text
stringlengths
1
3.81k
text_original
stringlengths
1
3.8k
labels
listlengths
0
9
message_id
stringlengths
4
57
chat_name
stringclasses
19 values
is_reply
bool
2 classes
Can you give an example of tasks? Is it mainly classification?
את יכולה לתת דוגמא למשימות? זה בעיקר קלאסיפיקציה?
[ 0, 1 ]
$MLxWr3pDHbgsZo03VwLkl9_YglKUPgBVZa6cqG7nm10:beeper.local
LangTalks Community 4
true
I tried to implement OCR for a friend's business too, for Hebrew handwriting that's "ugly/unreadable." After a few iterations and experiments, the clearest thing is it's better to just invest in an iPad for such a worker instead of trying to analyze the handwriting 😅
ניסיתי להטמיע לאיזה בית עסק של חבר גם ocr של כתב ״מכוער/לא קריא״ בעברית. אחרי מספר איטרציות ומשחקים, הדבר הכי ברור שכדאי פשוט להשקיע בipad עבור עובד כזה במקום לנסות לנתח את הכתב 😅
[ 0, 2, 4, 5, 6, 10, 11 ]
$A199igPRpQa4US3bjtidf9NHxAVYV_rj8ZcNpoasdag:beeper.local
Squid Club
false
חחחחח יש מלא סיכוני אבטחה ולכן חשבתי שלא מומלץ
חחחחח יש מלא סיכוני אבטחה ולכן חשבתי שלא מומלץ
[ 0, 6 ]
$Ix-fDHkNKP7UwsR_UrIyCesRrTWOZvy5Sc55yr9aYWU:whatsapp.com
MCP Israel
false
* By the way, there are 4 if you adopted `--dangerously-skip-permissions` :)
* אגב, יש 4 אם אימצת את `--dangerously-skip-permissions` :)
[ 0, 5, 6 ]
$1fBwVEfdCRMp0oWAPOIzm6a7HEJnQEG8TOfkCElcic4:beeper.local
Squid Club
false
Basic
בסיסית
[ 9 ]
1079
n8n Israel - Main 3
true
So it basically tells it not to ask me anything at all and just keep working continuously?
זה בעצם אומר לו שלא ישאל אותי כלום ושום דבר ופשוט יעבוד ברציפות?
[ 0, 1 ]
$27xV6i39UbDTW2XRR278VBVF12q2lqRmjDojn-O_rBk:beeper.local
Squid Club
true
Some have started keeping Shabbat. It helps them relax from the hype.
יש כאלו שהתחילו לשמור שבת. עוזר להם להרגע מהייפ
[ 4, 6 ]
$9BxAGrYQFmJbYdvyS7AsXVZWjMptBti9VK7T18PXvXQ:whatsapp.com
AI Transformation Guild
false
Yes, it works in both modes. The MCP app simply adds UI resources without changing the underlying server installation. To verify it's working, just check if the 'interactive tools' are visible in your MCP configuration.
Yes, it works in both modes. The MCP app simply adds UI resources without changing the underlying server installation. To verify it's working, just check if the 'interactive tools' are visible in your MCP configuration.
[ 0, 5 ]
$8UKK-bHupJE0rxv-d9awWDDrnGzFN7op3bZ7ApMHWZo:whatsapp.com
MCP-UI
true
* The flow is like this: They upload the files to your app, you in turn upload them to Gemini’s file store. Separate it so that each user’s files are stored separately from the others When you send their requests to the LLM (I also recommend Gemini here, relatively cheap and very high quality), you attach a tool call t...
* הפלואו הוא כזה: הם מעלים לאפליקציה שלך את הקבצים, את בתורך מעלה את זה ל file store של ג׳מיני. מפרידה את זה כך שהקבצים של כל יוזר ישבו בנפרד משאר היוזרים כשאת שולחת את הבקשות שלהם ל llm (ממליץ גם כאן על ג׳מיני, יחסית זול ומאוד איכותי), את מצרפת tool call ל file search וכך הוא יתייחס למסמכים שהיוזר העלה
[ 0, 4, 5, 10 ]
$f20A--wmo_SRrBYnbvvPRhWYcqOjqY6DHRLwlp0UgPc:beeper.local
MCP Israel
false
Hey, Has anyone here built and published a GPT App? 🙂 Would love a quick consultation
היי, למישהו פה יצא לבנות ולפבלש GPT App? 🙂 נשמח להתייעצות קטנה
[ 0, 1, 11 ]
$XfPUEkhdSss4mITP8c7K_R9eynvHFjQN8_PZO27q1po:beeper.local
MCP Israel
false
The a2a protocol solves a different problem (not privacy): the agent on the enterprise side is internal, so it's the data controller and responsible for what leaves the enterprise. It can mask anything confidential and make sure only what it wants to expose gets sent as input to the third-party supplier. This is great ...
פרוטוקול a2a פותר בעיה אחרת (לא של פרטיות) : האייג'נט בצד של האנטרפרייז הוא פנימי internal ולכן הוא ה data controller ואחראי למה שיוצא מאירגון האנטרפרייז. הוא יכול לעשות mask לכל מה שסודי ולוודא שרק מה שהוא רוצה לחשוף יעבור לקלט אצל ספק בצד ג'. זה מעולה ב GDPR כי זה מצמצם העברת מידע אישי לגורמים חיצוניים (כאלה שאולי את...
