Sanjeev Mohan

** Rising Concern Over an AI Bubble** Sam Altman, CEO of OpenAI, recently cautioned that investor excitement may be “overexcited,” even likening it to the dot-com bubble of the late 1990s. A MIT study revealed that 95% of enterprise generative AI initiatives show no financial return, igniting fears of overvaluation Firstpost recently covered this news well!! I loved the punch line at the end - ‘Revolutions don't move in a straight line. There will be hype and there will be crashes.’

13

2 Comments

Data integrity is the non-negotiable foundation for successful AI deployment - not the model itself. In my recent article for ET Edge Insights, I discuss how eliminating ROT (Redundant, Obsolete, or Trivial data), automating data checkpoints, and adopting a security-first approach are critical steps for building AI that earns trust. How is your organization prioritizing data integrity to truly unlock your AI strategy? Do share your thoughts in the comments. Iron Mountain #DoubleUp https://lnkd.in/d5NXQHjr

85

Recent analyses of large-scale citation behavior in conversational AI systems have suggested that visibility within citation graphs offers a proxy for influence and decision relevance. This article explains why that assumption fails in decision-adjacent contexts, and why reliance on citation data creates a false sense of governance control. Drawing on a multi-model evidentiary record of observed external AI behavior, we show that citation presence neither predicts nor explains decision eligibility, authority filtering, or entity substitution. Citation data describes a surface phenomenon. Decision exclusion occurs elsewhere, under different rules, and at later stages of model execution, leaving organisations unable to reconstruct or defend what influenced real decisions. #AIVOJournal #AIGovernance #AIAccountability #EnterpriseAI #AEO #GEO (link in comments)

4

3 Comments

I want to steel man this awesome blog from my colleague Eric Range on all the bells, whistles, agents and more we showcased at the Element451 Engage Summit at the end of last month to address some of the skeptics. (Note: I used AI as the skeptic) 🗣️Is the 381 work days claim just hype? -Multiple colleges echo similar productivity or enrollment gains. This is not a single cherry-picked pilot. https://lnkd.in/eTGvAMn7 ⚠️Autonomous scoring and risk flags will violate privacy or bake in bias. -FERPA & SOC 2 Compliant: Element451 encrypts all records at rest + in transit, and supports MFA. -No black-box scoring: The App Reader Agent only evaluates criteria that your admissions team explicitly writes. If reviewers omit a rubric item, the agent skips it; if they revise thresholds, the agent follows the update from that point forward. -Full audit trail: Every AI action scores, flags, and nudges. All of this is surfaced in the UI with rationale, timestamps, and user-level access logs that satisfy FERPA’s “legitimate educational interest” rule. 🙈Students won’t trust hidden AI nudges; transparency is weak. -Identity-verified chat/voice: Before any PII is revealed, Bolt Agents authenticate the caller or chatter, mirroring staff workflows. -Explainable outputs: App Reader, Fraud Detection, and Success Agents all expose the factors behind each score or flag so human staff can override or amend. -Institution-controlled messaging: Campaign templates and nudges are editable; schools decide tone, cadence, and disclosure language. Staff preview every outbound message before launch. 😬Over-automation will deskill staff or create unhealthy dependence. -The adoption guide positions AI as a relief valve for “boring, time-consuming tasks no one likes,” freeing staff for strategy and relationship-building. -Staff Agent shortcuts (“Catch Me Up,” task ribbon, on-demand drafts) are explicitly designed to augment daily work, not replace it. -Once repetitive triage moves to AI, teams re-allocate hours to high-touch outreach (e.g. 1:1 advising, doing more strategic work ) rather than shrinking headcount. Our clients tell us this. Closing Take Institutions still need governance guardrails, but the “AI workforce” is already producing measurable ROI without sacrificing trust, privacy, or human judgment. There is a better option than a wait and see approach or hoping things get better on the enrollment yield, student retention, and staff recruiting. Many of our clients were AI curious doing ad-hoc AI projects and using AI endpoint solutions. They now have an ever improving platform + AI workforce and are becoming Ai-first institutions. Read the blog and test the agents out yourself https://lnkd.in/eGspwdvv

18

1 Comment

95% of enterprises spent millions on AI. Zero meaningful returns. The math doesn't add up until you see what they actually bought. Most companies fell for the "silver bullet" promise. They bought flashy LLMs without clear use cases. The real issue? Random AI experiments without strategy. VCs are now predicting 2026 as the turning point. But only for companies that shift focus: • Target specific business problems • Integrate AI into existing workflows • Focus on data ownership over model performance • Cut vendor sprawl and concentrate budgets The winners won't be chasing the latest AI trend. They'll be solving real problems with proven tools. Startups need $1-2M ARR and clear enterprise adoption to survive. The AI consulting pivot is already happening. By late 2026, we might see universal AI agents. But first, companies need to master the basics. Stop buying AI for the sake of AI. Start building solutions that actually move the needle. What's your take on the AI ROI challenge? #AI #Enterprise #ROI 𝗦𝗼𝘂𝗿𝗰𝗲꞉ https://lnkd.in/giyX7R6E

14

4 Comments

There is a realization that foundation models enable differentiation when they have context. I’ve had multiple conversations recently that reinforce what this "When Every Company Can Use the Same AI Models, Context Becomes a Competitive Advantage" piece highlights: the advantage is shifting to how well organizations can encode their internal reality (workflows, decisions, and tacit knowledge) into systems that AI can actually use. The model is the engine, but context is the fuel. This becomes even more critical in asset-heavy industries like oil & gas, mining, and manufacturing where decades of technical debt are embedded in legacy systems running core operations. What’s emerging now is the ability to use AI to effectively create a digital twin of these systems; not just at the data layer, but at the intent layer. By surfacing and structuring what that old code was actually designed to do, we can begin to understand and map technical debt in a way that was previously impossible. Context graphs become the bridge between what exists and what should exist. And this is where the economics fundamentally change. Traditional modernization efforts cost hundreds of millions and take years, often becoming obsolete before they’re complete. But when AI can generate software rapidly against a well-structured context graph, the constraint shifts from coding cost to clarity of context. That unlocks a new lens for CFOs: capital allocation is no longer about funding massive rewrites, but about investing in context infrastructure that enables continuous, AI-driven system regeneration. The companies that internalize this shift will compress decades of technical debt into years, and redefine their operating leverage in the process. Read more here: https://lnkd.in/g-4gmkjN Atanu Roy, Sujoy Roy, Deepak Dalakoti, Yuho Kida, Amelia Taylor, Ashley Walker, Amanda Cohen, Jaime Wilde, Sri Elaprolu, Asa Kalavade, James Vulling #AI #DigitalTransformation #TechnicalDebt #ContextEngineering

46

3 Comments

There is no AI without data. Full stop. My pal Deepak Shah, CDO of the Department of Defense shares this 14 pillar framework for scaling AI. . What are your tips for scaling aI that actually yields roi? #data #ai ##aivillage #stratregy

31

9 Comments