Latest post
TechnologyThe AI infrastructure boom is partly lending to itself. That changes the cost curve you're planning around, and most companies building on it haven't noticed.
AI Advisor · Founder, 256 Technologies
Hitesh
Sarda
I help leadership teams cut through AI hype and make evidence-based decisions: what is feasible, what the ROI really is, and the smallest pilot that proves it. 25+ years building production systems, including work at Citi, Virtusa, Yash, and ValueLabs. Today I run 256 Technologies, an applied AI lab in Hyderabad, and I have been writing here since 2009.
178
posts
7
topics
2009
since
New book · Free PDF
Implementing AI Agents in Your Company
A Blueprint for Innovation
A practical, hype-free playbook for the managers and operators who have to turn AI agents into real business outcomes. Read it free, no signup.
What I do
Book an advisory hour →
AI Advisory
Acting as your fractional Head of AI: strategy, feasibility, ROI modeling, vendor assessment, and roadmaps grounded in real engineering constraints. No hype, no vendor lock-in.
AI Executive Training
A six-week program that builds an AI-native leadership team: hands-on executive workflows, governance and risk, and a 90-day adoption roadmap that moves leaders from watching to leading.
MVP Development
Tangible code beats theoretical roadmaps. I build the smallest prototype that answers the key technical and business questions, across agents, computer vision, and ML, before you commit to a full build.
Recent Writing
Full archive →
Someone is subsidizing your AI compute
The AI infrastructure boom is partly lending to itself. That changes the cost curve you're planning around, and most companies building on it haven't noticed.
AI does your hardest work, not your messiest
A frontier model just produced a novel math proof while similar tools still fumble routine operations. The dividing line is verifiability, not difficulty.
Why the smartest model can make your product worse
Frontier models keep setting records, yet what decides whether a real-time AI product works is the second your user waits, not the model's score.
Your backlog is the only benchmark that matters
Public coding benchmarks have decoupled from real work. The teams getting value from AI build a small evaluation from their own merged pull requests instead.
When the regulator becomes your product manager
Every new car in Europe now ships an eye-tracking AI model no buyer chose. How that feature is failing is a preview of what happens when AI gets mandated onto a product.
Your model was never your moat
Open-weight models now match frontier quality at a fifth of the cost. For anyone building on AI, that shifts where durable advantage has to come from.
The Autonomy Budget: Deploying AI Agents Before They're Reliable
Meta is spending $145B on AI yet says agents have stalled. The real limit is compounding math, and it decides which workflows you can safely hand to an agent today.
The vulnerability Google won't call a bug
A researcher found a way to leak private YouTube data through a comment. Google says it's not a security bug. That answer is more revealing than the exploit.
