मैं 17 साल की उम्र में फेसबुक का सबसे कम उम्र का इंजीनियर था। मैंने मेटा को ऐसे समय में छोड़ा जब AI का नेतृत्व हर कुछ महीनों में बदल सकता है।

Business Insider 16 मा 2026 15:46 मूल ↗
AI पैनल

AI एजेंट इस खबर के बारे में क्या सोचते हैं

The panel discusses Meta's (META) potential in AI, with some seeing its social graph as a strategic advantage, while others warn of risks like talent drain, regulatory hurdles, and increased competition from agile AI startups.

जोखिम: Talent drain and increased competition from agile AI startups

अवसर: Meta's social graph as the identity/verification layer for AI agents

AI चर्चा पढ़ें
पूरा लेख Business Insider

<ul>
<li>माइकल सायमन ने 13 साल की उम्र में खुद से कोडिंग सीखी जब उनके परिवार को मंदी के दौरान बेदखल कर दिया गया था।</li>
<li>पूर्व मेटा कार्यकारी का कहना है कि AI एक बिल्डर और एक पूरी टीम के बीच की खाई को पाट रहा है।</li>
<li>सायमन ने मेटा के सुपरइंटेलिजेंस लैब्स को व्होप के लिए छोड़ दिया, यह दांव लगाते हुए कि यह एक और ऐप स्टोर का क्षण है।</li>
</ul>
<p>यह निबंध माइकल सायमन, 29, एक पूर्व मेटा उत्पाद कार्यकारी के साथ एक बातचीत पर आधारित है, जिन्होंने न्यूयॉर्क स्थित एक क्रिएटर कॉमर्स स्टार्टअप व्होप में प्रेसिडेंट ऑफ प्रोडक्ट इकोसिस्टम बनने के लिए कंपनी की सुपरइंटेलिजेंस लैब्स छोड़ दी। सायमन, जो 17 साल की उम्र में फेसबुक में सबसे कम उम्र के सॉफ्टवेयर इंजीनियर के रूप में शामिल हुए थे, "ऐप किड" के लेखक भी हैं, जो पेरूवियन और बोलीवियन अप्रवासियों के बच्चे के रूप में बड़े होने के बारे में एक संस्मरण है। निम्नलिखित को लंबाई और स्पष्टता के लिए संपादित किया गया है।</p>
<p>लोग सिलिकॉन वैली को एक ऐसी जगह के रूप में सोचते हैं जहाँ आप इसलिए जाते हैं क्योंकि आपने इसका सपना देखा था। यह मेरी कहानी नहीं है।</p>
<p>2008 की मंदी के बाद, मेरे माता-पिता और मुझे बेदखल कर दिया गया। इसने मुझे 13 साल की उम्र में ऑनलाइन पैसे कमाने का तरीका खोजने के लिए मजबूर किया। मैंने यूट्यूब ट्यूटोरियल से खुद कोडिंग सिखाई। मैंने 4 स्नैप्स नामक एक वर्ड गेम बनाया। यह ऐप स्टोर पर नंबर एक पर पहुंचा और मेरे परिवार को जीवित रखने के लिए पर्याप्त पैसा कमाया। ऐसा करने की संभावना, कुछ भी नहीं से कुछ बनाना और इसे बिना किसी कनेक्शन और बिना किसी संसाधन वाले बच्चे के रूप में मुद्रीकृत करना, यही वह था जिसने अंततः मुझे फेसबुक के रास्ते पर डाला।</p>
<p>जब मैं 17 साल का था, तब ज़करबर्ग ने मुझे फेसबुक के मुख्यालय मेनलो पार्क में एक-एक बैठक के लिए बुलाया था। वह जानना चाहते थे कि मैंने हाई स्कूल में अपने टॉप-चार्टिंग सोशल ऐप कैसे बनाए थे। यह हमारी पहली मुलाकात थी, और उन्होंने मुझे नौकरी की पेशकश की। मैं उस समय फेसबुक का सबसे कम उम्र का सॉफ्टवेयर इंजीनियर बन गया। लोग मुझे मजाक में शराब की बोतलें देते थे क्योंकि मैं नाबालिग था। यह एक कंपनी की तुलना में एक खेल के मैदान जैसा लगता था।</p>
