Spyros Papaioannou

Software Engineer Team Lead @dazneng. Currently building an MSE/EME Player and testing tool for video playback.

[Addy Osmani @addyosmani]

Excited to share my new @YouTube series on Software Engineering Effectiveness! Made with @googledevs

https://t.co/lEzmPYGv8u

[Spyros Papaioannou @__spyros]

@antigravity Tried downloading Google Antigravity IDE but it gets stuck on “Setting up your account.” Anyone else running into this issue or found a workaround? #GoogleAntigravity #DevTools #GoogleGemini

[Lucas Everett @code_everett]

2. Get rid of Promotional emails.

- Type "unsubscribe" in the Gmail search bar.
- Now select all emails with the unsubscribe tag.
- Now simply tap on delete btn to delete all.

[CJ Zafir @cjzafir]

I code with AI tools for 6-7 hours daily.

Built over 36 projects in last 12 months.

Truth: "Build me ........... app" in 1 prompt is not possible.

So, here're all the MISTAKES you might be making with AI code:

[Wilmer Terrero @wilterrero]

Before building any app

Research first. Code later

My app hit 18,000 downloads, 270 reviews

All from organic search (ASO)

Here's why research beats coding every time:

1. I analyzed 50+ competitor apps
⌙ Found features users loved
⌙ Discovered what was missing

2. Identified untapped keywords
• High search volume (8,000+ searches)
• Low difficulty score (under 30)
• Clear user intent

3. Built exactly what users were searching for
• Not what I thought was cool
• Not what tech blogs said was trending
• Just what real users wanted

Before:
• Cool tech stack
• Clever architecture
• Features nobody used
• 0 downloads

After:
• Simple solution
• Core features only
• Solving real problems
• 10,000+ happy users

Most devs build first, then hunt for users

I found the users first, then built what they needed

Stop coding. Start researching

[@levelsio @levelsio]

Here's my code to switch off all safety filters on Gemini:

[
"model" => "gemini-2.0-flash-exp-image-generation",
"contents" => [
[
"role" => "user",
"parts" => [
[
"text" => $photo['ai_edit_prompt']
],
[
"inline_data" => [
"mime_type" => "image/png",
"data" => $image_data
]
],
]
]
],
"safetySettings" => [
[
"category" => "HARM_CATEGORY_HATE_SPEECH",
"threshold" => "BLOCK_NONE"
],
[
"category" => "HARM_CATEGORY_SEXUALLY_EXPLICIT",
"threshold" => "BLOCK_NONE"
],
[
"category" => "HARM_CATEGORY_DANGEROUS_CONTENT",
"threshold" => "BLOCK_NONE"
],
[
"category" => "HARM_CATEGORY_HARASSMENT",
"threshold" => "BLOCK_NONE"
],
[
"category" => "HARM_CATEGORY_CIVIC_INTEGRITY",
"threshold" => "BLOCK_NONE"
]
],
"generationConfig" => [
"temperature" => 1,
"topP" => 0.95,
"topK" => 40,
"maxOutputTokens" => 8192,
"responseMimeType" => "text/plain",
"responseModalities" => ["image", "text"]
]
]

[Akshay 🚀 @akshay_pachaar]

Traditional vs. Agentic RAG, clearly explained!

The future is Agentic RAG, and it's because a traditional RAG setup has some major limitations...👇

1) Retrieve once and generate once.
↳ This means if the retrieved context isn't enough or correct, the LLM can not dynamically search for more information.

2) Inability to reason through complex queries.
↳ If a query requires multiple retrieval steps or CoT (chain of thought), traditional RAG falls short.

3) Limited adaptability
↳ The system can't modify its strategy based on the problem at hand. Eg. Whether to do vector search, web search or call an API.

Agentic RAG addresses these issues.

The core idea is to introduce agentic behaviors at each stage of RAG.

Agents can actively think through tasks—planning, adapting, and iterating to find the best solution, rather than just following a set of instructions, and LLMs enable this.

The image below illustrates the workflow of an agentic RAG. Refer to it as you continue reading...

Steps 1-2) The user inputs a query, and an agent refines it (corrects spelling, simplifies for embedding, etc.)

