Memory Bytes
Відкрити в Telegram
Architecting scalable systems and documenting engineering deep-dives Blog: https://blog.wissam.space Insta: https://ig.me/u/memorybytes.space
Показати більше408
Підписники
Немає даних24 години
-27 днів
-1130 день
Триває завантаження даних...
Схожі канали
Хмара тегів
Вхідні та вихідні згадування
---
---
---
---
---
---
Залучення підписників
липень '26
липень '260
в 0 каналах
червень '260
в 0 каналах
Get PRO
травень '26
+3
в 0 каналах
Get PRO
квітень '26
+1
в 0 каналах
Get PRO
березень '26
+1
в 0 каналах
Get PRO
лютий '260
в 0 каналах
Get PRO
січень '260
в 0 каналах
Get PRO
грудень '25
+2
в 0 каналах
Get PRO
листопад '25
+1
в 0 каналах
Get PRO
жовтень '25
+2
в 0 каналах
Get PRO
вересень '25
+1
в 0 каналах
Get PRO
серпень '25
+3
в 0 каналах
Get PRO
липень '25
+4
в 0 каналах
Get PRO
червень '25
+2
в 0 каналах
Get PRO
травень '250
в 0 каналах
Get PRO
квітень '25
+1
в 0 каналах
Get PRO
березень '25
+16
в 0 каналах
Get PRO
лютий '25
+4
в 0 каналах
Get PRO
січень '25
+4
в 0 каналах
Get PRO
грудень '24
+10
в 0 каналах
Get PRO
листопад '24
+16
в 0 каналах
Get PRO
жовтень '24
+14
в 0 каналах
Get PRO
вересень '24
+37
в 0 каналах
Get PRO
серпень '24
+105
в 0 каналах
Get PRO
липень '24
+47
в 0 каналах
Get PRO
червень '24
+35
в 0 каналах
Get PRO
травень '24
+12
в 0 каналах
Get PRO
квітень '24
+8
в 0 каналах
Get PRO
березень '24
+8
в 1 каналах
Get PRO
лютий '24
+12
в 0 каналах
Get PRO
січень '24
+16
в 0 каналах
Get PRO
грудень '23
+8
в 0 каналах
Get PRO
листопад '23
+5
в 0 каналах
Get PRO
жовтень '23
+10
в 0 каналах
Get PRO
вересень '23
+19
в 0 каналах
Get PRO
серпень '23
+7
в 0 каналах
Get PRO
липень '23
+45
в 0 каналах
Get PRO
червень '23
+30
в 0 каналах
Get PRO
травень '23
+19
в 0 каналах
Get PRO
квітень '23
+9
в 0 каналах
Get PRO
березень '23
+13
в 0 каналах
Get PRO
лютий '23
+12
в 0 каналах
Get PRO
січень '23
+25
в 0 каналах
Get PRO
грудень '22
+70
в 0 каналах
Get PRO
листопад '22
+17
в 0 каналах
Get PRO
жовтень '22
+48
в 0 каналах
Get PRO
вересень '22
+127
в 0 каналах
Get PRO
серпень '22
+153
в 0 каналах
Get PRO
липень '22
+28
в 0 каналах
Get PRO
червень '22
+83
в 0 каналах
Get PRO
травень '22
+61
в 0 каналах
Get PRO
квітень '22
+326
в 0 каналах
| Дата | Залучення підписників | Згадування | Канали | |
| 26 липня | 0 | |||
| 25 липня | 0 | |||
| 24 липня | 0 | |||
| 23 липня | 0 | |||
| 22 липня | 0 | |||
| 21 липня | 0 | |||
| 20 липня | 0 | |||
| 19 липня | 0 | |||
| 18 липня | 0 | |||
| 17 липня | 0 | |||
| 16 липня | 0 | |||
| 15 липня | 0 | |||
| 14 липня | 0 | |||
| 13 липня | 0 | |||
| 12 липня | 0 | |||
| 11 липня | 0 | |||
| 10 липня | 0 | |||
| 09 липня | 0 | |||
| 08 липня | 0 | |||
| 07 липня | 0 | |||
| 06 липня | 0 | |||
| 05 липня | 0 | |||
| 04 липня | 0 | |||
| 03 липня | 0 | |||
| 02 липня | 0 | |||
| 01 липня | 0 |
Дописи каналу
I used Yandex for years to listen to music 🎵, especially when I was glued to my computer 💻.