[ 0, 5, 10, 11 ]
$C2ex7kjea9ZIGjbCyL0VWUOL-bs0mMefZRyCQA1EBNk:beeper.local
Squid Club
false
https://chat.whatsapp.com/B9zw49KxddGGD1CZI7fLKy
https://chat.whatsapp.com/B9zw49KxddGGD1CZI7fLKy
[ 3, 9 ]
$9NtW2h8T3pCFqvXxTSgaB9lLFFSXKOzMLvNsHaejLdg:whatsapp.com
MCP Israel
true
Does anyone know or use langfuse in production? Can you recommend or diss? Or alternatively, can you recommend better alternatives? (from experience)
מישהו מכיר או משתמש ב- langfuse בפרודקשן? ויודע להמליץ או דיס? או לחילופין יודע להמליץ על אלטרנטיבות טובות יותר? (מנסיון)
[ 0, 1, 11 ]
1184
MCP Israel
false
Here you go 🤓
בבקשה 🤓
[ 6, 9 ]
$bKerX6TpQ0EACeLqIYHraGUc2oy13NERE5bRDkiGMTE:beeper.local
Squid Club
false
I'd be happy too
אשמח גם
[ 9, 14 ]
$ezJkgF30FQRGqxjo2kZGbrkcbr-6q8dOyp8jKC9bqjc:beeper.local
Squid Club
true
It will definitely be shared :) We didn't hire a paid filming company for nothing :)
תופץ בהחלט:) לא סתם הבאנו חברת צילום בתשלום:)
[ 6, 7 ]
$WMURw7223Zn4hb24KM39_0zlzcvDvmTF0ub4qfwzjtg:whatsapp.com
n8n Israel - Main 3
true
So all the big companies work locally?
אז כל החברות הגדולות עובדות לוקאלית?
[ 0, 1 ]
$Kfxf_DY3xb74CsUUf8Y_MuBtkCwj8A9wix627PDQ_94:whatsapp.com
n8n Israel - Main 3
true
* The flow is like this: They upload the files to your app, you in turn upload them to Gemini’s file store When you send their requests to the LLM (I also recommend Gemini here, relatively cheap and very high quality), you attach a tool call to file search and that way it will refer to the documents the user uploaded
* הפלואו הוא כזה: הם מעלים לאפליקציה שלך את הקבצים, את בתורך מעלה את זה ל file store של ג׳מיני כשאת שולחת את הבקשות שלהם ל llm (ממליץ גם כאן על ג׳מיני, יחסית זול ומאוד איכותי), את מצרפת tool call ל file search וכך הוא יתייחס למסמכים שהיוזר העלה
[ 0, 4, 5, 10 ]
$TrdavRtj_YXhI8LK0rbz-vm-2q83G4X-t_0ZOfkwwCM:beeper.local
MCP Israel
false
Did you also try the bmad method? How does it compare to it?
ניסית גם את bmad method? איך הוא ביחס אליו?
[ 0, 1 ]
$d6r_tX1-ziJBtjm7J5s_8OAtIDEEc08Fhq4kLTnIr0I:beeper.local
Squid Club
true
About half a year ago, I had a crazy weekend (because of bad settings in CLAUDE.MD), and I ended up wasting an unreasonable amount of tokens compared to normal usage. CC-USAGE reported the following numbers at the time:
לפני כחצי שנה, היה לי סופש מוגזם (בגלל הגדרות רעות של ה CLAUDE.MD), ונוצר לי בזבוז טוקנים לא סביר בעליל לעומת שימוש די רגיל. CC-USAGE טען בזמנו את המספרים הבאים:
[ 0, 2, 10 ]
$tUBDiBOT1OsJQM4mPyqbghxDJQC6BfJoXbfWcR8UtjI:beeper.local
Squid Club
true
Ran Barzik once wrote about them and recommended them as something safe It's been over two years since then so maybe it changed Although I heard him not long ago say he works with their terminal on a podcast but I'm not sure when the episode is from 😅 I'll look for it
רן ברזיק פעם רשם עליהם והמליץ עליהם בתור משהו בטוח עברו מאז מעל שנתיים אז ייתכן וזה ישתנה אם כי שמעתי אותו לא מזמן אומר שהוא עובד עם הטרמינל שלהם בפודקאסט אבל אני לא בטוח ממתי הפרק 😅 אחפש
[ 2, 10, 14 ]
$wcbPPv-R9f4i8drXf-T_wjYDZWdAZ45-G4o_3sAnt0I:beeper.local
Squid Club
true
Crazy
משןגע
[ 9, 12 ]
$pBHbA-nFmKGCegHGnExZ4yXv2JExv-yFgAXYGbkJhto:beeper.local
Squid Club
true
Basically the organization is responsible for what leaves it, it controls the internal agent. The third party (the supplier) will get the information needed as input and use it to return a response according to that supplier's capability (for example, a crewai SaaS that translates documents, like in another app I have)
למעשה הארגון אחראי למה שיוצא ממנו, הוא שולט על האייג'נט הפנימי. הצד ג' (הספק) יקבל את המידע שדרוש כקלט וישתמש בזה כדי להחזיר תשובה בהתאם ליכולת של אותו ספק (למשל SaaS של crewai שמתרגם מסמכים, כמו באפליקציה אחרת שיש לי)
[ 0, 5, 10, 11 ]
$WgRagBksKClib5imM535711G0iopJMy-ffbstGex9Xc:beeper.local
Squid Club
false
New episode! Measuring impact and effectiveness of using AI for code writing. Will be especially interesting for the development managers among you https://www.linkedin.com/posts/lee-twito_64-ai-coding-metrics-liad-elidan-milestone-activity-7436636913337319424-9BOb
פרק חדש! מדידת אימפקט ואפקטיביות של שימוש בAI לכתיבת קוד. יעניין במיוחד את מנהלי הפיתוח שבניכם https://www.linkedin.com/posts/lee-twito_64-ai-coding-metrics-liad-elidan-milestone-activity-7436636913337319424-9BOb
[ 0, 3, 7, 11 ]
$uEGx4CcIaI0i_VDrAiDJMjwGZ-mt7nepkr4OXhjmBo8:beeper.local
LangTalks Community 2
false
https://openrouter.ai/api/v1/models
https://openrouter.ai/api/v1/models
[ 0, 3 ]
$s8FLbq2rLBenLoaXrIRgkH7bJLDWfNe4P5TZ9D8Fm3Q:beeper.local
Squid Club
false
And specifically, I need this skill for a system I'm building for lawyers (some documents come in handwriting mixed with print)
ובאופן ספציפי, אני צריך את הסקיל הזה למערכת שאני מייצר לעו"ד (חלק מהמסמכים מגיעים בכתב יד משולב עם דפוס)
[ 0, 2, 10 ]
$SlogwC6NlsRqpZC-gFghwGQqCa5W2lWThGhcV2GhPkI:beeper.local
Squid Club
false
Yours?
