About Ashish
Nineteen years in enterprise technology, mostly in the unglamorous middle distance between what gets demoed and what actually ships.
My interest sits at three intersecting threads: enterprise CRM, the long arc of how large organisations sell and serve; customer experience, the surface where that promise either holds or breaks; and, more recently, agentic AI, where both of those threads meet a frontier still finding its shape.
This site is where I write longer-form, watch the field, and think out loud. The interest is in the architecture and craft of things that work, not in selling solutions. What I want to leave behind here are essays that are honest about what's hard, signals that are useful to the people doing the work, and a record of the field's evolution that ages well.
The recurring threads.
Three subjects this site keeps returning to, picked because each is going through a real, structural change and there's still a lot worth saying that nobody is saying clearly.
Generative AI
What the field is genuinely getting right, what's overhyped, and which patterns are actually surviving contact with production systems.
Agentic systems
Multi-agent architectures, MCP and A2A, orchestration, evaluation, the unsexy plumbing under the demos.
CRM, the long arc
Two decades of large platform transitions. What changes, what doesn't, and what AI is actually doing to the category.
An honest point of view.
Most enterprise AI failures aren't model failures, they're architectural and organisational ones, dressed up in technical language.
- Adoption is the hard part. Most "AI failures" are change-management problems in technical clothing.
- Prefer evaluation over intuition. The best architectures emerge from honest measurement, not vibes.
- Treat governance as a first-class concern, not a retrofit.
- The unglamorous parts, onboarding, observability, rollback, audit trails, are where credibility gets built.
Things I keep coming back to.
Restraint over novelty
New capabilities matter only when they raise the floor of what reliably ships. Hype-driven adoption is the costliest kind of technical debt.
Continuous learning
Nineteen years in, and the half-life of useful frameworks keeps shrinking. Staying current is the work, not a side project.
Two-coast view
Working between New York and Mumbai. Different markets, different constraints, the same insistence on what actually ships.