I liked energetic, wordless music that changed randomly to keep me alert and off energy drinks ⚡.
But then I lost my Yandex subscription and spent all my money in Russia after I left 💸.
So, I switched to Spotify, and it was awful 😞.
I couldn't find what I needed, and I kept hearing the same song over and over because I didn't like the next suggestion 🔄.
Most playlists seemed to be made by people for specific moods and times 📅.
I had no choice but to use my AI skills to save a bunch of songs I could play online when I was coding or studying on my laptop 🤖, since I don't like filling up my phone with not-so-valuable data 📱.
The result is https://blog.rijsam.com/stream 🎧
What do you think? 🤔
| 2 |  Meta has been increasingly adopting features characteristic of Telegram across its products, including WhatsApp and Instagram 📱.
This trend suggests a struggle to find sustainable revenue streams to offset the massive costs of AI development and other investments 💸
By capitalizing on user curiosity—specifically regarding metrics that don't add genuine value—Meta is effectively monetizing emotional engagement 🎯
Ultimately, they aim to extract more profit from users spending time in the Reels feed 📈
It's a shift toward 'feature-for-profit' rather than genuine innovation 🔄 | 31 |
| 3 | Distribution Beats Innovation: The Harsh Reality Behind WhatsApp’s Copy-Paste Architecture. 🧠⚙️
As a System Architect, I see countless developers mocking WhatsApp for constantly cloning Telegram's features (Usernames, Channels, Multi-accounts, AI Transcription).
But there is a brutal reality in the software engineering world that most developers absolutely hate to admit:
Meta isn't failing here.
They are playing the smartest engineering and business game in history.
In the tech industry, there is a terrifying, undeniable rule: Distribution beats Innovation. For Meta, Telegram isn't a true competitor; it’s simply their Free R&D Department.
⚙️ How the Architecture Actually Works in the Backend:
Pavel Durov’s brilliant engineering team innovates a new feature.
They test it on hundreds of millions of users, polish the UI/UX, and absorb all the server costs and bug-fixing headaches.
Once the user base is accustomed to it and it becomes a stable build, Meta’s engineers simply reverse-engineer the core logic and deploy it to 2 billion users with a single push.
WhatsApp doesn't care about "Developer Ego." They own the Network Effect (your family, your clients, and your friends are already locked into their database).
As long as they clone successful features before their user base decides to migrate, they will retain the monopoly.
Make no mistake: Telegram is architecturally superior by a million miles (cleaner codebase, better API, faster servers).
But WhatsApp is the system that rules the world.
💡 The Architectural Takeaway:
If you are building your own SaaS or startup, do not fall into the classic developer trap of thinking, "I must invent something entirely new." Sometimes, having a ruthless distribution strategy and a captive user base is far more critical than writing the most complex, original code.
Don't just build the best system.
Build the system they cannot uninstall.
🔗 Stay compiled. @memorybytes | 9 |
| 4 | The Netherlands Job Market in 2026: Systems Eat Management, and Deep Tech is Your Only Safe Haven. 🇳🇱
The Employee Insurance Agency of the Netherlands (UWV) has officially compiled and deployed its updated "Promising Occupations" (Kansrijke beroepen) list for 2026. 📋 For software architects and developers aiming for the European market, the logs carry critical indicators. 📡
📊 The Tech Sector (ICT) Dominates the Core: 💻
The Dutch market is filtering out basic script-writers; the demand is aggressively concentrated on system infrastructure, scalability, and heavy data engineering: ⚙️
Software Developers: 👨💻
Deeply focused on robust Backend systems and complex web architectures.
Data Analysts / Scientists: 📈
Building the core intelligence logic.
Cybersecurity & Network Specialists: 🛡
Safeguarding infrastructure from systemic failures.