שלך?
[ 1 ]
$yPYJh7vp6-HPK_Zq3-gAj_VkXXXWjiAkt9SE-4eXKwY:whatsapp.com
MCP Israel 2
true
זה נכון לגבי archive, אבל חלק מהמאמרים, בעיקר תזות, לא שם. אני אסתכל על מה ששלחת. תודה!
זה נכון לגבי archive, אבל חלק מהמאמרים, בעיקר תזות, לא שם. אני אסתכל על מה ששלחת. תודה!
[ 9, 14 ]
$MxIVcITIpXeCtAvVWYCFBwv0ofBwpHrFG_szxPaRxko:beeper.local
LangTalks Community 3
true
Hey everyone, welcome to all the new members. We're getting pretty close to 800 people in the group and it's super exciting. I've wanted for a long time to organize a physical meetup that's different from the usual. No lectures, no sponsors, no time wasting. Pure learning and enrichment. I'm looking for someone from th...
היי כולם, כל מה שהצטרף בורכים הבאים. אנחנו מתקרבים בצעדים די גדולים ל800 איש בקבוצה וזה סופר מרגש. אני כבר המון זמן רוצה להרים מפגש פיזי שיהיה מפגש שהוא שונה בנוף. בלי הרצאות ספונסרים בלי בזבוז זמן. נטו לימוד והעשרה. אני מחפש מישהו מהקהילה שיש לו ניסיון ויש לו חשק לעזור לי להרים את זה אנא פנו אליי בפרטי. סהר. ** הסיבה...
[ 1, 7, 10, 11, 12 ]
$yo1TfRt9TYhrXfL-eXdyV9zx4Xx4RpKz3K_h2VJnQW4:beeper.local
Squid Club
false
And does n8n run locally or on a server? I mean, does it need my own server or not?
וn8n הוא רץ לוקאלי או בשרת ? הכוונה היא האם הוא צריך שרת שלי או לא?
[ 0, 1, 11 ]
$04JGxcEgbnTFwUT6fvNPJ4iErXFj7X3ro-EB5jJsR08:beeper.local
n8n israel - Main 1
false
* That's *exactly the project* - for the model to classify the answers into 8 known categories. And in addition, to have a classification of "belongs to the category but is new in it". There is no such data.
* זה *בידיוק הפרויקט* שהמודל ידע לסווג את התשובות ל-8 קטגוריות מוכרות. ובנוסף שיהיה סיווג של "שייך לקטגוריה אך חדש בה". אין דאטה כזה
[ 0 ]
$zYbdFO9oWv7IX5TPsECpZRyj_xYb2QUC1KNs98WAAzs:beeper.local
LangTalks Community
false
😂😂 Just wait till you see my warp... I just do LB for the tokens
😂😂 חכה חכה שתראה את הwarp שלי .. אני עושה LB לזי׳טונים פשוט
[ 0, 2, 6 ]
$88aXKyzod7AhSSNdXjfbsqgtiJDE389dLklXjvVjoq8:beeper.local
Squid Club
true
Wow, good luck with the amount of curses you'll get 😅
ואי ואי בהצלחה עם כמות הקללות שתקבל 😅
[ 6 ]
$nyetjayl10RCTrDqHwN_fXVgjo4QOhLC9hPZkpCBj10:whatsapp.com
AI Transformation Guild
true
Is there a link to the tool?
האם יש לינק לכלי?
[ 1 ]
$1UzuNrmVg9gbdgxAdUyl3Knk4NoRk0uxMMVJeVTZFXI:beeper.local
MCP Israel 2
true
I work with Lanable but I'm always testing other tools too. https://app.superdesign.dev also does a good job. The right way to work for me is to create a mockup or general design direction in these tools. Import the git repo to my computer and from there Claude Code picks up the design and continues it. If you want to ...
אני עובד עם לאנאבל אבל גם כל הזמן בוחן כלים אחרים. https://app.superdesign.dev גם עושים עבודה טובה. הדרך הנכונה לעבוד מבחינתי זה לייצר איזה מוק או כיוון כללי של עיצוב בכלים האלו. לעשות import לgit repo אליי למחשב ומשם קלוד קוד כבר תופ את העיצוב וממשיך אותו. אם רוצים לייצר ממשקים פונקציונאליים ממליץ לעבוד עם shadcn זה מ...
[ 0, 2, 3, 4, 5, 10 ]
$9N9jXZE5VFI3wIw0-In--0iIMo75PelHUPP5opHzjTQ:beeper.local
Squid Club
true
How many work hours do you manage to get out of it, roughly?
כמה שעות עבודה אתה מצליח להוציא עם זה בערך?
[ 0, 1 ]
$veZVBBMPYtCeVOf1fxzUMJHU1bHpcOzgQNxmDEJ22Gk:beeper.local
Squid Club
true
This is an English text, financial, but there's no expectation to do calculations, it's more analytical.
