Jon Bishop (@igrowsaas) header
Jon Bishop (@igrowsaas) profile picture

Building https://teapawt.net/t.co/GfP7Fc8G85 Ex @heap (acq’d), @periscopedata (acq’d), @joinhomebase

Marina Del Rey
Joined August 2020
This is a ridiculous stat in a ridiculous story: "The marginal cost of running an agent, had collapsed to, essentially, the cost of electricity." The marginal cost of a coding agent is not even remotely close to "the cost of electricity." These agents are absurdly expensive to use and run. Why do you think AI labs are banning people from having multiple $200 subscriptions? Because those subscriptions are heavily, heavily discounted to drive demand. Why did labs stop folks from using their subscription costs in OpenClaw? The OpenClaw guy had five max subs and was losing 20k a month building and running his amazing project (because he was retired and had the money to set in fire) before AI labs banned this practice of having multiple subs. In case you just missed it: Because these agents are expensive as hell to run. The cost of running coding agents daily on eight hour shifts is thousands of dollars a month at API pricing and that is subsidized too. My team regularly burns anywhere from 4K-8K a month across three people using the latest and greatest for an AI driven building workflow. That's not even agents running 24x7 "making money while you sleep" which is utter and total nonsense. This is one of the most spectacularly unprofitable businesses in history so far. People talking about the end of all work because this stuff runs for "pennies" cannot do even the most basic math. New NVIDIA chips don't even break even for data centers for like 24-36 months and they are basically obsolete by then. That doesn't count power and cooling and people to run it all. Imagine if your car was basically worth zero after three years? I'm so sick of these idiotic Population Bomb level stories about the end of all work and running agents for pennies. It's a mass delusion for people who can't be bothered to bust out a calculator on their phone for five seconds.
$50B of Indian IT services market value was eroded in the last 30 days. The Citrini article predicts it will collapse even more. Niftya IT index: -15% Wipro: -25% Infosys: -25% TCS: -17% Cognizant: -24% HCL: -17% Accenture: -25% Capgemini: -30% LTI Mindtree: -25% TechMahindra: -18% Mphasis: -20% Palantir claims it can compress complex SAP ERP migrations (ECC to S4) from years to 2 weeks. GCCs (companies owning their own offshore IT departments in India) with Claude Cowork are far more ecomical than multi year IT services contracts. I do think the 18% rupee collapse is exaggerated though. The IT services business model absolutely breaks at the current capability of AI tooling, and its ~10% of Indian GDP.
269
589
114
4,679
1,200,173
You hear about a startup that’s 9/9/6 but they’re like including the 2 hours everyone hits the gym and the Sunday ping pong tourney That’s not grinding bro ur just 22 and didn’t get a college experience
37
48
26
3,838
237,054
i think one of the next-decade health trends = detoxing and repairing hidden inflammation from mold + environmental toxins. we spend ~90% of life indoors & ~50% of buildings have water damage.
10
5
68
6,513
How do we measure AI fluency at Zapier? Here are some role-by-role examples 🧵
We’re setting a new standard at Zapier. 100% of new hires must be fluent in AI.
83
251
68
2,256
722,019
Guys I hate to rain on this parade but json prompting isn’t better. This post doesn’t even try to provide evidence that it’s better, it’s just hype. It physically pains me that this is getting so much traction - I’ve actually done experiments on this and markdown or xml is better - “Models are trained on json” -> yes they’re also trained on a massive amount of plain text, markdown, etc - JSON isn’t token efficient and creates tons of noise/attention load with whitespace, escaping, and keeping track of closing characters - JSON puts the model in a “I’m reading/outputting code” part of the distribution, not always what you want
Holy sh*t… nobody’s talking about JSON prompting but it’s the cheat code to unlock God-tier outputs from GPT-4, Claude, and Gemini. Here’s how it works (with copy-paste templates): 👇
168
179
75
2,177
692,563
Ok, here it is. My deep dive into the various patents that underpin how Google's AI Mode works. ipullrank.com/how-ai-mode-wo… The main thought I came away after this analysis is that search results are no longer deterministic in this environment. They are highly probabilistic due to the LLM-driven reasoning chains. This inherently means that the things we're doing don't always work anymore because our content is at the mercy of what an LLM "decides" at numerous stages that we can't see. For me, that definitively puts the "it's just SEO" argument to bed. It's going to require engineering, specifically Relevance Engineering, for us to navigate this and figure out how to ensure we make it through the reasoning cascade and reliably drive results. Oh, and I think Query Fan Out is the place to start, so I included a tool called Qforia. It's my first attempt at replicating 6 out of the 7 synthetic query styles that Google generates (yep, I found the Query Fan-out patent too). I'm looking forward to continued discussion. Well, the non "it's just SEO" parts of it, at least. I'm eager for us to collectively land on what it's going to take for us to thrive in this new environment. Naturally, I provided my initial thoughts on how to structure content based on what we've seen be effective and how the systems work, but we'll need to test things out as a community to build the new common practices.
14
43
9
164
13,104
building free tools got me 50K clicks in 6 months from all new search queries here's how: don't require users to create an account to try it bc 99% of users dont know that clearing cookies resets usage limits add the word "tools" and "generator" in the title tag, meta description and all throughout the body link to them everywhere throughout your site google search "reddit [your free tool]" and comment with the link to your tool add tools to directories go to formula bot dot com, connect your google analytics and google search console accounts to... 1) identify landing pages with the highest conversion rates 2) track which search queries lead to the best-performing pages 3) find queries where you're not ranking top for these high-performing pages import queries from step 3 into a new google ads campaign that directs users to the respective landing pages
1
17
2,365
Let’s say I need to analyze a competitor quickly. I start by looking at its referring domains. How many and which sites. Not only to know how powerful they’re in search. To understand how the product is perceived by the market. If they ran PR campaigns and what’s their angle. Then opinions I can find on user-generated content sites starting with Reddit (I avoid Google because it’s usually affiliate sites). What keywords they’re ranking for. Commercial? Informational? In-depth look at content that’s ranking to evaluate how good it is. Engagement to followers ratio on each platform they’re on. What content does well and why. If it’s B2C - who is tagging them on Instagram. If it’s B2B - who’s mentioning them on X. Check these profiles to understand their ICP and find patterns. Google autocomplete with brand name+modifiers to see what people ask. This alone will give me a decent idea.
