Himanshu Sharma

If you are still manually typing SQL in 2026, you are not only behind but obsolete. optimizesmart.com/ga4-bigquery-c…

Influencers today, in a nutshell --> Post a selfie + a generic motivational (but largely useless) quote and call it a day. Watch the "This spoke to me " comments roll in". But post something actually useful (like when to split a voice agent into two or more specialised agents) and hear crickets. The algorithm (and audience) often rewards shallow motivation over deep, actionable insight. So here’s my challenge: Drop one piece of genuinely useful, non-generic advice you’ve applied in your work, especially in AI.

I don't claim to be self-made. I don't give constant reality checks to others. I don't dismiss my luck and give 100% credit to my hard work. I don't claim to have figured it all out. I don't completely dismiss the challenges/problems the younger generation is facing or label them lazy without any solid evidence, or give them oversimplified solutions (work hard, stop complaining or be like me if you want to succeed). I live a hard life, but that doesn't mean you should, too. Be like me lol. Life has always been a mix of timing, luck, personality, starting conditions, economic cycles, technology waves, and yes, personal effort. Pretending the personal efforts are the only ones that matter is a form of cosplay. Most "self-made" stories conveniently edit out the invisible help, good breaks, or different era they lived in. "They are too lazy....When I was their age, I did this ...They want everything handed over to them..." - previous generation. "They don't understand me. They don't care = they don't love." - current generation. Each new generation resents the previous one exactly because they fail to understand their unique challenges and circumstances, and give them a blueprint which no longer works. >> Share your story as "this worked in my circumstances" rather than "this is the blueprint." >> Listen first. The younger cohort's "they don't understand me = they don't love me" often comes from feeling dismissed, not from entitlement. The people who earn respect across age groups are usually the ones who can hold both: "I worked hard and caught some breaks. Your path has different obstacles. Here's what I see, but adapt it to your reality." You don't want to be part of the generation resented by the next. Break the generational curse.

Check out the latest article in my newsletter: When to split your voice agent - The Bleed Test Framework. linkedin.com/pulse/when-spl… via @LinkedIn

Why Your Voice Prompt Needs a Clean-up Pass? A clean up pass is a deliberate edit session that reorganises the voice prompt without changing what the agent does. Same behaviour, fewer rules, each instruction in one obvious place.

The most dangerous people in your circle are the ones loudly preaching: “Nobody is coming to save you.” They didn’t grind alone because they’re tough. They waited for help that never came (often without ever clearly asking), got bitter, and now turned their pain into a universal law. When someone else is struggling, these same individuals will likely withhold help not just out of indifference, but as a subtle form of retaliation or “fairness” enforcement: • “I had to grind it out with no one, why should you get a handout?” • “If I suffered in silence, you should too.” • It becomes a way to validate their own hardship: If I help you, it cheapens what I went through. Once you understand their psychology, you realize they are not virtue signalling and you are dealing with a monster who will backstab you at first opportunity . They are potentially dangerous in relationships, not just unhelpful, but actively undermining. If helping you “cheapens” their struggle, your success or ease can feel like a personal threat. They might smile, offer vague motivational slogans, then quietly sabotage (spreading doubts, withholding key info, or distancing when you need them most) to restore their narrative of “life is hard for everyone.” The internal script (“I suffered alone, so you should too”) can flip into active punishment when you bypass their rules, e.g., by asking directly for help, succeeding with support, or challenging their worldview. It’s not always conscious malice; it’s a wounded ego protecting itself. Loud preaching of self-reliance often masks vulnerability. People like this can seem strong and “realistic” until a situation tests loyalty. Then the mask slips: they prioritize their unresolved grudge over your well-being.

When I started blogging about Google Analytics (some 14 years ago), I heard crickets, day in and day out. It was just me and 2 or 3 others. Then, over time, interest in GA exploded, resulting in millions of website visits a year. Most people today have no idea how early I started or how long I’ve been covering Google Analytics. The bloggers who started with me quit over a decade ago. Today, it feels exactly the same with Voice AI. I’ve been consistently creating content about voice AI, while hearing mostly crickets again. But I’ve seen this movie before. In a couple of years, voice AI will be everywhere. The demand for practical knowledge, implementation guides, and real-world strategies will skyrocket. What feels quiet and overlooked right now will become one of the most competitive and valuable spaces in tech. This is my reminder to anyone building in emerging fields: delayed gratification still wins. The consistent creators who show up early and stay through the silence are the ones who win when the wave eventually arrives. The rest do catchups.