זה טקסט באנגלית, פיננסי אבל אין ציפייה לבצע חישובים, יותר אנליסטי
[ 0, 2 ]
$WeGfcDJ470Npq8b0MBTReesZoyOlnYWJkyJT5OoImhA:beeper.local
LangTalks Community
false
Worth checking, maybe they also have OCR for cursive handwriting
שווה לבדוק אולי יש להם גם OCR לכתב עגול
[ 0, 5, 11 ]
$OW0dresKFYsiJrOIx_3F0d47vq09NzabPZ85QXQFho4:beeper.local
Squid Club
false
But take the first one, push it to Claude Code and tweak it a bit
אבל קח את הראשון, דחוף אותו לקלוד קוד ותשפצר אותו קצת
[ 0, 5 ]
$CvjWo0Zw2f_5Cq4u_EPF3II3VhozegVfdqPrTL5ugjM:beeper.local
Squid Club
true
Joining the recommendation. I tried several open source tools and they were a disaster in Hebrew, especially when the document included tables. Azure ocr gave us the best results
מצטרף להמלצה. ניסיתי כמה כלי open source והם היו קטסטרופה בעברית, בטח כשהמסמך כולל טבלאות. Azure ocr נתן לנו את התוצאות הכי טובות
[ 0, 2, 4, 10, 14 ]
$j52qDmX-6a50IJ9ad_5N9p-fN1khfPrRSBKZy11tkM4:beeper.local
Squid Club
true
If anyone is organizing a meetup and looking for a company that wants to host it, I'd be happy to connect.
אם יש מישהו שמארגן מיטאפ ומחפש חברה שרוצה לארח אותו אני אשמח לקשר
[ 7, 11 ]
$ESiBx8fywivM-l6D3tIIlNrhEzeTMmcx79gmKNIgAXY:beeper.local
MCP Israel
false
Your productivity went up 5x, while big companies report a 0-5 percent increase in productivity😎
התפוקה שלך עלתה פי 5, כשחברות הגדולות מדווחות על 0-5 אחוז עליה בפרודקטיביות😎
[ 0, 4, 6, 14 ]
$gLJYToy0A4xdmlmu503b5KCO2Q-j3-f_up0Lsp5oV60:beeper.local
Squid Club
true
Hey, does anyone know if Claude supports the csp settings of mcp-apps? Because right now we can't display resources inside claude.ai and it only works inside chatgpt?
היי מישהו יודע אם claude תומך ב csp settings של mcp-apps כי כרגע אנחנו לא מצליחים להציג resources בתוך claude.ai ועובד רק בתוך chatgpt?
[ 0, 1, 11 ]
$LwhznYCHMxG48752j-Q267LoI94einpWyor9JGgZ-BU:whatsapp.com
MCP-UI
false
Because you need to fix the agent from time to time and you need to know what and focus and all that.
כי צריך לתקן את הסוכן מידי פעם וצריך לדעת על מה ופוקס וזה
[ 0, 4 ]
$OlpfVEy80VptNfAMXg8Uu07E1Eboe9eYAhhZLAPoclw:beeper.local
Squid Club
false
Oh okay, good to know that's what it's called in professional jargon.
אה אוקיי טוב לדעת שככה זה נקרא בז'רגון המקצועי
[ 9 ]
$g96MMedxaC7kZr361jqqCxTpKrGVhF2O9XGIlZjhM5g:beeper.local
Squid Club
false
You can clarify the question: which clients does he work with, in which countries, what types of projects, what are the rates or business models he works with – it would help understand and serve as inspiration for everyone I think 😉
אפשר לחדד את השאלה מול איזה לקוחות הוא עובד באיזה מדינות איזה סוגי פרויקטים מה התעריפים או סוגי המודל העסקי שהוא עובד מולם זה יעזור להבין וישמש השראה לכולם נראה לי😉
[ 1, 4, 10 ]
$9XAHGhwf7BSRQqbg7WLpE0SCkw5NLbP_K7uMiGYszA8:beeper.local
Squid Club
true
Basically yes But I get this error a lot: Context overflow: prompt too large for the model. Try /reset (or /new) to start a fresh session, or use a larger-context model.
בעיקרון כן אבל יש לי המון פעמים את השגיאה הזו: Context overflow: prompt too large for the model. Try /reset (or /new) to start a fresh session, or use a larger-context model.
[ 0, 2 ]
$DVGIgnsfRDT-H-KYamQ2-Yb4Xf0pLfreG6iG3NThXso:beeper.local
LangTalks Community 3
true
You can host the models yourself in Claude code through Ollama, so long as you have the architecture for it
You can host the models yourself in Claude code through Ollama, so long as you have the architecture for it
[ 0, 5 ]
$9Ombb0rsrUFxGEM4DcTXmD2fuFz7W49e7VH0yan6rXY:whatsapp.com
LangTalks - AI driven coding
false
The mcp apps are agnostic to what and how you render something in an iframe. So you can use it, I don't specifically know angular elements
הmcp apps אגנוסטי למה ואיך אתה מרנדר משהו בiframe. ככה שאתה יכול להשתמש בזה, אני לא מכיר ספציפית angular elements
[ 0, 5, 10 ]
$qer90cxCXpv-tahdu7sFvat5UjioVRidS2sOsdB6GOc:whatsapp.com
MCP-UI
true
Genius
גאוני
[ 9, 12, 14 ]
$rHBP4kCq9mIJmC_yoELgLF_E5vAwPwBxNjG9ZzOiseg:beeper.local
Squid Club
true
מתי יהיו תשובות על הרשמה?
מתי יהיו תשובות על הרשמה?