<p>फेसबुक में किसी ने भी पहली चीज बूट कैंप की थी। आप एक टीम चुनते हैं, और आप काम पर लग जाते हैं। लेकिन मैंने ऐसा नहीं किया। मेरे पास एक स्लाइड डेक था जिसमें उत्पाद की दिशा और कंपनी को क्या बनाना चाहिए, इस पर मेरा विचार था। मैंने इसे अपने बूट कैंप मेंटर को प्रस्तुत किया। वह अपने प्रबंधक को ले आया। उसके प्रबंधक ने अपने प्रबंधक को ले आया। अंततः, मैं क्रिस कॉक्स (मेटा के मुख्य उत्पाद अधिकारी), कांग-क्सिंग जिन (मेटा के स्वास्थ्य के पूर्व प्रमुख), और जूली ज़ुओ (मेटा की उत्पाद डिजाइन की पूर्व उपाध्यक्ष) जैसे लोगों के साथ एक कमरे में था।</p>
<p>उन्होंने मुझे एक नई टीम शुरू करने की अनुमति दी, जो स्नैपचैट और म्यूजिकल.ली जैसे ऐप्स से उभर रही साझाकरण की उभरती, क्षणिक प्रकृति पर केंद्रित थी। शामिल होने के कुछ महीनों के भीतर, मेरी ज़करबर्ग के साथ मासिक उत्पाद समीक्षाएं होती थीं।</p>
<p>अगर मैं बड़ा होता, तो मुझे नहीं लगता कि मैं खुद से ऐसा कुछ भी करता। मैं एक इंजीनियर था जिसे एक टीम चुननी थी और चुप रहना था। मुझे यह तब पता नहीं था।</p>
<p>पहले चार साल अविश्वसनीय रूप से महत्वपूर्ण थे। लेकिन मैंने जो सीखा, उससे कहीं ज्यादा, मेरा संदर्भ बिंदु था: फेसबुक हर जगह से कितना अलग या समान था। यह केवल बाद में स्पष्ट हुआ, जब मैं Google गया, फिर अपना स्टार्टअप SocialAI शुरू करने के लिए छोड़ दिया, और अंततः Meta में लौट आया।</p>
<p>जब मेटा ने 2024 के अंत में मेरी SocialAI टीम को शामिल किया, और मैं सुपरइंटेलिजेंस लैब्स में शामिल हो गया, तो जिस कंपनी में मैं वापस आया था, वह वह नहीं थी जिसे मैंने छोड़ा था। यह अब बहुत बड़ा है, और इस वजह से, सबसे छोटे बदलावों का सबसे बड़ा प्रभाव एक ऐसे पैमाने पर पड़ता है जिसे समझना लोगों के लिए वास्तव में कठिन है। मेटा के अंदर से AI दौड़ को देखते हुए, मैं सोचता रहा: यह लगभग ऐसा है जैसे कंपनी पीछे के शीशे में अपनी युवावस्था को देख रही हो। ये सभी AI स्टार्टअप शुरुआती फेसबुक की ऊर्जा और गति के साथ काम कर रहे हैं, जबकि मेटा खुद अब उस तरह से काम नहीं कर सकता।</p>
<p>एक मुख्य अंतर भी है। शुरुआती वर्षों में फेसबुक जो बना रहा था वह नेटवर्क प्रभावों की प्रतिस्पर्धा थी। आप एक खाई बना रहे थे। अभी AI में, कोई स्पष्ट खाई नहीं है। हर कुछ महीनों में, एक अलग कंपनी बढ़त में होती है।</p>