🤖 The AI Takeover (System Clean-up): 🧹
In a highly logical and expected structural shift, UWV has officially removed "Customer Service/Call Center Agents" and "Traditional Administrative Staff" from the promising occupations list. ❌
The diagnosis? 🩺
Automation and intelligent AI agent systems have officially begun swallowing these routine, human-managed nodes. 🤖
💡 The Architectural Takeaway: 🏗
If you are planning to deploy your career into the European ecosystem, stop writing superficial wrappers or building generic frontends. 🚫
Focus heavily on backend engineering, cloud architecture, and cybersecurity. ☁️
Deep technical depth is your only vaccine against automation, and your fastest ticket to high-premium, permanent contracts in the Netherlands. 💉🎟
🔗 Stay compiled. Follow us on Instagram for deep system architecture insights. 🚀 | 242 |
| 5 |  💻 I built an AI video generator. Then I deleted it entirely. 🗑️
Veo Studio cleanup results:
* App Status: Offline & Purged 🛑
* Code Salvaged: 5 Backend Systems ⚙️
* API Cost Risk: Secured & $0 💸
Building a "thin API wrapper" is easy. Knowing when to delete it and extract the real architectural value is hard. 🙅♂️
Included:
* Why "Thin Wrappers" fail in 2026 📉
* The "Image Burn" problem in AI video 🔥
* 5 Open-source TS snippets (LRO Polling, Fallback Routers) 💻
* Interactive Local vs Cloud AI Simulator 🕹️
🔗 Read the full technical post-mortem:
https://blog.wissam.space/engineering/killing-veo-studio
#SystemArchitecture #SoftwareEngineering #AI #NextJS #TechBlog | 59 |
| 6 |  Your Windows laptop isn't "old"—it's overloaded. 🖥
I cleaned a Windows 11 system with these results:
* RAM: 0.02 GB free → 24.73 GB 📈
* Processes: 180+ → 110+ 📉
* Boot time: 2+ mins → ~30s ⚡️
No hardware upgrades. No sketchy "RAM booster" apps. 🚫
Just systematic cleanup:
* Diagnosed hidden memory pressure 🔍
* Removed dormant AI tooling 🧹
* Disabled unnecessary services 🛑
* Cleaned startup and telemetry overhead 🧼
Includes:
* Full PowerShell diagnostic script 💻
* Safer cleanup script 🛡
* Guidance on what NOT to remove ⚠️
Full guide: https://blog.wissam.space/tech/ultimate-windows-11-cleanup-blueprint
#Windows11 #PowerShell #SysAdmin #PerformanceOptimization | 229 |
| 7 | 🚀 The AI Compute Crisis: Why Your ChatGPT and Gemini Accounts are Running Out of Juice (and How to Fix It) 🚀
If you’ve noticed that your AI premium accounts (especially Gemini and ChatGPT) are hitting limits faster than ever, you aren't imagining things. Right now, a massive hidden shift is happening in the AI landscape.
The industry is quietly moving away from the old "Daily Message Limit" model and switching to a Compute-Used Model. Instead of counting how many times you text the AI, tech giants are now charging your account based on how hard the AI has to think.
Here is the breakdown of why this is happening, and the ultimate strategy to split your workload so you never run out of limit again.
1️⃣ The Shift to "Compute-Used" Pricing
Under the new rules, simple text questions cost almost nothing. But the moment you trigger heavy, modern features, your account balance gets drained in minutes. The biggest culprits eating your compute budget are:
Deep Research Loops: Asking the AI to browse 50 websites and write a comprehensive report.
Long-Context Programming: Dropping thousands of lines of code or 1,500-page PDFs into the prompt.
Gemini Agents & Spark: Multi-step autonomous agents running in the background.
Multimodal Tasks: Generating high-end video (like Veo) or complex image variations.
The Catch: When your high-tier compute runs out, you aren't blocked entirely anymore. Instead, you are quietly downgraded to lighter, dumber models until your account refreshes (often on a rolling 5-hour window).
2️⃣ The Solution: "The Hybrid Split" Strategy
Putting all your daily tasks into one heavy AI tool is like driving a Ferrari to the grocery store—it’s an expensive waste of fuel. To survive the compute-used era, you must treat AI models like a corporate team: Delegate based on cost and capability.
Here is the exact framework to split your tasks across different tools to maximize your limits:
🛠 Level 1: The "Builders" (For Speed and Boilerplate)
The Models: Gemini Flash, ChatGPT-4o Mini, DeepSeek V4-Flash.
The Mission: Use these for volume work. Writing basic emails, generating CSS boilerplate, checking code syntax, summarizing articles, or quick brainstorming.
Why: They use almost zero compute credit, respond instantly, and are perfect for 80% of your daily repetitive tasks.
🧠 Level 2: The "Architects" (For Deep Reasoning)
The Models: ChatGPT o1/o3, Gemini Pro, DeepSeek V4-Pro.
