When not to use AI
AISkcript Applied Labs
When not to use AI
Some tasks are better solved with a rule, a better form or a small process change. Here is how we spot them, and how we say so during a pilot.
Karthik Kamalakannan · · 4 min read
Blog
Skcript Applied Labs
What a 10-day pilot actually delivers
AISkcript Applied Labs
What a 10-day pilot actually delivers
Ten working days sounds short for AI work. Here is what happens on each day of our pilot, what you have at the end, and what ten days can't do.
Karthik Kamalakannan · · 4 min read
Keeping your data in the UAE
AISkcript Applied Labs
Keeping your data in the UAE
"Our data can't leave the UAE" is the first thing most security teams tell me. In AI work that one sentence hides three separate questions: where data is stored, where the model runs, and who can see it.
Karthik Kamalakannan · · 5 min read
AI and UAE e-invoicing
AISkcript Applied Labs
AI and UAE e-invoicing
E-invoicing will turn most supplier invoices in the UAE into structured data. That changes where AI is useful in accounts payable, and where plain rules do the job better.
Karthik Kamalakannan · · 4 min read
We tested Claude, OpenAI and open models on your work. Here's how.
AISkcript Applied Labs
We tested Claude, OpenAI and open models on your work. Here's how.
There is no single best AI model. There is the model that does your task well enough, at a cost you can live with. Here is how we build a scorecard from your real cases to find it.
Karthik Kamalakannan · · 3 min read
Why AI pilots stall
AISkcript Applied Labs
Why AI pilots stall
Most AI pilots don't fail because the model is weak. They stall in five gaps between a demo and a system people rely on. Here is how to spot each one in your own pilot.
Karthik Kamalakannan · · 3 min read
Introducing Skcript Applied Labs
AISkcript Applied Labs
Introducing Skcript Applied Labs
A personal note from Karthik Kamalakannan, Founder and CEO of Skcript, on why we are launching Skcript Applied Labs: one task, your own numbers, every model tested on your cases, and a working first piece in 10 days at a fixed price.
Karthik Kamalakannan · · 5 min read
Cost per task, not cost per token
AISkcript Applied Labs
Cost per task, not cost per token
AI prices are quoted per token, but your business pays per finished task. Here is how I think about the real cost of AI work, and how we measure it in a pilot.
Karthik Kamalakannan · · 4 min read
Do we need to fix our data first?
AISkcript Applied Labs
Do we need to fix our data first?
It is the most common reason I hear for waiting on AI. For most real work, a set of real examples is enough to start. Here is why, and what we actually need.
Karthik Kamalakannan · · 4 min read