LangTalks

LangTalks arrived to Big Data London (BDLN) !!! 🇬🇧🦜 🦜 We're here to geek out with Guy Fighel over how the Data Engineer role is evolving in the era of AI-first products! Come say hi at the Data Fabric Theatre! Guy Fighel Lee Twito Hetz Ventures LangChain

Last week I attended the LangTalks meetup on AI-driven development! A clear theme emerged: the shift from Vibe Coding towards Spec-Driven Development. From Lee Twito and Gal Peretz R&D managers’ playbook for AI coding transformation to Gili Nachum's demo of Amazon's agentic IDE, Kiro, the emphasis is on perfecting the initial spec to guide AI agents. Sahar Carmel (סהר כרמל)’s talk on brain temperature control for creativity in research and focus in designing the solution. Nimrod Kor’s session on dynamic context management further highlighted the need for well-structured inputs. Removing the clutter and keeping only what is needed. The key takeaway: as AI automates coding, the engineer’s role is evolving to orchestrating the entire SDLC. Huge thanks to Lee Twito and Gal Peretz for organizing another fantastic LangTalks meetup. For more great insights, I highly recommend joining the LangTalks community and listening to their podcasts. #SDLC #AgenticAI #Langtalks

We're coming to London for Big Data LDN! Join Guy Fighel - from keynote to meetup - for a bunch of chances to connect: 👉 Hetz VIP Drinks for senior executives, investors and founders 👉 A *live* LangTalks recording with Gal Peretz & Lee Twito 👉 Data Engineer Things Meetup, hosted by Hetz, Orchestra & Lightdash 👉 Let's Talk Context with Guy & Omri Lifshitz @ Upriver 👉 Governed Data Models panel including Guy + Hetz Data advisors - Shachar Meir & Joe Reis If you're in London next week, grab a ticket to BDL (it's free and registration is open). Links to talks and details below in comments. See you there!

Is now a bad time for Key takeaways from yesterday's LangTalks meetup? 💡 Ready or not, here they come 🥳 🔦 Software Development Life Cycle (SDLC) is back in the spotlight Spec → Design → Development → Testing → Deployment → Maintenance Lee Twito and Gal Peretz demonstrated how AI helps accelerate development. Now we are stuck addressing the rest lol Successful companies create AI-ready development processes and foster an AI mindset. The penalty for doing the wrong thing is amplified with AI as the rest of the funnel suffers. 👑 Software Engineers are shifting towards Team Leaders AI copilot → AI pair programming → AI autonomy The evolution of AI in Development continues. From tab completions to generating snippets to agents and agentic flows. Software engineers are expected to orchestrate features and projects. If you haven't done so already, it's time to align engineering KPIs with DORA metrics: lead time to production, deploy frequency, and failure/recovery time. 😰 Addressing SDLC issues with Kiro Gili Nachum demoed Amazon's new agentic IDE, Kiro, which comes with pre-baked concepts of how to address the first 3 SDLC steps: Spec, Design, and Development. ⏳ Context switching problems are back Just as with humans, LLMs are struggling when it comes to messy contexts. Nimrod Kor showed us how his startup Baz, can help clean up and optimize the context of LLMs to yield better results when executing agentic flows. They use simple concepts: remove duplication, summarize, create small tasks with clear deliverables, and batch. Does it sound familiar? Well, because it is, these are the same concepts highly productive developers use. Read more in my article attached in the first comment. 🎨 Creativity Sahar Carmel (סהר כרמל), who should probably do stand-up as well, connected the dots between creativity and LLMs. In short: optimize for deliverables. Use high temperature to allow for creative freedom in the research phases, and push for compliance (low temperature) in the design phase. If both are done correctly, development will feel effortless. — It is amazing to see that my personal predictions made almost two years ago at Augury are materializing in bigTech. I've been increasingly vocal about the need to shift focus to steps before Development (in SDLC). Development speed was never the issue; gathering the right context and choosing the right thing to do is. Maybe now that developers receive the role of team leaders and can directly influence product direction, we'll see my vision start coming true: developers working directly with customers. The future is definitely exciting. — Are you looking to amplify your R&D and lead complex projects? Let's connect 📩 #AI #SoftwareEngineering #LangTalks

Full house!! Thanks to 200 LangTalkers that joined us today for a meetup full of AI-coding, connections, and pizza The great speakers: Gili Nachum , Nimrod Kor , Sahar Carmel (סהר כרמל) And our sponsors LangChain , Baz , Amazon Web Services (AWS) , Sap.Hi The sessions recordings will be available on LangTalks YouTube channel next week: https://lnkd.in/dNfWjU7q Yours Lee Twito & Gal Peretz LangTalks

הקסם של LangTalks הוא ההפרדה בין הבאזז והכותרות לבין התכלס והפרקטיקה. זאת הסיבה שהגיעו למיטאפ היום כמה מאות ai engineers, וגם זה רק כי לא היה מקום ליותר (נרשמו 1100). התמונה הרנדומלית הזאת מהסדנה שהעבירו האבא והאבא של הקהילה, Lee Twito Gal Peretz , היא דוגמה לזה. טעימה. כמו הפיצה. כל מיטאפ פיצה ובירה. יש קבעונות שלא צריך לשנות. אם זה עובד אל תיגע. אם זה פיצה תיגע. אני ב derailment איפה Sahar Carmel (סהר כרמל) שיעצור אותי. ההקלטות של הסדנאות יעלו בשבוע הבא לקישור שבתגובהראשונה.