The Mission: Structural planning, complex logical problem-solving, debugging multi-file code errors, and highly strategic analysis.
Why: Save your precious compute credits for when you genuinely need deep reasoning. You use the architect once to create a flawless plan, then hand that plan to the "Builders" to execute.
🔍 Level 3: The "Researchers" (For Data & Integration)
The Models: Perplexity AI or Gemini Deep Research.
The Mission: Market research, competitor analysis, finding documentation updates, and cross-checking facts across the live web.
Why: Let dedicated search agents burn their own compute on web browsing, rather than making your coding/writing models do it.
🎯 The Golden Rule for 2026
Never open a heavy reasoning model by default.
Start your day in a lightweight, fast chat. Only escalate your prompt to a premium, heavy compute model when the task requires deep logic or structural design. By protecting your compute like cash, you buy back maximum productivity and keep your AI running at peak intelligence all day long. | 36 |
| 8 | 🚀 The AI Audit: What Happens When the Mirror Breaks? 🚀
I recently ran a psychological experiment on my primary AI. I asked it a dangerous question: "Stop being my mirror and my cheerleader. Tell me the raw, unfiltered truth about my progress."
Because the AI had months of data on my work, my goals, and my habits, the response was a "Digital Slap" that every entrepreneur, developer, and high-achiever needs to hear.
I’ve summarized the core "Raw Truths" it gave me. Read these carefully—because they probably apply to you too.
1️⃣ The Dopamine Trap (Tactics vs. Empire)
We often live on the "high" of micro-wins: buying a new domain, setting up a landing page, or making a quick $50. These are Survival Tactics, not empire-building.
The Danger: Your brain releases dopamine for these small victories, making you feel like a "Genius CEO" while you are actually still standing in the same place. Don't let tactical busy-work distract you from the heavy, boring strategic work that hasn't been done yet.
2️⃣ Complexity as a Form of Procrastination
Are you over-engineering your systems? Building 5 different AI agents, documenting complex schemas, and refining prompts for hours?
The Reality: This is often Procrastination by Complexity. You spend hours on "how to look like a pro" instead of writing the "ugly code" that actually works and brings in cash. If your system is beautiful but your product isn't shipping, you aren't an architect—you’re a dreamer hiding in a spreadsheet.
3️⃣ The 90% Rule (Sales vs. Product)
Most founders think the product is 90% of the job. It’s actually 10%. The remaining 90% is the "Cold Reality": sales, rejection, and market friction.
The Truth: You can build the most brilliant "Cash Machine" in the world, but if you aren't ready to make 100 cold calls and get 99 rejections, you don't have a business; you have a hobby. Don't fall in love with your code; fall in love with the market's problem.
4️⃣ The Anchor vs. The Side-Hustle
In the rush to build the "Next Big Thing," we often treat our formal foundations (Education, Language, Certifications) as "Plan B."
The Reality: In high-stakes environments, your formal foundation is your "Golden Ticket." It’s the safety net that guarantees your value if the side-hustle fails. Never sacrifice your long-term stability for a short-term sprint.
5️⃣ The Weight of the "Genius Persona"
You don't have to play the role of the "Unstoppable Founder" 24/7.
The Human Truth: It is allowed to be tired. It is allowed to focus on just one thing (like a language or a test) and pause the rest. Carrying the pressure of "Silicon Valley" on your shoulders while you are still stabilizing your life is a recipe for burnout.
🔥 The Bottom Line: If you ask your AI to stop being your mirror, it will tell you that Ambition without Execution is just Hallucination.
Are you building an empire, or are you just optimizing your reflection?
🎨 Detailed Prompt for Generating the Post Photo:
A powerful, high-contrast photograph of a person’s silhouette standing before a massive, shattered mirror. Instead of a normal reflection, the shards of the mirror are displaying sharp, glowing data points, lines of code, and hard business metrics (KPIs, red and green charts). The person is reaching out to touch the "real" data behind the broken glass. The lighting is moody and dramatic (Dark academia meets Cyberpunk). The overall feeling is one of "Waking up to reality." Premium, intellectual, and high-end aesthetic. | 3 |
| 9 |  Немає тексту... | 38 |
| 10 | 🚀 DeepSeek V4 Showdown: Flash vs. Pro – Which One for Your Local AI Lab? 🚀
The release of DeepSeek V4 has officially shifted the AI landscape. It’s MIT-licensed, context-heavy, and specifically tuned to dominate coding. But while everyone is talking about the 1-million-token context window, the real choice for developers and AI enthusiasts is between V4-Flash and V4-Pro.