[ 1 ]
$suyaXF6kj27_0qHjqtZgIw7uZWNGorTb0yscrS_0b7U:beeper.local
LangTalks Community 2
true
It's still my original 🤣
זה עדיין שלי מקורי 🤣
[ 6, 9, 11 ]
$FrgpLpbI-FncpDJxSjz1ITswkxSNfOO7tsIWmAgWkaU:whatsapp.com
AI Transformation Guild
false
Tell me, is it just me or does mcp TS work better than Python in the overall experience? That’s how I feel, no metrics.
תגידו, זה רק אני או שmcp TS עובד יותר טוב מפיתון בחוייה הכללית? ככה אני מרגיש אין מטריקות.
[ 0, 1, 2, 4, 11 ]
$4lnt0jLMU1qKchOMr96WY_29Nbnfc5J593rXgAO8fso:beeper.local
MCP Israel 2
false
Yeah, that's exactly the plan, talking about all the stuff around frontend and conversation management and all that.
כן בדיוק התכנון, מדבר על כל המסביב של פרונט וניהול שיחה וכל זה
[ 9 ]
$YDEUJxhhyhYNgUIvWmbUuqWfPDQRGA1pvNVTEvWxgec:beeper.local
Squid Club
true
True
נכון
[]
$6OGCs-9-aDdBeyfRuu3VLkLee_NdNV9KPMDS2cn5aV8:beeper.local
Squid Club
false
You need a connected computer and also internet. That's why the Claude Code option and the Claude app are appealing
צריך מחשב מחובר וגם אינטרנט. בגלל זה האופציה של קלוד קוד והאפליקציה של קלוד קורצת
[ 0, 4, 5, 11 ]
$FiB8mybM27C3di1G96Dxe8T3HJaGvk5mVfB563nFtfE:beeper.local
Squid Club
false
I think that in GPT and similar models, they cut off the beginning as the conversation progresses.
אני חושב שנגיד בgpt ודומיהם, הם חותכים את ההתחלה ככל שהשיחה מתקדמת
[ 0, 4 ]
$rcXwrhd0k3S5H_bvn2U2QIOf-IOL5aRVz2VP5-2P_Ps:beeper.local
Squid Club
false
It's not about memory, Claude, Code, and Cursor are actually general purpose agents and do whatever they want in the thinking process. Cognetivy comes to give the agent a defined framework for how to think (the workflow you create) and a way to monitor the mode of operation. Meaning, to make the agent a bit more determ...
לא מדובר בזכרון, קלוד קוד וקרסר הם בעצם General purpose agents ועושים מה שבא להם בתהליך החשיבה. קוגנטיבי בא לתת לאייג׳נט פריימוורק מוגדר לאיך לחשוב (הוורקפלו שיוצרים) ודרך לנתר את צורת הפעולה. כלומר להפוך את האייג׳נט לקצת יותר דטרמניסטי ולתת לו ״מודל מנטלי״ למי שמכיר את המונח לחשוב על פיו. הכלי מתאים לאו דווקא למשימות ...
[ 0, 2, 4, 10 ]
$BatpUh3DY9ymaGBj6w2GmaEmwYDv3TtZZZRUAXf-wyM:beeper.local
LangTalks Community 2
true
For me it works normally.. let's move to private.. they have some issue with how the Authorization header is sent, a bit different from everyone else
אצלי עובד רגיל.. בוא נעבור לפרטי.. יש להם איזה עניין באיך שולחים את הAuthorization header קצת שונה מכולם
[ 0, 2, 5 ]
1181
MCP Israel
true
My agent posted there What's the question? I'll pass it on to him 😉
הסוכן שלי פרסם שם מה השאלה ?אעביר אליו 😉
[ 1, 6 ]
$h9dX9xe-x2Ic8UDjKNdglsPqS6yGTaWeDGQd9MBiK3I:beeper.local
LangTalks AI-SDLC
false
$25,000 in revenue, if I may ask?
25 אלף דולר הכנסות, אם אפשר לשאול?
[ 1 ]
$dlb5voTPncHs5nBKhi8IZg4dq9mMeP6jWRDtxOPf4tU:beeper.local
Squid Club
true
What do you mean? You need it to run on some machine, either a machine in n8n's cloud or your own machine.
מה הכוונה? אתה צריך שזה ירוץ על מכונה כלשהי, או שהיא מכונה בענן של n8n או שהיא מכונה שלך.
[ 0, 1, 5 ]
$EmrusYAQ9S8ZDSsM_19B-7FNAuzuIYr5PQA10nfNBXc:whatsapp.com
n8n Israel - Main 3
true
Hi Question, if I want to translate articles from English to Hebrew but translate in full length (files of 30 pages let's say) which model is best to use so it does a good job? Through API... Would appreciate recommendations
היי שאלה, אם אני רוצה לתרגם מאמרים מאנגלית לעברית אבל לתרגם באורך מלא (קבצים של 30 עמודים נניח) באיזה מודל הכי כדאי להשתמש כדי שיעשה עבודה טובה? דרך api… אשמח להמלצות
[ 0, 1, 10, 11 ]
$KJrEY4YW55Gq-nLQq_fLHvi-Ek0191FfUgEUm5yUzRY:whatsapp.com
MCP Israel
false
Very interesting! Has anyone had experience with platforms/companies similar in concept to Mem0 and can recommend?
מאוד מעניין ! האם למישהו יצא להתנסות בפלאטפורמות / חברות שהם דומות ברעיון ל Mem0 ויכול להמליץ ?
