Fetch Saved Posts
GEThttps://api.tweetsmash.com/v1/bookmarks
Retrieve all the bookmarks that you've created. You can use the query parameters to filter and paginate results.
Header Parameters
Query Parameters
Response
{
"status": true,
"data": [
{
"post_id": "1938682607145034084",
"tweet_details": {
"text": "friendly reminder: productizing AI agents is the biggest zero-to-one play on the internet right now.\n\nstep 1\n\nstart with the workflow. open someone's \"how-to\" loom, watch for any section where the mouse repeats itself for more than 30 seconds. that is your extraction point. i call this step looking for drudge. \n\nstep 2\n\nadd the context. scrape the internal FAQ, the tribal acronyms, the edge-case screenshots that a general LLM will never see. that proprietary mess formatted into plain text or json is your first moat. sometimes you need an audience for this. it helps. usually i create an audience/community to start.\n\nstep 3\n\nwire up an agent on an LLM or agent platform: claude, gpt-4o, gemini, lindy, string dot com. keep the prompt brutal-simple: \"given this context and that input, return this output.\" test until the agent's average response time beats the human baseline by 10×. keep refining prompt, it will get longer, longer and more valuable. never share that prompt with anyone.\n\nstep 4\n\nwrap the agent in the thinnest UI possible. one drop zone, one action button, one log window that says \"done.\" hide the settings; hide the model; hide the magic. outcome is the only feature. make a brand people fall in love with.\n\nstep 5\n\nprice the saved hours. take the median hourly cost of the person who used to run the workflow, multiply by the time you shave off each week, and anchor the subscription somewhere south of pain but north of hobby.\n\nstep 6\n\nship video proof before a landing page. record old way vs. new way in a split-screen clip; caption it; post it where the operators hang out. every \"how did you do that?\" reply is a warm lead. you can use platforms like lindy ai etc to do cold dm/email automation.\n\nstep 7\n\nloop usage back into the context file. new ticket? new edge case? append it. your agent improves without code, and the context corpus becomes asset #2.\n\nstep 8\n\nlayer distribution inside the product. add a quiet \"powered by\" footer on every output, or trigger a shareable report by default. users spread the gospel while you sleep.\n\nstep 9\n\nwhen churn stabilizes and support tickets hit near-zero, bolt on adjacent micro-agents each with its own promise, its own subscription tier, no extra complexity in the UI.\n\nstep 10\n\nignore vanity metrics. screenshots of ARR can be faked (and are constantly). My favorite was watching someone screenshot test payments. Literally fake money lol. \n\nAnyways, time returned to a paying customer cannot be faked. compound that saved time across niches until you own a stack of invisible employees.\n\nyou own a portfolio of little products that save people time and you get paid for it. the market for this is in the billions.\n\ndo this on repeat and you won't need funding, permission, or glossy charts.\n\nyou will be free and be creating impact.\n\nthis is probably one of the biggest opportunities out there.",
"link": "https://twitter.com/gregisenberg/status/1938682607145034084",
"posted_at": "2025-06-27T19:35:27+00:00",
"attachments": [],
"quoted_tweet": null,
"meta": {
"tweet_reply_to_tweet_id": null,
"tweet_reply_to_user_id": null,
"tweet_conversation_id": "1938682607145034084",
"tweet_is_thread": false,
"is_sub_tweet_of_thread": false,
"non_real_time_engagement_metrics": {
"quote_count": 17,
"reply_count": 86,
"retweet_count": 172,
"bookmark_count": 5352,
"favorite_count": 2299
}
}
},
"author_id": "14642331",
"author_username": "gregisenberg",
"author_details": {
"name": "GREG ISENBERG",
"username": "gregisenberg",
"profile_image": "https://pbs.twimg.com/profile_images/1577116785656139776/5mi0qgTz_normal.jpg"
},
"tags": null,
"is_read": false,
"is_archived": false,
"imported_at": "2025-06-29T12:55:13.596508",
"sort_index": 1836198123519712800
}
],
"meta": {
"next_cursor": "MTc1OTYxNw==",
"limit": 1,
"total_count": 396
}
}