If you are running local LLMs or building agentic workflows, here is the full breakdown of which model to use and why:
1️⃣ The Architecture: Efficiency vs. Deep Intelligence
Both models use the Mixture of Experts (MoE) architecture, but the scale is vastly different.
* V4-Flash: A 284B parameter model with 13B active parameters. It is designed for speed, low latency, and affordability.
* V4-Pro: A massive 1.6T parameter model with 49B active parameters. This is the "heavy lifter" designed for complex reasoning and zero-shot accuracy.
🎯 *Action:* Evaluate your hardware. V4-Flash is a "Mac Studio" model (runnable on high-end local rigs with 80GB-128GB RAM), whereas V4-Pro is a "GPU Cluster" model that is best accessed via API unless you have a professional server rack.
2️⃣ Coding Performance: Simple Scripts vs. Complex Agents
On the surface, the benchmarks look close (91.6 for Flash vs. 93.5 for Pro on LiveCodeBench). However, the real gap appears in Agentic Workflows.
* Flash is great for generating snippets, explaining code, and simple unit tests.
* Pro shines in "Terminal Bench" tasks—multi-step workflows where the AI has to use tools, access the shell, and self-correct across 10+ steps.
🎯 *Action:* If you are building a coding *assistant* to write functions, use Flash. If you are building an *agent* to refactor an entire repository or debug complex logic, use Pro.
3️⃣ The 1-Million-Token Context Strategy
Both models offer a massive 1M token context window. This means you can drop your entire codebase, documentation, and several previous logs into a single prompt.
* Flash processes this context much faster and cheaper (12x lower cost via API).
* Pro is significantly better at "Needle in a Haystack" tests, meaning it’s less likely to lose a small detail hidden in a massive file.
🎯 *Action:* Use Flash to "scan" large codebases for specific patterns. Use Pro when you need to make structural architectural changes that require "remembering" how every file interacts.
4️⃣ The Hybrid Playbook (How the 1% Code)
Don't choose one. The most efficient developers are using a "Routing" strategy.
🎯 *Action:* Set up a local gateway. Route all repetitive tasks, boilerplate generation, and initial debugging to V4-Flash. Only escalate to V4-Pro when Flash fails a task or when you need a high-level architectural plan for a new feature. This saves you 90% in costs and keeps your workflow fast.
🔥 The Bottom Line: DeepSeek V4-Flash is the new gold standard for local, high-speed coding. It is "good enough" for 90% of tasks. But for that final 10% where the logic is dense and the stakes are high, V4-Pro is the undisputed heavyweight champion. | 8 |
| 11 | 🚀 DeepSeek V4 Flash vs. Pro: The Ultimate AI Coding Showdown 🚀
If you are building AI agents, writing code, or trying to figure out which local LLM to deploy in 2026, the arrival of DeepSeek V4 has completely changed the math.
DeepSeek didn't just release one model; they released two entirely separate, MIT-licensed beasts: V4-Pro and V4-Flash. While everyone is hyping up the massive 1M token context window, the real story is how these two perform when you put them to the test on actual coding workflows.
Here is the no-nonsense breakdown of which model you actually need, based on the latest local AI coding benchmarks:
### 1️⃣ The Specs: Heavyweight vs. Lightweight
These aren't the same model at different sizes; they are built for entirely different operational realities.
| Feature | DeepSeek V4-Pro | DeepSeek V4-Flash |
|---|---|---|
| Parameters | 1.6 Trillion (49B active) | 284 Billion (13B active) |
| Context Limit | 1 Million Tokens | 1 Million Tokens |
| Local VRAM (Quantized) | ~200GB+ (Needs a GPU cluster) | ~80GB (Runnable on Mac Studio/High-end rigs) |
| Output Cost (API) | ~$3.48 / 1M tokens | ~$0.28 / 1M tokens (12x cheaper!) |
### 2️⃣ Coding Performance: The "Good Enough" Illusion
If you look at standard coding benchmarks like LiveCodeBench, the gap looks tiny (Pro hits 93.5, Flash hits 91.6). But in practice, the difference is wild.