4
3
40
3,200
I get hella requests for Twitter advice. Here are some easy tips for founders/leaders: - Niche down. Don’t start trying to be like Austen Allred or Naval. Start super narrow and focused in one area. - Content is king. Watch your favorite founders and study their content. Focus on improving your posts; keep iterating until you find a winning formula. Nail the hook, share a unique insight, and make it sufficiently provocative to get some comments. - Don’t @-mention, share links, or use hashtags. These will hurt your post’s reach. - Comments drive reach, so if you need to, beg, borrow and steal for them. - Frequency and consistency. Ideally 2-3x/week, but if that’s too hard just do 1x/week. Make a commitment and stick to that as a minimum. - Reach matters more than follower count. Make impressions your primary metric, not followers. - Most of your posts will be misses. You gotta be resilient to them, because this is a hits business and most posts aren’t going to be hits. - The best posts are either thoughtful, in-depth insights or hot takes from the toilet. Find what works for you. Honestly, I thought social was just a vanity thing when I started. I did it because it was fun and gave me some status while I was in-between companies. Once we started Maven, I realized my social presence was hugely valuable - hiring, recruiting instructors, engaging with our customer base, and building brand awareness.
7
8
5
109
39,909
Kevin Scott, CTO of Microsoft, framed the opportunity in AI this way : work on problems that used to be impossible, but are now really hard. My immediate reaction : what technical problems are now possible but still hard? What can computers achieve by themselves that two or three years ago would be intractable? But, this point isn’t purely about technical innovation. What business problems are newly solvable? Nearly 10 years ago, I wrote a post about the minimum viable average contract value to justify a sales team. Most companies typically hire account executives & development reps at $15k or greater in contract value because the economics of lower ACVs aren’t viable for most. Imagine a product offerd at a $10k contract value sold by an AE. The on-target earnings of the salesperson might be $120k. To break even, this team would need to book 12 contracts per year or one per month. But most sales leaders operate their GTM teams with a 4-5x ratio between sales costs & quota. In this case, this team would need to book $480k or 48 contracts per year - 1 per week. For most products, this is extremely difficult, bordering on impossible. AI-based sales development changes the economics of this calculation. If AI can reduce the overall cost of sales, the minimum viable ACV drops. Assume an account executive can close two transactions per month without AI but AI improves productivity by 50% (similar to coding productivity gains from copilots.) ACVs remain the same. The team operates at a 3x ratio of on-target earnings (OTE) to quota. It’s tight, but the economics work for a sales team to engage with a $10k ACV product. There may be a compounding effect. Assisted sales from freemium/PLG typically convert at 15% compared to unassisted sales at around 2-4%, a 5x+ improvement in yield. This hypothetical demonstrates a GTM motion that used to be impossible but is now really viable (& hard). Executed properly, this strategy could become a competitive differentiator in a market segment, enough to win disproportionate market share. AI has the potential to change the unit economics of a business, expanding addressable segments with better efficiency. tomtunguz.com/the-marginal-t…
5
10
1
96
17,011
Over the last year, we've sponsored 17 different email newsletters. One thing I've learned is that the success rate/response varies dramatically. Here are the three that have performed best for us: - @GrowthTactics by @NealOGrady - @mkt1cap by @emilykramer - @thegrowthmemo by @Kevin_Indig A few other learnings so far: - Higher Price ≠ Success — We've gone out on a limb and sponsored a handful of more expensive placements, and I've always found myself a little disappointed. If anything, I would say that the lower the placement cost, the better it has generally performed for us. - Don't Go Too Broad — We've tried sponsoring a few very general newsletters (eg "AI"). And in some cases, they actually drove a ton of demos, but the quality of the demos was a lot lower. Folks were curious but weren't actually in our ICP. And I wasted a ton of time demoing as a result. - Target Different ICPs with Different Placements — For example, the Growth Memo serves a very different ICP than MKT1, and you'll need to tailor the copy to the ICP you are targeting with that placement. For the Growth Memo, I might pitch our internal linking tool very specifically, given that the audience caters to harder-core SEOs. In contrast, I might pitch ourselves more broadly as an all-in-one tool for content marketing and SEO if the newsletter serves marketers/growth folks more broadly. - Double Down on Winners — When you find a placement that works, do it again. But not right away; maybe wait a month or a few. It's better to double down on the placements you know worked previously versus risking it with another unknown placement. But you do need to burn some capital on unknown placements to find those winners ultimately. - Dedicated Landing Pages Didn't Work as Well — In the beginning, we pushed everyone to our homepage. But then we experimented with dedicated landing pages custom to that placement. We actually found that pushing folks into the homepage performed better. 🤔 You might see us in a few more placements soon! 👀
5
5
21
3,155
Jon Bishop retweeted
Replying to @ommakes
it’s more … 1. Users are often more than one step removed from buyer — lack direct access/frequent communication 2. Try and sell in and fail to position OR understand buyer value — easily sidelined 3. Don’t want exec team knowing they are leveraging something not approved — didn’t ask for permission
2
1
10
483