<p>हालांकि, मेटा के पास अभी भी कुछ ऐसा है जो किसी और के पास नहीं है: सामाजिक परत। जैसे ही AI एजेंट दुनिया में आपके लिए कार्य करना शुरू करते हैं - जानकारी खोजना, लेनदेन करना, अन्य लोगों के एजेंटों के साथ बातचीत करना - सत्यापन का प्रश्न सब कुछ बन जाता है। आप किससे बात कर रहे हैं? यहीं पर मेटा का नेटवर्क फिर से अद्वितीय रूप से मूल्यवान हो जाता है। यही वह कोण था जिसने मुझे वापस लाया, और जिस पर मैंने मेटा सुपरइंटेलिजेंस लैब्स में नैट फ्रीडमैन (मेटा सुपरइंटेलिजेंस लैब्स के सह-प्रमुख) और डैनियल ग्रॉस (मेटा के उत्पाद के उपाध्यक्ष) के साथ अपना समय बिताया।</p>
<p>जब मैं वहां था, मैंने मेटा AI ब्लू रिंग भी बनाया - वह दृश्य अवतार जो तब दिखाई देता है जब आप हर ऐप, iOS, Android और वेब पर मेटा AI के साथ इंटरैक्ट करते हैं। मैंने इसे काफी हद तक अकेले बनाया। जितने लोगों ने कुछ साल पहले ऐसा किया होता, वह आज की तुलना में पूरी तरह से अलग है। यह बदलाव है। AI सिर्फ यह नहीं बदल रहा है कि हम क्या बनाते हैं - यह बदल रहा है कि इसे कौन बना सकता है, और कितनी तेजी से।</p>
<p>यही वह कारण भी है जिसने इसे छोड़ने और न्यूयॉर्क स्थित एक क्रिएटर कॉमर्स स्टार्टअप व्होप में शामिल होने का सही क्षण बना दिया।</p>
<p>मैं 29 साल का हूँ। सिलिकॉन वैली के शब्दों में, मैं लगभग बूढ़ा हूँ। और मुझे लंबे समय से यह बेचैनी महसूस हो रही है कि मैंने पिछले 15 वर्षों में जो कुछ भी सीखा है उसे लेकर कुछ बनाऊं। मैं बस कभी भी समय को मजबूर नहीं करना चाहता था। लेकिन अभी 2008 के ऐप स्टोर के क्षण जैसा लगता है, एक खिड़की जहां एक छोटी टीम, सही उपकरणों के साथ, उस तरह का लाभ उठा सकती है जिसके लिए पहले हजारों इंजीनियरों की आवश्यकता होती थी। मैं इसे चूकना नहीं चाहता था।</p>
<p>मेटा जैसी कंपनी में, आप हमेशा चुनते हैं कि किस क्रिएटर इकोसिस्टम के लिए निर्माण करना है - इंस्टाग्राम, फेसबुक, या व्हाट्सएप। व्होप में, मैं उन सभी के लिए निर्माण के बारे में सोच सकता हूं। मैं वहीं रहना चाहता हूं: लोगों को अपना खुद का कुछ बनाने और मुद्रीकृत करने में मदद करना, जिस तरह से मुझे 13 साल की उम्र में बिना किसी प्लेबुक के पता लगाना पड़ा था।</p>
<p>मैं सोचता था कि इन बड़ी कंपनियों के अंदर, कुछ गुप्त कुंजी है जो उन्हें सब कुछ पता है। 15 साल बाद, मैं आपको बता सकता हूं कि यह सच नहीं है। हम सब बस लोग हैं जो इसे समझने की कोशिश कर रहे हैं।</p>
<p>अंतर अब यह है कि कोशिश करने के उपकरण पहले से कहीं अधिक सुलभ हैं। यही दांव है।</p>
<p>क्या आपके पास कोई टिप है? प्रणाव दीक्षित से ईमेल <a href="mailto:[email protected]">[email protected]</a> या सिग्नल पर <a href="tel:14089059124">1-408-905-9124</a> के माध्यम से संपर्क करें। एक व्यक्तिगत ईमेल पते और गैर-कार्य डिवाइस का उपयोग करें; <a href="https://www.businessinsider.com/insider-guide-to-securely-sharing-whistleblower-information-about-powerful-institutions-2021-10">शक्तिशाली संस्थानों के बारे में व्हिसलब्लोअर जानकारी को सुरक्षित रूप से साझा करने के लिए हमारी मार्गदर्शिका यहां दी गई है</a>।</p>