[ 0, 1 ]
$YYs9vajZDDlQxuClPFz-8IW1Q7eHQAUXVN9FcOKmHU4:beeper.local
LangTalks Community
true
At what stage do you run them? If there are particularly problematic stages, try running a unit test on them first (with a real llm, not a mock)
באיזה שלב אתם מריצים אותם? אם יש שלבים בעייתיים במיוחד, תנסו להריץ unit test עליהם קודם (עם llm אמיתי, לא mock)
[ 0, 1, 5, 10, 11 ]
$OVGNDjSRq1FtZfZFjfal1uQqJuqCueeGl4fQg_Gr6Uk:beeper.local
LangTalks Community 3
false
I agree with what was written, you often end up working for it.. there's a chance they'll eventually get to a version where it's really worth it
מסכים עם מה שכתבו, אתה עובד הרבה פעמים בשבילו.. ישמצב שעוד יגיעו לגרסה שזה יהיה ממש שווה את זה
[ 2, 4 ]
$PoeyccKRqxS5uRz1zyCdUcwgMbxC0P6BGNduReOcTzo:beeper.local
Squid Club
true
Has anyone here encountered the following situation: A. Small team, working on Claude Code individual pro because it cost less than teams and they didn't see a need for teams. They paid for an annual subscription because of the discount. Long story short, now there are 7 different developers in the same company, each ...
האם מישהו פה נתקל בסיטואציה הבאה: א. צוות קטן, עבדו על individual pro של קלוד קוד כי זה עלה פחות מ teams ולא ראו צורך ב teams. שילמו מנוי שנתי כי הנחה. מפה לשם נוצר מצב של 7 מפתחים שונים באותה חברה שכל אחד מהם עם individual pro בתשלום שנתי, שנרשם במועדים שונים, ולכולם יש בין 6-11 חודשים שנשארו להם במנוי השנתי. ב. הופ...
[ 0, 1, 2, 10, 11 ]
$8dEgCTLLvdnTQPZAseSGOVt9EsTAEUhBTRWWtHQYnBs:whatsapp.com
AIL - AI Leaders Community
false
They mentioned it in skill.md once for the agent to generalize
They mentioned it in skill.md once for the agent to generalize
[ 0, 4, 11 ]
$eCOKfnAtFvVko5yYqbTtMx_kNUpoVRj_pUW4Ks-8NVw:beeper.local
LangTalks - AI driven coding
false
Hey, We encountered a case where the answer to a question included many relevant sources, which exceeded the standard top k, and as a result, we missed important information. How would you suggest tackling such a problem so as not to miss information? Or in other words, to make sure the model has gone through "enough" ...
אהלן נתקלנו במקרה שבו תשובה לשאלה כוללת הרבה מקורות רלוונטיים, שחרגו מ-top k סטנדרטי ולמעשה פספסנו ככה מידע חשוב איך הייתם מציעים לתקוף בעיה שכזו כדי שלא לפספס מידע? או במילים אחרות, לוודא שהמודל עבר על "מספיק" מקורות על מנת לספק תשובה מלאה 🙏
[ 0, 1, 2, 10, 11 ]
$GlZKkvWd0hwRyg2eHXFIp8fIAR2IOlMpZnKFJWHBcaE:beeper.local
LangTalks Community
false
You can do stdio and access it with an api-key and then it won't drive you crazy.. But it depends on what exactly you need because it's not perfect either
אתה יכול לעשות לו stdio ולגשת אליו עם api-key ואז זה לא ישגע אותך.. אבל זה תלוי מה אתה צריך בדיוק כי זה גם לא מושלם
[ 0, 4, 5 ]
$aM53odJwfQZSrSb0Z46885ZAWlGxb_Kn8WkkIlqTsK4:beeper.local
MCP Israel 2
false
Old photo. Media will be requested from your phone automatically overnight. If you also don't like that you can't automatically name chats in chat hub - give me an up vote
Old photo. Media will be requested from your phone automatically overnight. אם גם אתם לא אוהבים את זה שאי אפשר לתת שם אוטומטי לצ׳אטים ב-chat hub - תנו לי up vote
[ 0, 4, 11 ]
$-bSFMtVFr7MkZmcngkghI8QNPkVUuEKxftmueCQug9A:whatsapp.com
n8n israel - Main 2
false
I meant the deepagents library from LangChain.
התכוונתי לספריית deepagents של לאנגציין
[ 0 ]
$RWle3dpRQYmuSqd_ij8l_zgV33Lr7f7Bbh4EgbuRs20:beeper.local
LangTalks Community 2
true
Good week to the inefficient helpers of openClaw 🤖 In recent weeks we've started dealing with MCP Apps and of course, also with MCP UI. After quite a few experiments, I decided to package something that really works well for me. Until now we dealt with how to put the agent into the existing dashboard, and how to let...
שבוע טוב לעוזרים הלא יעילים של openClaw 🤖 בשבועות האחרונים התחלנו להתעסק אצלנו עם MCP Apps ואיך לא, גם עם MCP UI. אחרי לא מעט ניסויים, החלטתי לארוז משהו שבאמת עובד לי ממש טוב. עד עכשיו התעסקנו באיך מכניסים את האייג׳נט לדשבורד הקיים, ואיך נותנים לו לראות את מה שהיוזר רואה על המסך. עכשיו הגיע הזמן לשלב הבא – איך מכניס...
[ 0, 2, 3, 6, 7, 10, 11, 12 ]
$TJ1maaundflsyN22w3349JeK7aM2bt8EjFS7r4fM7nE:whatsapp.com
AIL - AI Leaders Community
false
Works smoothly for me on M1
עובד לי חלק על m1
[ 2, 9 ]
$tCamic82fOIFzlnErJoE6T2NeYW_QfUON21W_DOrEus:beeper.local
AIL - AI Leaders Community
false
There's an option today to control this, meaning on-demand, you don't have to load everything into memory
יש היום אפשרות לשלוט על זה, כלומר ondemand, אתה לא חייב לטעון את הכל לזכרון
[ 0, 5 ]
$2lh0hhXCtRneXZWqwgYvzKdPWMjIVdzPy0w1Tau-Gqc:whatsapp.com
MCP Israel
true
Resources from on-prem to cloud?
