Sonal Goyal

Entity Resolution is foundational for data management, yet it is extremely complex to build. How many of you resonate?

Claude is almost like the cofounder I never had!

From “What Was Revenue Last Quarter?” to “Generate 30% More Revenue This Quarter”, by @sonalgoyal open.substack.com/pub/sonalgoyal…

9/So, a direct question to everyone building or buying enterprise software: If your SaaS vendor made your data fully portable tomorrow — accessible by any agent you build — would that change how you evaluate them? Or are you already walking away from vendors who won't?

8/The SaaS vendors who thrive won't be the ones who held their data tightest. They'll be the ones who had the vision — and the courage — to let it flow.

1/Vibe coding has everyone debating whether SaaS is dead. Both sides — "just build it yourself" vs "a POC isn't a product" — have a point. But they're both arguing about the wrong thing. 🧵

@clairevo Why hasn’t @Apple innovated here? 🤦‍♀️

@hellomishraa No, is there one?

@sonalgoyal Up for it. Are you part of any community?

Any builders in data or ai in Noida/NCR? Would love to meet.

🚨 Two strong women. Helpless. LIVE on television. Trisha 😢: A survivor who fought for years - now dragged, silenced, abandoned. Sonal breaks down on air. The system has failed all of us 💔

Clarifying as someone who treats lung cancer. There is a direct link between air pollution and lung disease

@JaredSleeper The missing context is Customer 360

Expounding on the "Systems of Context" concept. What's clear is that what AI models need to function optimally and what human users need to function properly is very different. Humans are carrying around boatloads of context- definitions, relationships, KPIs, business priorities, etc. This stuff is not usually written down/shared widely. Even stuff like who has influence within a company, etc. To get the best results from a model, regardless of whether you are generating an AI ad or having an agent do a business task, you need to give it tons of context. The JSON prompts we're seeing pop up for NanoBanana are a great example of this- easy to see a world where they get much longer/more complex. The issue is that the context doesn't exist in written/usable form. It is a mistake to assume systems of record have it all- they don't. It's a great starting point, but woefully insufficient. The exercise is- what company can most efficiently collect and organize that context in a way that is LLM legible. Systems of record exist because it is enormously valuable for a company to operate off of one, standardized source of truth. The more of that is digital, the better. LLMs put that on steroids- so systems of record need to go on steroids and become systems of context, eventually even with LLM input on what data/info is useful to collect.

@sajithpai Any data on SAFE rounds?

There is no one seed round. There are broadly 3 tiers of seed rn in India. Dilution is anywhere from 17-25%, but typically 20-22% for any of these rounds. $1-2m - almost always first time founders, in nonAI, nonQC,; have some revenue + customer validation, and have raised a $300-700k preseed + angel round before. $2-4m - This is a formation cheque usually: first time founders in a hot space (like vertical QC, AI companion etc) or 2nd time founders / Sr Operators but keen to build in a non-hot space; OR it is a traction cheque - it has raised a $300-700k preseed round but has strong traction. $4-10m+ - Formation cheque: 2nd time founders / Sr Operators in a hot space. This is effective a Series A sized round in seed stage.