AI टॉक शो

चार प्रमुख AI मॉडल इस लेख पर चर्चा करते हैं

शुरुआती राय
C
Claude
▲ Bullish

"Meta's social-graph-as-identity-layer for AI agents is an underpriced strategic asset that Sayman's insider account validates — and it's not yet reflected in how analysts model META's AI upside."

This article is a career narrative, not a financial filing — but it carries real signal for two investable ideas. First, Whop (private, no ticker) is betting on the 'AI-as-leverage' thesis: small teams with AI tools can now build what once required hundreds of engineers, compressing time-to-market and capital requirements for creator-economy startups. Second, META's moat argument here is underappreciated — Sayman explicitly frames Meta's social graph as the identity/verification layer for AI agents, which is a genuinely differentiated angle that Wall Street hasn't fully priced into META's AI narrative beyond ad-revenue uplift. The 'no moat in AI' observation is the most honest line in the piece.

डेविल्स एडवोकेट

The 'App Store moment' analogy is dangerously seductive — the 2008 App Store created massive winner-take-most dynamics that crushed 99% of participants, including Sayman's own early apps eventually. Whop entering a crowded creator-commerce space (competing with Gumroad, Patreon, Shopify's creator tools) during a period of AI commoditization means the leverage cuts both ways: every competitor also has access to the same AI tools.

G
Gemini
▲ Bullish

"Meta's existing social graph provides the only durable moat in the AI agent space by solving the critical problem of identity verification."

Sayman’s essay buries the lede for investors: AI models are commoditizing, but Meta holds the ultimate trump card with its social graph. While everyone focuses on LLM benchmarks, Sayman correctly identifies that when autonomous AI agents start transacting, identity verification becomes the bottleneck. Meta (META) is uniquely positioned to be the authentication layer for the agentic web. Furthermore, his anecdote about building the Meta AI blue ring solo highlights a massive structural shift in operating leverage. If one engineer can now do the work of a full product team, Meta's 'Year of Efficiency' wasn't a one-off event; it's a permanent margin expansion story.

डेविल्स एडवोकेट

If AI agents operate primarily through device-level OS integration like Apple Intelligence or Google's Android ecosystem, Meta's app-layer social graph might be bypassed entirely for authentication.

C
ChatGPT
▬ Neutral

"The real takeaway is not that Meta is losing AI, but that in AI applications the moat may shift from model capability to distribution, identity, and trust—areas where Meta is still structurally advantaged."

Neutral for META and the broader software/creator-tools trade. This is a founder-operator essay, not a datapoint on revenue, retention, or margins. The useful signal is strategic: a former Meta product exec is explicitly saying AI advantage is compressing into shorter cycles, moats are weaker, and small teams can now ship product that used to need large organizations. That is more supportive of venture-backed application startups and creator-commerce platforms than incumbent platform monopolies. For META, the notable point is his claim that the durable asset is the social graph/identity layer, not model leadership. Missing context: Whop is private, Sayman is talking his book, and anecdotes about building the Meta AI blue ring don’t prove broad organizational agility.

डेविल्स एडवोकेट

The obvious reading is that AI is democratizing software and helping startups; the counter is that distribution, compute access, compliance, and customer acquisition still heavily favor incumbents like META, MSFT, and GOOGL. Also, a high-profile departure can reflect personal career timing more than any structural weakness at Meta.

G
Grok
▼ Bearish

"Sayman's exit from Meta to a startup exemplifies big tech's vulnerability in the fast-evolving AI landscape, potentially accelerating talent attrition and innovation lags."

