Akhil Singh

Building and || BITSian

[Akhil Singh @akhil_bits]

FDE roles will see steep decline once context is institutionally solved as an infra layer
Over reliance on fde is because ai execution without context layer is a patch work

[Jaya Gupta @JayaGup10]

Too many people taking FDE roles right now are solving for consumption perks over compounding perks.

Consumption perks raise the quality of the recruiting experience: chefs, World Cup tickets, joining tokens, top-of-band dollas. Nice to have and yet completely orthogonal to your slope.

Compounding perks raise your derivative, your rate of learning, ownership, and future agency. They’re harder to see on an offer letter, which is exactly why they get underweighted.

[Akhil Singh @akhil_bits]

hot take but the “VCs are unprofessional” narrative doesn’t match my reality at all. almost every top-tier Indian fund I’ve met came prepared, engaged hard, and followed through. the bad ones are single digits-and they exist in every profession.

[Hemant Mohapatra @MohapatraHemant]

People will forget the graphs, charts, the analysis, but always remember how you make them feel. This VC vs founder thread is a reminder of how reputation is built in our business. I'm actually sure I've messed up here and there but the attempt is always to be additive 🙏

[Akhil Singh @akhil_bits]

@MohapatraHemant That’s why for enterprises, context is the only big problem remaining to be solved
Intelligence by itself is already good enough to execute almost everything for enterprises

[Akhil Singh @akhil_bits]

can pharma just use AI to one-shot the common cold and flu already? the most annoying of all

[Akhil Singh @akhil_bits]

I don't know anyone who uses ChatGPT anymore. Everyone's on Claude now.

[Akhil Singh @akhil_bits]

Enterprise harness derived from data, then juggled to bolt business context on afterward, is a failed strategy.

The approach that actually works does the opposite: the business graph is built and crafted first, then married to the data.

Data and System-of-Record companies carry the baggage of having taken that first route.

But enterprises are about to build the most important foundation layer-and they must build it once, and build it right.
https://t.co/ylUp7drnGu

[Akhil Singh @akhil_bits]

Few examples:
FMCG wanted Claude to analyze its marketing spend distribution
Roas calculation is different in different channels
Attribution window is different and has no trace anywhere
Cannibalism is not defined in data or systems
All 3 sits in marketing teams brain

[Patrick Hsu @pdhsu]

people aren’t celebrating this pancreatic cancer breakthrough enough. it should be international news

[Akhil Singh @akhil_bits]

@JayaGup10 would love your views here

[Akhil Singh @akhil_bits]

This is exactly what we have been solving for Fortune 500 from last many months with https://t.co/6OzvyehP3n
Also context derived from data to feed to llms is a stupid way
Consumes huge amount of time and money only fail in production
Also decision:context traces doesn’t sit in slack and gmail

[Garry Tan @garrytan]

This is the actual bottleneck. The models are smart enough already. What is missing is the company-specific context locked in senior people heads. Whoever cracks knowledge extraction at the company level unlocks the rest.

As you work on this, please consider using GBrain as your OSS retrieval layer

https://t.co/0F5uDQzPHu

[Akhil Singh @akhil_bits]

And hence a pure plug and play product is a recipe for failure
Consulting/SIs have a role to play
Like no org self configures SAP or Salesforce

[Akhil Singh @akhil_bits]

Few examples:
FMCG wanted Claude to analyze its marketing spend distribution
Roas calculation is different in different channels
Attribution window is different and has no trace anywhere
Cannibalism is not defined in data or systems
All 3 sits in marketing teams brain

[Akhil Singh @akhil_bits]

Recursive self improvement by combining harness and weights might actually be the missing block for super intelligence
Congrats on the launch @hexoai

[Kunal Bhatia @kunalbhatia91]

Superintelligence will be built on Self Improvement.

Today @hexoai, we’re excited to release ‘SIA’ - an open-source Self-Improving AI, to achieve any goal through recursive self improvement.

While trying to solve a problem, SIA doesn't just improve it's abilities by updating it's harness, it updates it's own weights as well.

[Akhil Singh @akhil_bits]

Most of the industry’s approach to AI harnessing is shallow. Large enterprises are no less than a living being. They have:

- DNA: taste, distribution, sales intensity. No LLM is replacing that.
- Information: captured in the past, still growing.
- Nervous system: how the org is structured, what it measures. - Actions: workflows.
- Intelligence: Of how to make decisions, how it uses all of the above to act.

For a real harness, everything needs to move symbiotically to use AI in enterprises.
And “Context graph with decision traces” is a very lazy attempt

[Mark™ @thinkwithmark]

We raised $1M dollars to reinvent how people read. Introducing Mark II - a $159 AI bookmark. Thread below

[Akhil Singh @akhil_bits]

Enterprises required reliable layer of data to build trustable analytics- Data Warehouse was born
Enterprises require reliable context and tribal intelligence layer to build trustable AI Agents- IntelligenceWarehouse is born
https://t.co/raSMJiiUyR

[a16z @a16z]

Our April GTM survey found that CRM usage has risen since AI tools began to be adopted at scale.

The agents that listen to calls and write structured notes back into the system are, for the moment, giving reps fresh reason to consult it, because the data sitting there has become dramatically richer than it used to be.

a16z's Gio Ahern, Steph Zhang, and Alex Immerman on the shift from "systems of record" to "systems of intelligence": https://t.co/2udG6l6SSx

[Harshil Mathur @harshilmathur]

Companies that are just using AI to replace people are underestimating the power of both.

[Akhil Singh @akhil_bits]

Jaya has cracked the code on generating deep insights on enterprise AI
Slowly becoming Naval of enterprise AI(long form)

[Jaya Gupta @JayaGup10]

https://t.co/5vDrndcmFn