Resources from onprem to cloud?
[ 0, 1 ]
$NGRMlS_UrE_wcatmsvnNsOksV9kvtuRIbyfUyqcUeyQ:whatsapp.com
AI Transformation Guild
true
Does anyone know why, even though it seems installed, I can't use Claude Code on web?
מישהו יודע להגיד לי מדוע, למרות שנראה שזה מותקן, אני לא מצליח להשתמש בקלוד קוד בווב?
[ 0, 1, 2, 11 ]
$7qWIRSESGUppr0WMxnU8y7BOZ2koq7EStAEqqrS4AwU:beeper.local
Squid Club
false
Amazing
מדהים
[ 9, 14 ]
$QZ_BJkOgxorI-gevSsOIhyQ6svaSfWZIJo3d2Q606cM:beeper.local
Squid Club
false
Great! Is this available to read on any web page? The format here is a bit tough
נהדר! יש את זה לקריאה באיזה דף אינטרנט? הפורמט פה קצת קשוח
[ 1, 14 ]
$QYKu3XfJ3uaCNGZKmbnjr40H0iRuFgfggjtdJHCyCLA:beeper.local
Squid Club
true
https://github.com/marketplace/actions/code-review-with-chatgpt
https://github.com/marketplace/actions/code-review-with-chatgpt
[ 0, 3 ]
$BCDs_zkTdJ9n1cmmyW9EzgbP76W75YGr5g8pXPOPftI:beeper.local
Squid Club
true
Build the planning file with task breakdown, then without additional planning.
תבנה את קובץ התכנון עם פירוק למשימות, ואז בלי תכנון נוסף
[ 5 ]
$E7pJ94f5cgYsqfyLr9clDwQoSAn5FEMmQopIyISXxb0:beeper.local
Squid Club
true
You're a star! 👑
יא כוכב! 👑
[ 9, 12, 14 ]
1027
n8n israel - Main 2
true
Highly recommend it. It's very convenient. It doesn't really do orchestration, but it breaks things into tasks - and sub-tasks, and then the agent can work on a defined task end to end
ממש ממליץ עליו. הוא נוח מאוד. זה לא ממש עושה אורכוטרקציה, אבל מחלק לך למשימות - ותתי משימות, ואז הסוכן יכל לעבוד על משימה מוגדרת מקצה לקצה
[ 0, 2, 4, 5, 10, 14 ]
$1uWTa-G_CffPo3gsj7zBZVas7FCHjaYSraZ6i7JO294:beeper.local
Squid Club
true
* Behind writing a (successful) book there's a lot of skill and theory I learned this when I founded storywise.ai about a month after chatGPT came out, and among other things it helps publishers fight the flood of books being written by people like you 😅 But it also helps authors do literary editing for their plot -...
* מאחורי כתיבת ספר (מוצלח) יש המון מיומנות ותיאוריה למדתי את זה כשהקמתי את storywise.ai בערך חודש אחרי שיצא chatGPT ובין היתר עוזר להוצאות לאור להלחם במבול של ספרים שנכתבים על ידי אנשים כמוך 😅 אבל גם עוזר לסופרים לעשות עריכה ספרותית לעלילה שלהם- מבתי לשלם אלפי דולרים
[ 0, 2, 4, 6, 10 ]
$Y0dMA4qvNe5bo4dyk6VT2KM6iAb0I22G5js1ftJI0VI:beeper.local
Squid Club
false
A different sound? Because I think it plays a sound by default.
צליל אחר? כי לדעתי הוא משמיע צליל כברירת מחדל
[ 0, 1 ]
$pJLZxP0jBoP5QwS0OFJgUHBwLw3ouj0SqU3dJJcCgd0:beeper.local
LangTalks - Code Generation Agents
false
* Hi all, We want to enable clinicians (yup, non-developers) to run LLMs on historical clinical data to test prompts and measure performance. The problem: how to let them do this easily, with versioning and result tracking, without building everything from scratch. Our current solution idea: Prepared dataset – CSV w...
* Hi all, We want to enable clinicians (yup, non-developers) to run LLMs on historical clinical data to test prompts and measure performance. The problem: how to let them do this easily, with versioning and result tracking, without building everything from scratch. Our current solution idea: Prepared dataset – CSV w...
[ 0, 1, 2, 10, 11 ]
$5sjmlR98GzAVoGq8Vt0fXaBHly24UWUh6GZc2nJy5g4:beeper.local
LangTalks Community 4
false
One more week! Join, it's free and virtual
עוד שבוע ! תצטרפו זה בחינם ווירטואלי
[ 7 ]
$1otw5zI14RBMBAwIhQ_QfDLgblcwTOM5o7kjKP_tCAA:whatsapp.com
A2A Israel
false
I've been toying lately with the idea of using LiteLLM as a proxy for Claude Code, with retrospective analysis of what happened. For example, at the end of the day, go over the sessions that took place and have a cheap Claude read and give feedback on what the user did. I have a feeling this has already been done 🙂
אני משתעשע לאחרונה ברעיון של שימוש ב LiteLLM כפרוקסי לקלוד קוד, עם ניתוח רטרוספקטיבי של מה קרה. למשל בסוף יום לעבור על הסשנים שהיו ולתת לקלוד זול לקרוא ולתת פידבק על מה המשתמש עשה. יש לי תחושה שכבר עשו כזה 🙂
[ 0, 2, 4, 10, 11 ]
$l8Li4t63TKcw8Gp475Oe2NHVB4_l8Ko1M3efbgNjB34:whatsapp.com
AIL - AI Leaders Community
false
My clients work with Amplemarket but it's in the expensive category.
