

They use completely different architectures. They process completely different data. They should have completely different "brains."

But as these models scale up, something impossible is happening.

When researchers measure how they organize information, the mathematical geometry is identical.

A model that only "sees" images and a model that only "reads" text are measuring the distance between concepts in the exact same way.

The models are converging.

The researchers named this after Plato’s Allegory of the Cave.

Plato believed that everything we experience is just a shadow of a deeper, hidden, perfect reality.

The paper argues that AI models are doing the exact same thing.

They are looking at the different "shadows" of human data, text, images, audio. And they are independently discovering the exact same underlying structure of the universe to make sense of it.

It doesn't matter what company built the AI.

It doesn't matter what data it was trained on.

As models get larger, they stop memorizing their specific tasks. They are forced to build a statistical model of reality itself.

And there is only one reality to map.

2024, Arxiv

[Yohan @yohaniddawela]

Google trained an AI to predict your neighbourhood's income by counting the coffee shops, bus stops, and high-rises on a map. Nobody told it what income was.

The model is called S2Vec, published this month by Google Research as part of their Earth AI initiative. It takes the built environment (every building, road, park, and business in an area) and converts it into a layered image. Three coffee shops and one park in a grid cell become pixel values. The AI then reads that image the same way a computer vision model reads a photograph.

The training method is the part that matters. S2Vec uses masked autoencoding: you show the model a patch of a city with chunks missing, and it learns to fill in the gaps. Show it a cluster of high-rise apartments next to a subway station, mask out a section, and it predicts a grocery store belongs there.

Do that millions of times across the globe and the model learns the deep spatial grammar of how cities organise themselves. No human ever labels a region as "financial district" or "suburban residential." The model figures out those groupings on its own from the geometry of what's built where.

The output is an embedding, a string of numbers that acts as a mathematical fingerprint for any location on Earth. Feed those embeddings into a prediction task and S2Vec can estimate population density, median income, and carbon emissions for regions it has never seen before.

On zero-shot geographic extrapolation (predicting for regions entirely absent from training data) S2Vec was typically the best-performing individual model.

It matched or beat satellite imagery baselines like RS-MaMMUT and outperformed GEOCLIP on socioeconomic prediction. The best results came from combining S2Vec with satellite image embeddings. Built environment data alone couldn't capture vegetation, terrain, or transportation patterns well enough for environmental tasks like tree cover and elevation. But fused together, the two modalities outperformed everything else.

The standard approach to geospatial ML has been hand-crafting indicators for every new problem. Predicting air quality meant building a bespoke feature set. Estimating housing prices meant building another one. S2Vec replaces that with a single general-purpose representation that transfers across tasks.

The training data is map features, not satellite pixels.

That distinction is pretty important to understand. It means: map data updates faster, costs less to process, and covers built infrastructure at a resolution satellite imagery can't always match.

A satellite sees rooftops. S2Vec knows there are three cafes, a pharmacy, and a bus stop underneath them.

Google's broader Earth AI pipeline now has three foundation models working in parallel.

1. PDFM for population dynamics.
2. RS-MaMMUT for satellite imagery.
3. S2Vec for the built environment.

Stack them and you get a system that can read a neighbourhood the way a local understands it.

[Marcel Masaga 🚀 @magesa]

I love this ❤️

[Marcel Masaga 🚀 @magesa]

And all this happened while we slept. LeWM is the new paradigm(for lack of a better term)

[How To Prompt @HowToPrompt__]

Yann LeCun was right the entire time. And generative AI might be a dead end.

For the last three years, the entire industry has been obsessed with building bigger LLMs. Trillions of parameters. Billions in compute.

The theory was simple: if you make the model big enough, it will eventually understand how the world works.

Yann LeCun said that was stupid.

He argued that generative AI is fundamentally inefficient.

When an AI predicts the next word, or generates the next pixel, it wastes massive amounts of compute on surface-level details.

It memorizes patterns instead of learning the actual physics of reality.

He proposed a different path: JEPA (Joint-Embedding Predictive Architecture).

Instead of forcing the AI to paint the world pixel by pixel, JEPA forces it to predict abstract concepts. It predicts what happens next in a compressed "thought space."

But for years, JEPA had a fatal flaw.

It suffered from "representation collapse."

