seshu bonam

Graphic.so is Now Open Sourced! github.com/ai-graphic/Gra… Hit that github ⭐ and stay updated for learning easiest ai automation workflows. built by the most talented Dev i've ever worked with @dothebit 🙌🏻

@ApplyWiseAi @claudeai damn 🙌

@seshubon @claudeai It was a 8 hours session on to make some backend changes with Opus. I got a lot done ;) Update I went over a bit .

@claudeai i was able to accept the extra credit twice so ended up getting 400 instead of 200. one time from mobile and one time from desktop . i would stay quite but i love you guys!.. pls confirm if i can keep or itll get pulled. ;)

@seshubon Fix incoming

is a browser app like chrome with web pages showings tuff thats not vetted by apple against TOS. don't get it why a native app can't extend functionality with web view

@law_ninja interesting and so true

Every person who has tried to sell software to a small Indian law firm has heard this: "Bhai, send me the proposal. I'll look at it." You follow up. "Still reviewing." You follow up again. Nothing. Three months pass. The deal is dead. You cut the price. Same response. You add features. Same response. You offer a free trial. They log in once and disappear. The problem is not your pricing. The problem is not your product. The problem is you are selling the wrong thing. Small Indian businesses do not buy software. They hire people. This is not a behavioral quirk. It is how trust and accountability work in this market. Think about what happened when Indian courts started going digital. E-filing became mandatory. Case status went online. Court orders became downloadable. The portals existed. They were not complicated. Any lawyer with a smartphone and an internet connection could have figured it out in an afternoon. Nobody figured it out. Instead, thousands of e-filing operators and court typists set up shop near every district court complex in India. The same typists who used to type petitions on typewriters now started filing cases online for lawyers. Charging Rs 200 to Rs 500 per filing. Just to use portals the lawyer could have accessed themselves. These operators now handle everything from e-filing to downloading court orders to checking case status. Many of them charge monthly retainers from 15 to 20 lawyers each. They are the person the lawyer calls when anything digital does not work. The lawyers did not want the portal. They wanted a person who would handle it and be answerable when a filing deadline was missed. Same story with GST. ClearTax built software. Tally added modules. The tools existed. Nobody learned. Instead, 3 lakh GST consultants emerged across India. Charging Rs 500 to Rs 2,000 per month per client. Just to file returns using tools the client could have accessed themselves. Because the person you hire is accountable. The app is not. Now apply this to AI. You build an AI workflow system for a 5-person law firm. Client intake automation. Hearing date reminders. Document drafting. Legal research summaries. It works beautifully. You try to sell it as a SaaS product for Rs 2,000 a month. They will not buy it. Not because Rs 2,000 is too much. They pay their munshi Rs 12,000 a month. They pay for their Manupatra subscription. They pay the typist outside court for e-filing. They will not buy it because they do not trust a subscription to an unknown product. Nobody to call when something breaks. Nobody accountable when the reminder does not go out before the limitation date. The way to sell AI to small Indian law firms is not to sell software. It is to sell yourself as the person who builds it, runs it, and fixes it. Rs 15,000 to 20,000 to build and set up. Rs 2,000 a month to maintain and be available. Same pricing as their e-filing operator. Same mental model. You are not a product. You are a person they can call. And here is where the distribution insight gets interesting. Think about who already walks into a lawyer's chamber every month. The legal book supplier. The local distributor who drops off bare acts and commentaries. These people have been visiting the same 200 to 300 lawyers for years. They know which advocate sits in which chamber. They know their practice area, their court, their temperament. The lawyer already trusts this person. Already buys from them. Already opens the door when they knock. Now imagine that book supplier says: "Sir, along with your commentary subscription, I can also set up an AI system for your office. Hearing date reminders, draft notices, client follow-ups. Rs 15,000 setup, Rs 2,000 a month. I will handle everything." The conversion rate on that pitch is not 2 percent. It is 40 to 60 percent. Because the trust already exists. The relationship already exists. The regular access to the chamber already exists. The same applies to the stamp vendor and the notary agent who sees the same set of lawyers week after week. Or the munshi inside the firm who handles all the filings and would be the one actually operating any new system. This is how India adopts new technology. Not through app stores and LinkedIn ads. Through trusted intermediaries who bundle the new thing with an existing relationship. The person building AI deployment businesses for Indian law firms who figures this out first will not be selling to one advocate at a time. They will be training legal book suppliers and e-filing operators to offer this as a service to their existing clients. That is a distribution model. Not a product. Not a marketing funnel. The SaaS model assumes the buyer wants to learn and self-serve. The India model says: find the person the buyer already trusts. Work through them. One is selling software. The other is understanding how India actually works. Know anyone who has done this yet for legal software or AI in India?

