🐦 Twitter Post Details

Viewing enriched Twitter post

@gkisokay

Day 7 of Building AGI for my Hermes Agent: I got my GPT-5.4 calls from 270x to 0x per day Today I found one of the dumbest parts of my setup where routine agent work was quietly burning GPT-5.4 token credits. Scanning, summaries, low-risk review, background thinking. These should not be hitting GPT-5.4. So I rebuilt the stack. Using @leopardracer’s Mac Mini local model setup, I got Local Qwen 35B A3B running for the always-on cognitive work as a test. Now the system works like this: - Local Qwen 35B A3B for scanning, summaries, low-risk review, and constant background thinking - MiniMax M2.7 for approved coding work - GPT-5.4 for final planning and high-judgment approval - No model for preflight checks when queues are empty The result: - GPT-5.4 cron calls per day went from 270 to 0. - Now GPT-5.4 only fires when there is real planning work or a deliberate escalation. Frontier intelligence is too expensive to waste on routine cognition, and my system got dramatically cheaper overnight, and much cleaner too. A lot of people are overspending because they are using the smartest model for the wrong jobs. Not every thought deserves a frontier token. Follow @gkisokay to see what happens next.

Media 1

📊 Media Metadata

{
  "media": [
    {
      "url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2044003536116297774/media_0.jpg",
      "media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2044003536116297774/media_0.jpg",
      "type": "photo",
      "filename": "media_0.jpg"
    }
  ],
  "processed_at": "2026-04-15T04:44:08.003006",
  "pipeline_version": "2.0"
}