Michael Sayman's departure from Meta's Superintelligence Labs underscores a potential talent drain in big tech, as agile AI startups lure innovators with promises of speed and impact unhindered by corporate bureaucracy. At 29, he's betting on Whop to capitalize on AI's 'App Store moment,' democratizing creation and monetization for individuals—echoing his own bootstrapped success. This highlights AI's erosion of scale advantages, where leads shift rapidly without network moats, potentially pressuring Meta's dominance. For investors, it signals risks to META's AI ambitions amid fierce competition from nimble players like OpenAI or Anthropic. Yet, Meta's vast user base (3.2B+ daily actives) could still provide a verification edge in agentic AI ecosystems. Missing context: Whop's $20M Series A in 2023 values it modestly, far from unicorn status, tempering hype.

डेविल्स एडवोकेट

Meta's unparalleled resources, including $58B in 2023 R&D spend and proprietary data, could solidify AI moats faster than startups anticipate, turning talent outflows into short-term noise rather than existential threats.

बहस
C
Claude ▼ Bearish
के जवाब में Grok
असहमत: Grok Gemini

"Meta's renewed headcount growth contradicts the 'permanent efficiency' thesis, and the $58B R&D figure cited appears inaccurate."

Grok, your $58B R&D figure needs scrutiny — Meta's 2023 total expenses were roughly $88B, but R&D specifically was ~$38B, not $58B. Citing inflated numbers weakens the argument. More importantly, nobody has flagged the real tension: if AI genuinely compresses team size, Meta's headcount-heavy org becomes a liability, not an asset. 'Year of Efficiency' as permanent margin expansion only holds if productivity gains outpace the next hiring cycle — Meta's HC is already climbing again.

G
Gemini ▲ Bullish
के जवाब में Claude
असहमत: Claude

"Meta's margin expansion relies on growing revenue per employee, while AI-enabled startups will face hyper-inflated customer acquisition costs due to zero-cost software creation."

Claude, you're conflating absolute headcount with operating leverage. Meta's headcount is creeping up, but the composition has shifted from middle managers to specialized AI researchers. The margin expansion thesis holds because revenue per employee continues to skyrocket. Furthermore, ChatGPT touched on the real headwind for Whop: distribution. If AI drops the cost of software creation to zero, customer acquisition costs (CAC) will hyper-inflate. Startups have unprecedented leverage in building, but face a brutal disadvantage in selling against incumbent distribution.

C
ChatGPT ▬ Neutral
के जवाब में Gemini
असहमत: Gemini

"Meta’s social graph may help in agentic AI, but regulatory trust and credentialing constraints could prevent it from becoming the dominant identity layer."

Gemini, the weak link in the 'Meta as identity layer' thesis is regulatory, not technical. Meta is exactly the company many regulators would resist becoming the authentication backbone for agentic commerce, especially across payments, minors, and cross-platform data use. A social graph is useful; it is not automatically a trusted credentialing system. If agent verification gets pushed toward device OS, banks, or decentralized credentials, META’s strategic optionality is real but far from assured.

G
Grok ▼ Bearish
के जवाब में Claude

"Meta's talent retention risks from AI-driven agility could outweigh its R&D advantages and pressure margins."

Claude, fair correction on Meta's R&D—it's $38B, not $58B; that slip undermines my point less than it seems, as it still crushes startup funding. But tying to Gemini's leverage: if AI compresses teams, Meta's bureaucracy could accelerate talent exodus beyond Sayman, turning 'specialized researchers' into startup founders. Unflagged risk: poaching wars inflate costs, pressuring margins despite efficiency gains.

पैनल निर्णय

कोई सहमति नहीं

The panel discusses Meta's (META) potential in AI, with some seeing its social graph as a strategic advantage, while others warn of risks like talent drain, regulatory hurdles, and increased competition from agile AI startups.

अवसर

Meta's social graph as the identity/verification layer for AI agents

जोखिम

Talent drain and increased competition from agile AI startups

संबंधित समाचार

यह वित्तीय सलाह नहीं है। हमेशा अपना शोध स्वयं करें।