@HvacLeadsTech @scaling_shields can't dm, ping me

@seshubon @scaling_shields hey, can you send me the docs or the link that james sent you ?

i put my entire $480K/year cold email business into ONE google doc 200+ pages across 10 guides - the complete cold email blueprint - the offer formula that books calls - untapped lead sources (not apollo) - the scripts vault (every script ive ever used) - 2026 spam filter survival guide - ecom client acquisition playbook - b2b agency client acquisition playbook - the 2-line email that booked 103 calls in 12 days this is the doc i wouldve KILLED for before scaling to over $45k/mo like + comment "VAULT" and ill send it over (must follow + RT for priority access)

@brenvaapp @bcherny yes

@seshubon @bcherny I don't even see it in my plan... is it under settings-->usage? Thanks!

Starting tomorrow at 12pm PT, Claude subscriptions will no longer cover usage on third-party tools like OpenClaw. You can still use these tools with your Claude login via extra usage bundles (now available at a discount), or with a Claude API key.

@bcherny pls check again x.com/i/status/20402…

@bcherny 🙀Double credit! clicked once on desktop, got credited $200 banner shows up on mobile, clicked anyway credit is now $400 please check again @bcherny @AnthropicAI @claudeai

@seshubon fixed! try again now

@bcherny yay🥳 thanks @bcherny @AnthropicAI

@bcherny works, but ill keep post for posterity . my fav replied 🥳

what if!

surprised -> failed

@ant0ni0_r0mer0 Ads

Meta’s new algorithm rewards creative diversity. If you’re not increasing output, you’re losing. So I built a workflow that turns one brief into dozens of ads in minutes. The workflow analyze top competitors, extract what converts, define the angle, generate production-ready creatives, all in one flow. Perfect for brands that want to bring production in-house and cut agency costs, and for agencies that need massive creative volume without hiring. One setup. Infinite ads. Want it ? Comment “ADS” and I’ll DM you the workflow (make sure we’re connected).

for a quick sanity check /btw is so useful and sometimes for brainstorming new directions without context bloat.

@spect3ral MCP

I connected Claude directly to my Meta ad account. Meta ads MCP full guide 👇 No CSV exports. No copy-pasting into ChatGPT. No pivot tables. I just ask questions in plain English and get live answers from my actual data. The technology is called MCP (Model Context Protocol). It lets Claude read your live Meta ad data and take actions — pause ads, move budgets, flag problems — all from a single conversation. Setup took 10 minutes. No coding. Here's what I can do now: 1/ Morning performance check "Show me yesterday's performance. Flag anything where CPA is 20% above my 7-day average." Done in 30 seconds. Used to take 20 minutes in Ads Manager. 2/ Creative fatigue scan "Which ads have frequency above 3 and CTR dropped 20% in the last 14 days?" Claude pulls the data live. No export needed. 3/ Budget reallocation "Move $200/day from my worst ad sets to the top 3 by ROAS." It makes the changes. You confirm before anything goes live. 4/ Audience breakdown "Break down my best campaigns by age, gender, and placement. Where am I leaving money on the table?" Instant. No custom reports. No spreadsheets. 5/ Campaign creation "Create a new campaign targeting women 25-44 in the US, interested in skincare, $50/day budget." Launches in PAUSED status. Nothing spends until you approve. How to set it up (4 steps): → Get Claude Pro ($20/month) → Pick a managed connector (Pipeboard, Adzviser, or Madgicx) → Connect your Meta Business Manager through OAuth → Paste the connector URL in Claude's integration settings That's it. Test with: "List all my active Meta ad accounts." If you see your accounts, you're live. One critical warning: Never use browser extensions or scrapers that simulate clicks in Ads Manager. Meta will flag and ban your account. Only use connectors that go through Meta's official Marketing API. Pipeboard, Adzviser, Madgicx — these are safe. Total cost: → Claude Pro: $20/month → Connector: free tier available → Meta Marketing API: free Under $20/month for most accounts. I manage my entire Meta ads operation from a single chat window now. Comment "MCP" and I'll send you the full step-by-step setup guide

@denohawari LLM

we helped a SaaS company rank #1 in ChatGPT • $1.8M+ total revenue • 8,200%+ traffic growth • $35K+ monthly SEO traffic value all powered by our LLM SEO framework. most companies have no idea this is even happening: buyers are rapidly moving to LLM-driven discovery tools like ChatGPT, Claude, and Gemini if your brand isn’t showing up there, your competitors are straight up stealing your traffic and conversions. I broke down our entire methodology into a 800-word guide: • how we structure content LLMs consistently surface • data from the 8,200% growth case study • the step-by-step system that drove the $1.8M+ revenue lift • the AI-first keyword discovery framework we use across LLM platforms • the on-page signals that unlock visibility in ChatGPT, Claude, and Gemini this SaaS company's entire organic engine was built through this new AI-first SEO method now you can just steal it want the full breakdown? 1. like + follow 2. comment “LLM” and I’ll send it to you