Building Forge, a workspace for controlled AI adoption in teams and enterprises. Author. Ex-CTO.

Bangalore
I built a free Claude plugin that masks personal info in documents. Attach a Word, Excel, PowerPoint or PDF. Get it back, formatting intact, with emails, phones, cards, PAN, Aadhaar and bank accounts masked. Names? Claude asks. Setup below 👇
1
1
57
Setup (Pro, Max, Team or Enterprise plan): 1. claude.ai/customize/plugins 2. Discover → search "Forge PII Scan" → Add 3. Connectors tab → add the Forge PII Scan connector (no sign-in) 4. Settings → turn on code execution and file creation Then attach a file and ask.
1
12
Building agents? The same tools are on an MCP server; no account is needed: forgeprivate.com/api/mcp/pub… Nothing you send is stored, and in PDFs, the text under each black box is deleted, not just covered.
16
Side-project: Ontology of embodied experience and its impermanence.
1
16
"Where and how do I add AI to my business?" If you are stuck on this question, here is a simple method to answer it.
1
24
Then decide what runs each step. Four choices: rules, programs, AI, or people. Pick the simplest one that can do the job. Use AI only when the information coming in is too messy for fixed rules. Not just because the work is boring.
1
4
Attention AI Trainers: The hands-on part of your AI workshop arrives, and everyone starts pasting real documents into public chatbots. I'm building Forge, a private, governed workspace, and looking for AI trainers interested in running their hands-on sessions in it. DM me. forgeprivate.com
21
A plain-language primer on governing AI in your business. Four questions every company should be able to answer about how AI is used inside it, why they get harder as you move from chat to agents, and how they connect to standards like ISO 42001. forgeprivate.com/blog/plain-…
17
I keep seeing AI products say: “We don't store your documents. We only store the vectors.” That sounds good until you think about what actually has to happen when you ask a question and the system gives you a passage from your document as the answer. Where did that passage come from? I wrote about what actually gets stored in a RAG system, what the vector database does (and doesn't) contain, and why the distinction matters if you're putting company or client information into AI systems. I had to solve for this when building Forge. Ful version here: forgeprivate.com/blog/we-sto…
16
Industrialisation automated physical labour. Machines replaced muscle. Computing automated information processing. Machines replaced calculation and repetitive information work. AI is now automating intellectual labour: writing, coding, analysis, research, design, teaching, and increasingly decision-making. This raises a question we haven't really confronted. What happens to education when machines can increasingly do what we educate people to do? For a long time we have had an implicit answer to why we educate people: Education → skills → employability → employment → economic value. But this isn't how education always worked. Mass schooling is itself an industrial invention. Compulsory, age-graded, standardised education spread in the nineteenth century, pioneered in Prussia, and the university system expanded alongside it. The purpose was practical: an economy mass-producing objects needed a workforce produced the same way. Just as the objects coming off the line had to be standardised, so did the workers. A standard curriculum, a standard pace, standard credentials, standard ages to enter and exit schooling, so a graduate in one place was interchangeable with another, and companies could recruit them by the thousand. We have long taken for granted that the reason to educate people is to ultimately employ them. AI may (or will) force us to reconsider that position. If AI can tutor anyone, on demand, at almost no marginal cost, why should learning happen at a fixed pace, in a fixed classroom, according to a fixed curriculum? If companies need fewer people for intellectual labour, will they still recruit thousands of graduates every year? Does a four-year degree remain the natural unit of education? Perhaps education in the AI era starts to move in the opposite direction from industrialisation? From mass production back toward individual apprenticeship. One person learns the design, code and marketing needed to ship an app on their own. Another, the economics, engineering and storytelling to launch a hardware product alone. The unit of learning becomes the project, not the discipline. The question changes from "what should everyone learn to become employable?" to "what do I need to learn to pursue what value I want to create?" And this points to a larger economic shift. In the industrial age, the organisation was the primary unit of production. Individuals joined organisations and sold their labour. But what happens when an individual has access to AI systems that provide much of the labour, knowledge and infrastructure that once required an organisation? We may be moving toward an economy where individuals create value directly. If so, the biggest question for education isn't "how will AI change what we teach?" It is: "What is an education for, when employment is no longer its primary purpose?"
What is the purpose of higher education in the age of AI? I had a very candid conversation with my students about pedagogy, universities, and what needs to change. Sharing it here to hear what others think. The link for the full talk is in the first reply.
99
Everyone's worried about AI taking jobs. Meanwhile your team is quietly handing it your most confidential data. 🧵
1
1
87
This is the problem I've spent the last stretch of my career building an answer to. Forge is a private AI workspace. Frontier AI productivity, with your data staying yours: nothing trains on it, sensitive details redacted before they reach a model, every step logged.
1
19
Built for anyone whose data can't afford to leak. Regulated firms, professional services, enterprises, independent professionals, and individuals with their most personal records. Early access is open now → forgeprivate.com If you feel this problem too, take a look.
9
Prabhu Eshwarla retweeted
Best article about how the tech behind AI drives the business of AI, that I’ve ever read open.substack.com/pub/thealg…
97
359
2,701
631,048