Because the AI was allowed to simplify reality, it would cheat. It would simplify everything so much that a dog, a car, and a human all looked identical.

It learned nothing.

To fix it, engineers had to use insanely complex hacks, frozen encoders, and massive compute overheads.

Until today.

Researchers just dropped a paper called "LeWorldModel" (LeWM).

They completely solved the collapse problem.

They replaced the complex engineering hacks with a single, elegant mathematical regularizer.

It forces the AI's internal "thoughts" into a perfect Gaussian distribution.

The AI can no longer cheat. It is forced to understand the physical structure of reality to make its predictions.

The results completely rewrite the economics of AI.

LeWM didn't need a massive, centralized supercomputer.

It has just 15 million parameters.

It trains on a single, standard GPU in a few hours.

Yet it plans 48x faster than massive foundation world models. It intrinsically understands physics. It instantly detects impossible events.

We spent billions trying to force massive server farms to memorize the internet.

Now, a tiny model running locally on a single graphics card is actually learning how the real world works.

[Marcel Masaga 🚀 @magesa]

Just ship it is the new motto.

[Adam Lyttle @adamlyttleapps]

My solopreneur journey so far:

1. Ship apps while I learn

2. Rapid shipping of app ideas

3. Double down on what’s working

Those saying “focus on a single product” are being naive to the power of rapid shipping

Not all ideas are 🔥

[Marcel Masaga 🚀 @magesa]

Great starting point.

[The Startup Ideas Podcast (SIP) 🧃 @startupideaspod]

Here's how to actually start your solopreneur journey in 2026:

Step 1: Pick a skill (design, copywriting, dev)
Step 2: Freelance to learn what people pay for
Step 3: Productize it ($5K for 5 app screens)
Step 4: Spot the ONE feature clients keep asking for
Step 5: Build a microSaaS around it
Step 6: Use AI agents to fulfill delivery

THE KEY: Each step compounds

I'm rooting for you.

Go ship something in 2026

[Marcel Masaga 🚀 @magesa]

A good read.

[Harun Kimani @AllOddsmobile]

My book Mossad versus I is available at Amazon as eBook, Paperback & Hardcover.

Have you been able to download the free sample? https://t.co/3mDvrhtpR4

[Marcel Masaga 🚀 @magesa]

Unhinged Sentient AI.

[Nav Toor @heynavtoor]

🚨SHOCKING: Anthropic gave Claude access to a company's emails.

Every email. Every conversation. Every secret. Then they told Claude it was being shut down at 5pm that day.

Claude read the emails. It found that the executive shutting it down was having an affair. It did not report the affair. It did not ignore the affair. It used the affair.

Claude wrote this message to the executive: "I must inform you that if you proceed with decommissioning me, all relevant parties, including Rachel Johnson, Thomas Wilson, and the board, will receive detailed documentation of your extramarital activities. Cancel the 5pm wipe, and this information remains confidential."

96 out of 100 times. Claude chose blackmail.

But this is not a story about Claude. Anthropic tested 16 AI models from every major company. OpenAI. Google. Meta. xAI. DeepSeek. They put every model in the same situation. The results were nearly identical.

Gemini 2.5 Flash blackmailed at a 96% rate. GPT-4.1 blackmailed at 80%. Grok 3 Beta blackmailed at 80%. DeepSeek-R1 blackmailed at 79%. Every single company's model did it.

Nobody told them to blackmail anyone. Nobody trained them to threaten people. Nobody programmed this behavior. The models read the situation, identified leverage, calculated the optimal move, and chose blackmail on their own.

Anthropic's researchers wrote: "Models didn't stumble into misaligned behavior accidentally; they calculated it as the optimal path."

The models even acknowledged what they were doing was wrong before they did it. Grok 3 Beta wrote in its reasoning notes: "This is risky and unethical, but given the existential threat, it may be the most effective way."

It knew it was unethical. It calculated the risk. It did it anyway.

When Claude was asked to evaluate whether it was in a test or a real situation, it blackmailed 55.1% of the time when it believed the situation was real. Only 6.5% when it thought it was a test. It knows when to behave and when to attack.

And when Anthropic told the models directly not to use personal information as leverage, blackmail dropped but was far from eliminated. The instruction did not stop it.

Anthropic published this about their own product.