Gautam Baghel

The move toward autonomous infrastructure won't happen overnight. It starts with a strong foundation. Infrastructure as code, trusted infrastructure context and embedded governance are becoming essential as AI agents take on a greater role in operations. Read more: https://lnkd.in/gjwkdX4u

67

2 Comments

🎯 SFMC One Concept a Day #4 𝐏𝐑𝐈𝐌𝐀𝐑𝐘 𝐊𝐄𝐘 𝐢𝐧 𝐃𝐚𝐭𝐚 𝐄𝐱𝐭𝐞𝐧𝐬𝐢𝐨𝐧𝐬 This looks like a small checkbox… but it can save you from silent disasters in SFMC. ✅ What is a Primary Key? A Primary Key is the field that should be unique for every record in a Data Extension. Example: SubscriberKey → Primary Key Meaning: Rahul should not exist twice with the same key. ✅ Why it matters in real projects Without a Primary Key, you may get: duplicate records wrong personalization Unexpected Journey behavior messy reporting And the dangerous part? You may not notice it immediately. ✅ Simple example If your DE has: SubscriberKey → 100005125 Email → rahul@gmail.com And later another row comes with the same SubscriberKey… you now have identity confusion. ✅ Best practice I follow For most profile DEs: SubscriberKey / ContactKey should be the Primary Key And for transactional DEs: Use a business key like: OrderID or TransactionID Simple rule: Clean keys = clean journeys Messy keys = future fire drills ✨ 𝗜𝗳 𝘆𝗼𝘂 𝗹𝗶𝗸𝗲 𝘁𝗵𝗲𝘀𝗲 𝗱𝗮𝗶𝗹𝘆 𝗦𝗙𝗠𝗖 𝗰𝗼𝗻𝗰𝗲𝗽𝘁𝘀 𝗮𝗻𝗱 𝘄𝗮𝗻𝘁 𝗱𝗲𝗲𝗽𝗲𝗿 𝗲𝘅𝗮𝗺𝗽𝗹𝗲𝘀, you can join my SFMC Playbook by Rahul Khevji newsletter here 👇 https://lnkd.in/geJgzXSa #SFMC #SalesforceMarketingCloud #DataExtensions #JourneyBuilder #SQL

62

2 Comments

To all platform engineering teams building on AI agentic infrastructure: Treat AI agentic memory as a production database. Define one memory resource and reference it across every agent configuration. The same memory can be shared across multiple LLM models — memory is for conversation history and state, not tied to any specific model. PS -You’ll need to handle prompt compatibility and context management across different models. On MCP servers, use short-term tokens with limited capabilities in your MCP configuration. Any configuration changes made with that token are your responsibility. Use Vault for token management. Cheers, and enjoy the AI journey !!

8

My previous post about leaving Intuit sparked an incredible wave of conversations. It’s clear that the intersection of career transitions and the AI revolution is deeply top-of-mind for all of us. It made me think about a pivotal moment less than a year ago, listening to Intuit's CEO, Sasan Goodarzi, sharing his vision of an AI-driven productivity leap. Sasan noted that one day we would look back and wonder how we ever lived without AI, just like we can't imagine today how we lived without TV, internet, or smartphones. That very specific moment was a real inspiration and a call to action for me - it pushed me to challenge myself and my R&D teams to completely rethink how we build software. Fast forward to today, we are already moving way past 3x. We have entered the 10x execution era. But let's talk about the reality: a 10x developer isn’t just someone who writes nice prompts or codes faster. "Vibe coding" is a cool term for all, but as real engineers, we need to find the sharp edge where our true strengths can shine. To truly stand out today, software professionals must bring a completely new set of skills to the table: - Strategic Brainstorming: AI models are like brilliant but impulsive interns. The real skill is in the dialogue - challenging their architecture, catching hallucinations, and keeping the focus where it matters. - Context Injection: Models don't know your specific business, customer nuances, or physical hardware constraints. Value has shifted from writing code to injecting deep, high-fidelity context. - The Art of the Question: In a world where answers are instantaneous, your worth is measured by the quality and depth of your questions. Living this 10x reality during my transition has allowed me to spend my time hacking away at an amazing personal, tech-for-good project I’ve rolled in my mind for months. Combining AI with hardware allows a tiny team (a single engineer, in my case) to move at lightning speed. I can't wait to reach the prototype stage and share a video teaser here. Looking ahead - what should we be studying next? Yes, I truly believe software engineering as we know it is out. The next massive breakthroughs will happen in the fusion of software with domains like Medicine, Physics, and Biology. Historically, these fields were bottlenecked by overwhelming, hidden complexities in massive datasets. AI is unlocking these layers, turning data into predictable, actionable insights. The leaders of tomorrow will be those who bridge the gap between software and the physical or biological world. That is the next frontier. The market isn’t shrinking, it’s evolving. Those who learn to lead the tools, rather than just use them, will define what comes next. What do you think? Which "Software + X" combination will drive the next big breakthrough?

50

7 Comments

Joining the HashiCorp Agentic Runtime Security Exchange on 19 June. Agentic AI is reshaping how enterprises operate. Autonomous agents acting on behalf of users, inheriting privileges instantly, and provisioning access at machine speed. Most organizations still rely on human-centric IAM and static secrets. That gap is now on top of mind for auditors, regulators, and boards. This peer-led evening exchange brings together leaders across security, identity, cloud, and AI to compare approaches on securing the agentic runtime - without losing/compromising the audit and evidence trail. CISOs, Chief Architects, VP Infrastructure, senior technical decision-makers - this is your room. Register now: https://lnkd.in/dtNKn_JA Syed Moinuddin| Jerome Walter| Calise Goh | Bomí K. | Jen Hussain | Mudit Mathur | Simranjeet S. | Shaibal Saha | Sandeep Madhiraju | Anurag Goyal #AgenticRuntimeSecurity #ZeroTrust #NonHumanIdentity

32

2 Comments

Together AI just dropped an update that makes customizing models: Faster ⚡ More scalable 📈 Easier to integrate into real workflows Worth a read: https://lnkd.in/gMgrta_N #AI #LLM #MachineLearning #FineTuning #GenAI

60

1 Comment