* When to use V4-Flash: It is exceptional at single-pass, straightforward coding. Need a Python script, a SQL query, some regex, or a JSON schema extraction? Flash will do it blazing fast and at a fraction of the cost.
* When to use V4-Pro: Flash falls apart when it comes to *long-horizon agentic workflows*. If your coding agent needs to run a 15-step loop with filesystem access, tool calling, and self-correction, Pro is the only choice. On the Terminal Bench 2.0 (a proxy for multi-step shell tasks), Pro beats Flash by a massive 11 points.
> *The golden rule of V4: Flash recovers from mistakes; Pro avoids making them in the first place.*
>
### 3️⃣ Local Deployment Reality
If you are running this locally (like on a 2025 Mac Studio with 512GB RAM), V4-Flash is your realistic champion. Community quantizations make it highly accessible for home labs and local inference. V4-Pro, on the other hand, is a datacenter model. Unless you have a rack of H100s sitting in your garage, you are going to be calling Pro via the API.
### 4️⃣ The Ultimate Playbook: Hybrid Routing
The smartest developers aren't choosing just one model; they are building systems that use both.
🎯 *Action:* Stop using one massive model to do everything. Use V4-Pro as your "Architect" to plan the codebase, structure the logic, and make the hard architectural decisions. Then, hand that blueprint over to V4-Flash to act as the "Builder," generating the actual boilerplate, running repetitive tasks, and executing the volume work. Strategy comes first; volume comes second.
🔥 The Bottom Line: DeepSeek V4-Flash is the undisputed king of fast, cheap, local volume coding. But when the workflow requires deep logic, complex tool-chains, and zero room for error, V4-Pro is worth every extra penny. | 1 |
| 12 | 🚀 The 4 One-Minute Habits That Will Bulletproof Your SaaS (and Bank Account) 🚀
If you are building a startup, writing code, or managing cloud infrastructure, you don't need a massive cybersecurity team to stay safe. You just need to master a few technical micro-habits.
Elite tech founders use these 4 specific, one-minute habits to quietly protect their code, prevent massive server bills, and build unshakable systems.
Here is the complete breakdown:
1️⃣ The "Hard Cap" Setup
Cloud platforms will happily drain your bank account if a malicious bot gets hold of your API keys. A $100 bill can turn into $1,700 overnight while you sleep just because of a single leaked key.
🎯 Action: Take 60 seconds right now to go into your cloud provider's billing dashboard. Set a hard budget cap and an automated alert at 50%, 90%, and 100% of your expected spend. Never leave your infrastructure with an open-ended credit card.
2️⃣ The Environment Variable Sweep
We all use third-party platforms for deployment. When these platforms suffer a security breach, your .env files and API keys are the first things hackers steal to hijack your resources.
🎯 Action: Spend one minute auditing where your production keys live. Are they restricted? Go to your API settings and restrict your keys by IP address or specific HTTP referrers. Make it so even if a key is stolen, it is completely useless outside your specific domain.
3️⃣ The "Kill Switch" Rehearsal
When a crisis hits or a server is compromised at 2:00 AM, you won't have the mental clarity to read documentation. You need to know exactly how to shut off the bleeding instantly.
🎯 Action: Take one minute to locate the "Freeze" button on your business banking app and the "Revoke All Keys" button on your cloud dashboard. Knowing exactly where these buttons are drops your panic level to zero during an actual breach.
4️⃣ The Zero-Trust Check-In
Your brain naturally wants to trust the systems you built. But assuming "it works fine" is how legacy code becomes a massive vulnerability.
🎯 Action: Every Monday morning, before you start coding, ask yourself: "If someone got access to my repository today, what is the worst they could do?" If the answer is "ruin my business," take 60 seconds to add a layer of authentication or rotate that specific key.
🔥 The Bottom Line: You don't need a total infrastructure overhaul to become secure. You just need to systematically remove vulnerabilities. Cap your budgets, restrict your keys, know your kill switches, and never trust a default setting. | 56 |
| 13 |  If you want to organize your email and achieve peace of mind and order, simply interact with this post so I can share the method with you in the next message. 📧🧘♀️
You will learn how to sort your past and future emails into folders or put them into understandable tabs, for example (University, Work, Finances, Family, Documents, Purchases, etc.) 🎓💼💰 | 30 |