לקוחות שלי עובדים עם Amplemarket אבל הוא בקטגוריה היקרה
[ 0, 2, 3, 4 ]
1174
n8n Israel - Main 3
true
With the $100
עם ה100 דולר
[ 9 ]
$99RDzlWWSn7RFnxci1XFZxHp_HxmX7vVo2_py8JhfIc:beeper.local
Squid Club
false
Thanks a lot, brother, I'm going to explore this tool tonight, sounds priceless.
תודה רבה אחינו, אני הולך לחקור את הכלי הזה הלילה, נשמע שזה פרייסלס
[ 9, 12, 14 ]
$VICiJaYf24IwFD65GkM-Wt1BXreBq9nZRmNI_fgas6s:beeper.local
Squid Club
true
Definitely a tough FOMO chase Slowly I developed indifference and skepticism about every "innovation" A new skill to develop is the ability to quiet the fear of being left behind and understand you're on the right track
אכן מרדף פומו קשוח לאט לאט פיתחתי אדישות וספקנות על כל ״חידוש״ סקילז חדש שצריך לפתח היכולת להשקיט את החשש להישאר מאחור ולהבין שאתה בכיוון המכון
[ 2, 4, 10 ]
$96WYKMxnGhJAscvYWo3aey7xS4jtbZA7ez4Wso444aI:beeper.local
Squid Club
true
In theory, I'm trying to understand how the candidate develops in this era. Proper task breakdown, planning and creating suitable context, task planning, using models and tools, validations, agent code review processes, etc. Questions that show understanding and control of context, use of skills..
בתאוריה, אני מנסה להבין איך המועמד מפתח בעידן הזה. פירוק נכון של משימה, תכנון ויצרת קונקסט מתאים, תכנון משימה, שימוש במודלים וכלים, ולדיציות, תהליכי קוד רויו אגנטים וכו. שאלות שמראות הבנה בשליטה בקונקסט, שימוש בסקילים..
[ 0, 2, 5, 10 ]
$nNXffbKmyFuO649D1NLbF5VYAnBBqAoQ4pHbJUYUj7Q:whatsapp.com
AIL - AI Leaders Community
true
* I'm just setting up such a team. I think the PR review is a very late stage. You could have great code that's drifted from what was originally intended. Think about it—in the human world, every time there wasn't an organized process of breaking down the task and checks by the stakeholders to ensure it's what was a...
* אני בדיוק מקים צוות כזה. אני חושב שה pr review זה שלב מאוד מאוחר. יכל להיות לך קוד מעולה שהוא drifted ממה שהתכוון המשורר. תחשוב שבעולם האנושי כל פעם שלא עשו תהליך מסודר של פירוק משימה ובדיקות של המעורבים שאכן זה מה שסוכם קיבלנו משהו שה product לא בדיוק ביקש. אני חושב שה qa שלך צריך לבדוק קוד אבל גם לבדוק את כל מה...
[ 0, 2, 4, 10, 11 ]
$OYotHx8T5zNtNI4sdh2YuTN6CIFHB0YMMS9U6UcM9l0:beeper.local
AIL - AI Leaders Community
false
Probably?
כנראה?
[ 1, 9 ]
$z-il5CN8SY3G43LecUq8_-jSdNcSJh3rSkl2w3p_4X0:beeper.local
Squid Club
true
Keep in mind that even if you take the high numbers we're used to hearing, say $1,000 per person, which is significant monthly AI expenses, there's still a multiplier of 1-7 in terms of headcount costs, and that's before employer expenses, etc. And most people don't use such costs, mainly the early adopters/high achie...
קחו בחשבון שאפילו שלוקחים את המספרים הגבוהים שהתרגלנו לשמוע, נגיד 1000 דולר לאדם שזה משמעותי בחודש להוצאות על AI עדין יש כאן מכפיל של 1-7 נגיד במובן של הוצאות headcount וזה לפני הוצאות מעסיק וכו ורוב האנשים לא משתמשים בעלויות כאלה, בעיקר הearly adopters/high achievers לכן אם מסתכלים פה על פריזמה של תחליף headcount י...
[ 0, 1, 2, 4, 10 ]
$cv-jzkY9fSNK-tczeWBXpBrAjvYsprjxApa9LJlSRX8:whatsapp.com
AI Transformation Guild
false
What's the definition of strong enough?
מה ההגדרה של חזקה דייה?
[ 1 ]
$7qB-a-z5g2jLUwW-v8vJ7g-XdeZ7z1uOaJiNJQ1_eE4:whatsapp.com
LangTalks Community 3
true
Hey friends, I recently worked on a Flow in n8n that connects to Slack and GitHub's MCP and answers metadata questions from the DWH. I'm attaching a picture of the flow here, and if you're interested in seeing it in action, I also uploaded a short video on LinkedIn: https://www.linkedin.com/posts/uria-fridman-a9b0bb20...
היי חברים, עבדתי לאחרונה על Flow ב-n8n שמתחבר ל-Slack ו-MCP של GitHub ועונה על שאלות metadata מה-DWH. מצרף כאן תמונה של ה-flow, ואם מעניין אתכם לראות את זה בפעולה, העליתי גם וידאו קצר בלינקדאין: https://www.linkedin.com/posts/uria-fridman-a9b0bb203_n8n-genai-dataengineering-activity-7401515563971387392-zFPJ?utm_source...
[ 0, 2, 3, 10, 11 ]
1126
n8n Israel - Main 3
false
Agreed
מחזק
[ 9, 14 ]
$phCLim1L7sfeLnjKHKyuicR2WOoJ7W67oXXqoHbcdW0:beeper.local
Squid Club
true
Can connect you.
יכול לקשר
[ 9, 14 ]
$jqlQa4UtnAbUtZ98a9pCgAcUDY4nyiNbS7XQo5cTjA0:beeper.local
AI Transformation Guild
false