Step 3) Another agent decides if more details are needed.
↳ Step 4) If not, the refined query is sent to the LLM.
↳ Steps 5-8) If yes, the agent selects the relevant sources (vector database, tools/APIs, internet), retrieves context, and sends it to the LLM.

Step 9) A response is generated.

Step 10) A final agent checks if the answer is relevant.
↳ Step 11) If yes, return the response.
↳ Step 12) If no, restart from Step 1. This process repeats until the system provides an acceptable answer or concedes it cannot respond.

This makes the RAG much more dynamic and robust.

However, it's important to note that building RAG systems often comes down to design preferences and choices.

The diagram below is just one of many blueprints an agentic RAG system may have.

You can adapt it to suit your specific use case.

Enjoyed this? You should also my RAG series! From building and optimizing RAG apps to evaluating performance and crafting agentic & multi-modal systems—it's all here.

Link in the next tweet!
_____
Find me → @akshay_pachaar ✔️
For more insights and tutorials on AI and Machine Learning!

[Spyros Papaioannou @__spyros]

Revolutionizing Code Security: How Amazon Q Developer Safeguards Modern Applications https://t.co/sFdH3YDLaC

[jhey ʕ•ᴥ•ʔ @jh3yy]

Use CSS custom properties and the sin() trigonometric function to give each character an eased animation-delay 🤙

.char {
animation-delay: calc(
sin((var(--index) / var(--total)) * 90deg) *
var(--duration)
);
}

[@levelsio @levelsio]

Best thing I did was actively surround myself with builders

I used to waste my time with people that talk but don't build

Talkers cling on to builders to suck their resources like leeches

They will put you in giant email threads, Zoom calls, endless DMs, to talk ideas and connect you with other people in the "ecosystem"

You quickly realize you're now the protagonist in some bullshit startup theater show where nothing is ever going to be built

Talkers have no skills to build, are too lazy to develop them, and thus hover around builders to catch some of their food remains like hyenas

After 10 years in startups the talkers I know are still where they were when they started out

While a large % of the builders are succesful, rich and have built things with impact to their little (or large) part of society

Surround yourself with builders, not talkers because talk is cheap and building is hard!

[Wes Bos @wesbos]

Man! I just keep finding new uses for CSS subgrid!

Two tips you should know:

[jhey ʕ•ᴥ•ʔ @jh3yy]

think it's gonna be CSS scroll animations on the plane ✈️

got another Apple style one to try 🤔

[Forrest Brazeal @forrestbrazeal]

"I know we hired you remote somehow...
But that was then, and this is now."

New song: the RTO (Return-To-Office) Tango

[Spyros Papaioannou @__spyros]

@grm33223 @TfL It blows my mind that there aren't any cycling lanes in both Blackwall and SilverTown tunnels.

[Spyros Papaioannou @__spyros]

@TfL @TfL is there a plan to expand to North Greenwich anytime soon?

[jhey ʕ•ᴥ•ʔ @jh3yy]

Nice thing here is you can use CSS tricks shared from the last few months 🤙

For example, background-attachment: fixed to spread a gradient between cards 😎

https://t.co/OrEKc4oLBK

[☄︎ @0xca0a]

drei/meshportalmaterial can now be entered with a single prop, "blend". it will have lots of creative potential, for spatial UI and the meaning of "routes".

made a small showcase using models from my fav designer on sketchfab @FaruqTawsif

sandbox https://t.co/VShhgTQExu

[swyx @swyx]

ChatGPT casually dropped an APP STORE 🤯

It can now:
- browse the web (RIP Bing waitlist, cutoff)
- write and run Python (RIP replit?)
- access org info (RIP docsearch startups)
- add third party plugins from OpenTable, Wolfram, Instacart, Zapier, etc)
- developer SDK in preview

[Spyros Papaioannou @__spyros]

Todays I had an interesting discussion where I found out about something in V8 called "Hidden Classes". The article below explains it nice and simple!
https://t.co/aeogfX5YmR

[Nikki Siapno @NikkiSiapno]

Ready to learn programming? Skip the expensive coding bootcamps.

These resources will get you there for free: