Vishveshwar Jatain

SO: Brands waste ~$8.1 Billion of their advertising spend every year on low-quality YouTube placements like AI slop, MFA, kids' content, etc. That’s just one of the findings in the new “The Cost of Default” report published by Filament. When I first started working on this report, Erik and Scott, co-founders of Filament, pointed me to a database they had built to track the web of brands, agencies, and campaigns they run on YouTube. This database tracks nearly 700 billion public YouTube views across 230,125 ads from 9,576 advertisers, assembled through a collection of sources. Over the course of digging up this information, I tried connecting with and speaking to a few agency leaders. I explained what Filament does, and in a lot of cases, what I heard back was: “We’re good, thanks.” “This isn’t a problem for us.” “We’ve got something for this.” …but then you look at the data and realize that they don’t have something for it. If this was a solved problem, then we'd look at the numbers and say, “Okay, this is not a problem at all!” But this is what the numbers show instead: 1 / Upwards of $8 Billion worth of spend pointed towards YouTube is flushed down the drain on low-quality placements. 2 / Second, 45% of brands don’t own the accounts where the ads run, so unless they’re told, they never learn about this waste. 3 / Third, the waste is entirely preventable with the right countermeasures, part of which includes using human verification of inventory quality. Thanks to @vlotero, @elevin11, @kolibriefie, @adamheimlich, and @acfou, who generously offered their perspective and platform to help us get the word out about this report. Grab your copy: wearefilament.com?utm_source… This was many months of collaborative work, so a like, comment, and repost is much appreciated! -- @TimRowe_SOS @EricTilbury_RTB @RioLongacre @trishlaostwal @KendraEBarnett @notmiramurati @pranavpiyush @AdamSinger @bmorrissey @JonathanMoffie @tkawaja @BeelerTech @anthonykatsur @JamesBorow @JamesHercher @rishad @Mark_Penn @aripap @Beet_TV @JeremyBloomHere @billwise @davemorgannyc @mattprohaska @jason_kint @NairiHourdaj @kippbodnar @larakiara @PPCKirk @ericfranchi @JoeZawadzki @joe_zappa @oliviaakory @vinny_rinaldi3 @AdtechGod @JeromySonne @verklin @PatrickCoffee @bristei @xpangler @SarahSluis @SarahPerezTC @jason_koebler @CaseyNewton @inspiredtaste @EthanZ

Please do not truncate tomorrow to “tmrw”. I understand tomorrow perfectly well. My brain has to work extra hard to process your cool slang.

Choy recalls an estimate that AI agents could mediate $3–5T in consumer commerce by 2030. That’s not ad revenue; ads would be a smaller slice.

At this point, I could easily automate most of the "work" I do. This is a strange thing to think out loud, but it's true. Everything from product development, content creation, agency projects, creator deals, etc. 80% of the execution on all this stuff is made up of a specific set of actions and routines. I don't know if this is the case with everyone, but for me, AI never really delivered on the promise of "it'll free up your time", but instead opened up my vision to a much bigger superset of things that I simply could not do before. I'm starting to think that the only way to intelligently reclaim that time is to actually, in earnest, try to make as much of your own self redundant as you possibly can.

Prospecting used to mean wading through 10 different tabs: Apollo, Hunter, a verifier, and a spreadsheet full of no-intent numbered lead list. Now it's one prompt in the agent I already use. Find the people, verify the emails, save it to Sheets. $0.012 a company to find, $0.0032 an email to verify. No new subscription.

Who’s managing product at @X after @nikitabier left? This giant hand popped up to nudge me towards the new “promote” button. Thanks for the jump scare. I’ve deleted your app and exorcised my phone. (cc: @benjitaylor, wish I had agentation here to delete tf out of this)

Me after building ChatGPT / Claude connectors for my product, finishing an industry report for a client, submitting a grant application for a startup I advise, and participating in 3 creator campaigns on X—all in one day.

Time to get my preciou... er, I mean, my agentic cat herding ring. I would love to experience life again outside of the 100 terminal tabs I shuffle between all day.

I just spun up a Live 3 Duplex agent called “Dex”, and as someone with a fine-tuned ear for AI / IVR bots, I gotta say this is a big leap 🤯 There are two big problems with real-time voice AI: Latency and context drop during interruptions. If Dex called me on my phone, I would have a hard time telling it apart from a human.

Of course the big time influencers think that micro influencers doing brand deals on X is scummy. Mate, it’s not for you. And in case it doesn’t occur to you, this is almost a textbook application of free market economic principles. A product needs visibility, there happens to be someone who has an audience, and a deal is made. If this makes you feel a special kind of ick, maybe you could check your privilege.

Current life stage

Unfortunately, as an inductee of the Nvidia Inception program, I’m not at liberty to comment on current events surrounding certain key company executives and/or reports of alleged foreign interference. I can, however, state with freedom that I’ve never understood the appeal of leather apparel and accessories, including but not limited to jackets. Note that is not in any manner, shape or form, a reflection on any person (or persons) who choose to wear such apparel, but simply a matter of personal taste. This concludes my statement.

Lack of speed / responsiveness is the number one killer of purchase intent. The urgency with which a buyer feels the need to have their problem solved is greatest right after they’ve filled up the contact form on your site. And yet, almost every B2B startup squanders the golden opportunity to meet the buyer when their intent is at its peak. I think this is the first use case I’ll test with my @ElevenLabs grant credits. In case you haven’t been keeping up, voice AI has gotten really, really good. And ElevenLabs is the uncontested category leader. You can now hook these agents up to real-time search APIs, so they’ll show up knowing who they are speaking with. Imagine a voice AI agent giving your prospect a callback a few seconds after the form fill. It can give them useful info, qualify them at their own pace, and turn a typical ball drop into a deal in your pipeline. Of course, if you really want to push this idea, you can have Snitcher pass a company ID via a webhook to an agent that can research the buying committee and then dial parallel calls to everyone. What’s the worst that happens? Someone gets annoyed and hangs up. How’s that any different from a cold email being moved into the trash folder? New technology always provides these tiny windows of opportunity where you can swing home runs before the novelty wears off and it’s ruined for good. You can call it voice engineering if you like.

To the moon 😅

Always funny to see mfs with 20K+ followers on LinkedIn that they sourced through aggressive comment gating flop around with 124 followers here on X. What’s up Mr. Like-And-Comment, why don’t you share your 700-step n8n framework with us???

Best deal of all time right here.