Check out the latest article in my newsletter: When your voice prompt becomes the thing your tests wrote. linkedin.com/pulse/when-you… via @LinkedIn

"Think with AI and NOT think through AI." Most developers accept whatever solution the AI suggest. There is almost always a better solution if you can challenge the AI output.

GPT Realtime 2 is now available in Retell AI. However, watch out for its limitations and the fact that it is the most expensive voice LLM.

Many times, I feel Keir Starmer is trolling. How can someone be so hopelessly clueless? One has to ask where he gets his news from.

Here is what diet does to your age. Both guys on the left are younger than me but look like my dad. One is on a prince's diet, the other is on an actor's diet, and I am on a strict diet (no alcohol, no meat, no sugar, no dairy, no bread).

Check out the latest article in my newsletter: These three skills set you apart as an AI agent developer. linkedin.com/pulse/three-sk… via @LinkedIn

Don’t bet your company’s future on AI demos. Many companies (like Salesforce) learned this the hard way. Just because a system crushes routine tasks in a controlled pilot doesn’t mean it’s ready to replace humans at scale. Salesforce cut customer support headcount from ~9,000 to 5,000 using Agentforce. Executives later admitted they had overestimated the reliability of LLMs in real-world conditions. What looked promising in demos created headaches around accuracy, complex queries, and customer satisfaction. And they’re far from alone. Forrester’s 2026 analysis shows 55% of employers who made AI-driven layoffs now regret it. The fallout includes: >> Degraded service quality. >> Loss of tribal and institutional knowledge. >> Hidden costs (humans constantly fixing AI mistakes). >> Hits to employee morale and customer experience. Many companies are now quietly rehiring (often 25-50%+ of the roles they cut), sometimes at higher salaries. Just because AI can automate a task doesn’t mean it can match human accuracy and judgment in the foreseeable future, not without "months" of rigorous real-life testing, optimisation, feedback loops, and human oversight. Most organisations never reach clear, consistent performance above 80% in both efficiency AND accuracy before making big replacement bets. AI is an incredible tool. But it’s still a junior teammate that needs training, supervision, and iteration, not a plug-and-play replacement for experienced professionals. Leaders pause before the next big AI-driven restructure. Test thoroughly in production-like conditions. Measure what actually matters (customer outcomes, not just cost per ticket).

‘No one is coming to save you’ is often parroted by those who have been saved countless times by others. Don’t expect any help/empathy/sympathy from such people cos they have forgotten the face of their father and mother, mentors, communities, institutions who did save, support, or scaffold them. The timely loan, the stable home, the country they live in, the inherited cultural capital, the safety net of a functioning society, the quiet interventions they no longer acknowledge. To a narcissist“I did it all myself” feels way better than “I got crucial assists along the way.”

Lifetime access doesn’t mean access for your entire life. It means access for as long as the product exist. So it refers to the lifetime of the product. You get access for as long as the company continues to offer and support that specific version/product. If they shut down the service, discontinue the product, or stop maintaining the servers, your “lifetime” access ends, even if you’re still alive and well.

I am transferring the 'sales and roi acceleator' course to the main website as it's the flagship course, and then retiring the old e-learning platform at training .opimizesmart. com before June 1, 2026. Dont enroll into any course there. If you have an existing unfinished course there, now is the time to complete it. The main website already has the GA4 BigQuery Course. It's time to move on.

Here is how you know this person doesn't know much about voice AI. They say something along the lines, "I built a voice agent that sounds like Human. And here is how you can do the same". Sounding human is table stakes in the voice AI field. And most of the time, it's the TTS doing the job. So you can't even claim it's you who did that. Slapping a good TTS on top of a basic LLM pipeline and calling it “your” voice agent is like saying you built a car because you chose a nice paint job. The real flex is building a voice agent that handles interruptions, background noise, or mid-sentence corrections without derailing and actually completes the task (booking, troubleshooting, ordering, whatever) instead of sounding empathetic while failing. The graveyard of voice startups is full of beautiful-sounding agents that couldn’t book a haircut without hallucinating the time or looping endlessly when the user said, “never mind.”

ChatGPT users are high intent (ready to buy or deeply engaged in a specific action). Actually, there is no data to prove that. ChatGPT isn't like a pure search engine, where every query screams buying signals; much of it is exploratory, learning, or just for fun.

Usage limits are up, effective today we're: 1) Doubling Claude Code's 5-hour limits for Pro, Max, Team and seat-based Enterprise plans 2) Removing peak hours limit reduction on Claude Code for Pro and Max plans 3) Substantially raising our API rate limits for Opus models

Check out the latest article in my newsletter: Gemini 3.1 Flash-Lite Works Well for Voice AI. linkedin.com/pulse/gemini-3… via @LinkedIn