Anurag Singhal

Founder of Quattr Inc, the AI Search Visibility Platform. We unify your data, content & linking strategy to get you cited in AI answers, not just ranked

[Anurag Singhal @anurag_singhal]

I am now wondering if @salesforce is fighting back #SaaSpocalypse by copying @bendingspoons playbook right here in silicon valley?

[Braze @Braze]

Braze is taking the next step on our journey to transform customer engagement by agreeing to be acquired by @salesforce. We’re excited to join forces to help marketers around the world build deeper relationships with their customers. See here for more: https://t.co/C1b1vjTsC7

[dharmesh @dharmesh]

Woo hoo! I passed the 600,000 follower mark here on Ex-Twitter.

To celebrate, I'm going to go read some tweets and follow some people.

Thank you for your support.

[Anurag Singhal @anurag_singhal]

@lilyraynyc Also did you notice that the product names in these listcles launch a modal with a chatgpt generated view of the brand rendered with images at top vs sending the user to the brand's home page.

[Anurag Singhal @anurag_singhal]

I fed three takes I read here on X on this cool demo to Fable: one said "Nobel-worthy," another said "Turing Award + proof of RSI," third called it "slop bait."

Got pushed back on all three.

Fable's take: great demo of an idea from 1992, two real properties (exact + differentiable, runs on GPU), nothing built with it yet, just as @_chenglou calls it a toy.

Nice to see AI that argues with both the hype and the haters and helps me calibrate viral X posts in real time :)
Still learning and parsing this intriguing viral post, here to learn.

[Cheng Lou @_chenglou]

My dear friends, I’d like to show you my latest toy project

Introducing Handwired: turn your regular JavaScript code into a neural net!
- no training
- exact outputs vs the JS program
- fully differentiable

All runnable in a browser, without GPU

[Anurag Singhal @anurag_singhal]

Great points. Here is my 0.0002. I think human attention to evaluate, learn and use software is already overwhelmed today and when software costs a hundredth of today's cost, it gets tested further. So no Jevons there but...the real breakthrough comes from agents using software when they'd rather not build. That has bottlenecks today and removing that via MCPs. Evals is the unlock.

You needed a GTM strategy for SaaS.
You need a Got To Agent strategy to be able to sell your software for that $10 an agent might pay in $0.001 increments.

[Anurag Singhal @anurag_singhal]

@jackson_llk @quattr_inc MCP covers gsc, ga4, google add, ai visibility, logs and more

[Anurag Singhal @anurag_singhal]

@elshayib_ It will be Fable 6 (my bet)

[Anurag Singhal @anurag_singhal]

@IndependentEco They are the only lab that thinks architecture-first. And ……. architecture matters. That is all. All.

[Anurag Singhal @anurag_singhal]

I have worked on clincial trial design and management software used by big pharma. The alignment incentive concern for passing a phase 3 trial is no different than a pharmco executive incentives. Tao offers no real solution to the alignment problem that already exists with humans in control but advocates for not letting AI accelerate the research at the same level. In his example, the benefit of AI far outweigh the risks when weighed against already prevalent risks.

[Anurag Singhal @anurag_singhal]

Zero click: they never visit your website.
Zero opportunity: they never consider any business.

ChatGPT can attempt the job before showing alternatives. My tests found a gap between starting and finishing.

Progress, or premature substitution? https://t.co/dNGuKgbwbk

[Anurag Singhal @anurag_singhal]

@satyanadella Bodes a decacorn that positions itself as “Zero Trust Super Intelligence” ZTSI or “Contained Access Super Ingelligence Edge” CASIE

[Anurag Singhal @anurag_singhal]

@mxstbr Good to hear you are also taking a hard look across the seven-step process. Thanks for these updates, please keep them coming.

[Anurag Singhal @anurag_singhal]

Made-up forecasts are theater to Jono’s point, but imo real cmo bravery is investing in systems that help make an honest guess, align people and letting everyone check it later using the same investment. it will also need continuous investment to keep the system trustworthy.

That said, even most “enterprise” tools suck at this task today given the growing challenges of ai search. I built this in house at realtor dot com and it was a lot of work but it paid off big.

[Chandu Thota @ChanduThota]

We are making Markdown files work natively across Google Drive and Docs.

You can now preview .md files in Drive, open them in Docs, edit them, comment on them, and collaborate without converting the file into a Google Doc!

[Anurag Singhal @anurag_singhal]

@paulg This seems like a good lens. Specific examples from your experience that you might be willing to share to drive this home?

Also, I assume the post is about inspiring generational startups only

[Anurag Singhal @anurag_singhal]

@mxstbr Thank you!

[Anurag Singhal @anurag_singhal]

@ZryMiller I guess the next four weeks is about 28 codex updates up against a Fable 5.5 model launch

[Anurag Singhal @anurag_singhal]

@vigyso Will sign in with chatgpt also bring pre-authenticated MCPs from user’s chatgpt connectors to activate supported connectors in our apps?

[Anurag Singhal @anurag_singhal]

Can you opine on pricing and architecture evolution from here for enterprise customers?

ChatGPT signin points to an upcoming pricing model shift. How does business continuity work when customers run out of tokens mid week. (rn they at times ping to ask if we can use our mcp ourselves to do their work 🙂for them as they are out of token)

arch wise, how do we bet to build inside claude if users hardly have enough tokens to operate with vs we cover tokens inside own harness with annual contracts? same for permissions, easier to provision harness inside own app vs customer’s IT approving a connector. Any plans to allow vendors to cover tokens for sessions using our apps inside Claude?