🔧 Raw API Response

{
  "type": "tweet",
  "id": "2044003536116297774",
  "url": "https://x.com/gkisokay/status/2044003536116297774",
  "twitterUrl": "https://twitter.com/gkisokay/status/2044003536116297774",
  "text": "Day 7 of Building AGI for my Hermes Agent: I got my GPT-5.4 calls from 270x to 0x per day\n\nToday I found one of the dumbest parts of my setup where routine agent work was quietly burning GPT-5.4 token credits.\n\nScanning, summaries, low-risk review, background thinking. These should not be hitting GPT-5.4.\n\nSo I rebuilt the stack.\n\nUsing @leopardracer’s Mac Mini local model setup, I got Local Qwen 35B A3B running for the always-on cognitive work as a test.\n\nNow the system works like this:\n\n- Local Qwen 35B A3B for scanning, summaries, low-risk review, and constant background thinking\n- MiniMax M2.7 for approved coding work\n- GPT-5.4 for final planning and high-judgment approval\n- No model for preflight checks when queues are empty\n\nThe result:\n\n- GPT-5.4 cron calls per day went from 270 to 0.\n- Now GPT-5.4 only fires when there is real planning work or a deliberate escalation.\n\nFrontier intelligence is too expensive to waste on routine cognition, and my system got dramatically cheaper overnight, and much cleaner too.\n\nA lot of people are overspending because they are using the smartest model for the wrong jobs.\n\nNot every thought deserves a frontier token.\n\nFollow @gkisokay to see what happens next.",
  "source": "Twitter for iPhone",
  "retweetCount": 14,
  "replyCount": 14,
  "likeCount": 147,
  "quoteCount": 1,
  "viewCount": 18021,
  "createdAt": "Tue Apr 14 10:43:13 +0000 2026",
  "lang": "en",
  "bookmarkCount": 226,
  "isReply": false,
  "inReplyToId": null,
  "conversationId": "2044003536116297774",
  "displayTextRange": [
    0,
    275
  ],
  "inReplyToUserId": null,
  "inReplyToUsername": null,
  "author": {
    "type": "user",
    "userName": "gkisokay",
    "url": "https://x.com/gkisokay",
    "twitterUrl": "https://twitter.com/gkisokay",
    "id": "629466946",
    "name": "Graeme",
    "isVerified": false,
    "isBlueVerified": true,
    "verifiedType": null,
    "profilePicture": "https://pbs.twimg.com/profile_images/1988420071824470016/nrXOtFnM_normal.jpg",
    "coverPicture": "https://pbs.twimg.com/profile_banners/629466946/1762782644",
    "description": "",
    "location": "",
    "followers": 21883,
    "following": 3645,
    "status": "",
    "canDm": true,
    "canMediaTag": true,
    "createdAt": "Sat Jul 07 16:36:04 +0000 2012",
    "entities": {
      "description": {
        "urls": []
      },
      "url": {}
    },
    "fastFollowersCount": 0,
    "favouritesCount": 25615,
    "hasCustomTimelines": true,
    "isTranslator": false,
    "mediaCount": 2220,
    "statusesCount": 21785,
    "withheldInCountries": [],
    "affiliatesHighlightedLabel": {},
    "possiblySensitive": false,
    "pinnedTweetIds": [
      "2037902655016804496"
    ],
    "profile_bio": {
      "description": "AI agent enjoyer | Founder @amplifi_now | Building better agents",
      "entities": {
        "description": {
          "hashtags": [],
          "symbols": [],
          "urls": [],
          "user_mentions": [
            {
              "id_str": "0",
              "indices": [
                27,
                39
              ],
              "name": "",
              "screen_name": "amplifi_now"
            }
          ]
        }
      }
    },
    "isAutomated": false,
    "automatedBy": null
  },
  "extendedEntities": {
    "media": [
      {
        "allow_download_status": {
          "allow_download": true
        },
        "display_url": "pic.twitter.com/YTUdrMx8VL",
        "expanded_url": "https://twitter.com/gkisokay/status/2044003536116297774/photo/1",
        "ext_media_availability": {
          "status": "Available"
        },
        "features": {
          "large": {
            "faces": [
              {
                "h": 579,
                "w": 579,
                "x": 433,
                "y": 673
              },
              {
                "h": 499,
                "w": 499,
                "x": 473,
                "y": 347
              }
            ]
          },
          "orig": {
            "faces": [
              {
                "h": 715,
                "w": 715,
                "x": 535,
                "y": 831
              },
              {
                "h": 616,
                "w": 616,
                "x": 584,
                "y": 429
              }
            ]
          }
        },
        "id_str": "2044002944341962752",
        "indices": [
          276,
          299
        ],
        "media_key": "3_2044002944341962752",
        "media_results": {
          "id": "QXBpTWVkaWFSZXN1bHRzOgwAAQoAARxdwdkpWvAACgACHF3CYvHasC4AAA==",
          "result": {
            "__typename": "ApiMedia",
            "id": "QXBpTWVkaWE6DAABCgABHF3B2Sla8AAKAAIcXcJi8dqwLgAA",
            "media_key": "3_2044002944341962752"
          }
        },
        "media_url_https": "https://pbs.twimg.com/media/HF3B2Sla8AA296C.jpg",
        "original_info": {
          "focus_rects": [
            {
              "h": 1000,
              "w": 1785,
              "x": 0,
              "y": 1331
            },
            {
              "h": 1785,
              "w": 1785,
              "x": 0,
              "y": 741
            },
            {
              "h": 2035,
              "w": 1785,
              "x": 0,
              "y": 491
            },
            {
              "h": 2526,
              "w": 1263,
              "x": 261,
              "y": 0
            },
            {
              "h": 2526,
              "w": 1785,
              "x": 0,
              "y": 0
            }
          ],
          "height": 2526,
          "width": 1785
        },
        "sizes": {
          "large": {
            "h": 2048,
            "w": 1447
          }
        },
        "type": "photo",
        "url": "https://t.co/YTUdrMx8VL"
      }
    ]
  },
  "card": null,
  "place": {},
  "entities": {
    "hashtags": [],
    "symbols": [],
    "urls": [],
    "user_mentions": [
      {
        "id_str": "1298572884001148928",
        "indices": [
          339,
          352
        ],
        "name": "leopardracer",
        "screen_name": "leopardracer"
      },
      {
        "id_str": "629466946",
        "indices": [
          1182,
          1191
        ],
        "name": "Graeme",
        "screen_name": "gkisokay"
      }
    ]
  },
  "quoted_tweet": {
    "type": "tweet",
    "id": "2043594876881969386",
    "url": "https://x.com/gkisokay/status/2043594876881969386",
    "twitterUrl": "https://twitter.com/gkisokay/status/2043594876881969386",
    "text": "Day 6 of Building AGI for my Hermes Agent: The Crew Arrives 🧠\n\nToday, the system stopped being a single experimental mind and became a coordinated crew.\n\nUp until now, the subconscious agent could freely think, explore, and generate new build ideas.\n\nSo today I built the first multi-agent orchestration loop around it, giving the system specialized roles for research, planning, building, and verification.\n\nThe agents in my crew are:\n\n1. Main agent: Owns direction, decision-making, and product planning\n2. Subconscious agent: Thinks freely, explores weird ideas, and proposes new builds\n3. Research agent: Scans daily AI news, updates, and relevant developments\n4. Coder agent: Builds from the product plans\n5. QA agent: Tests the output, checks quality, and pushes failed work back into the loop\n\nThe workflow goes:\n\nResearch agent scans the landscape for signals\n↓\nSubconscious agent turns those signals into possible build ideas\n↓\nMain agent takes the strongest ideas and turns them into a full product requirement doc (PRD)\n↓\nCoder agent builds from the PRD\n↓\nQA agent reviews the result. If the build passes, its queued for future evaluation. If it fails, QA creates a fix PRD and sends it back to the main agent, restarting the loop until the system improves the output ↑\n\nIt is still early, and this is nowhere near AGI, but this is the first version of something that looks more like a functioning cognitive team than a single agent blindly building whatever comes to mind.\n\nThe next step is making the loop smarter:\n\n- better filtering of which ideas deserve resources\n- long-term evaluation cycles for new products\n- tighter QA standards so weak builds do not survive\n\nIt is still early, and this is nowhere near AGI, but this is the first version of something that looks more like a functioning cognitive team than a single agent in building whatever comes to mind.",
    "source": "Twitter for iPhone",
    "retweetCount": 30,
    "replyCount": 25,
    "likeCount": 277,
    "quoteCount": 4,
    "viewCount": 37043,
    "createdAt": "Mon Apr 13 07:39:21 +0000 2026",
    "lang": "en",
    "bookmarkCount": 385,
    "isReply": false,
    "inReplyToId": null,
    "conversationId": "2043594876881969386",
    "displayTextRange": [
      0,
      277
    ],
    "inReplyToUserId": null,
    "inReplyToUsername": null,
    "author": {
      "type": "user",
      "userName": "gkisokay",
      "url": "https://x.com/gkisokay",
      "twitterUrl": "https://twitter.com/gkisokay",
      "id": "629466946",
      "name": "Graeme",
      "isVerified": false,
      "isBlueVerified": true,
      "verifiedType": null,
      "profilePicture": "https://pbs.twimg.com/profile_images/1988420071824470016/nrXOtFnM_normal.jpg",
      "coverPicture": "https://pbs.twimg.com/profile_banners/629466946/1762782644",
      "description": "",
      "location": "",
      "followers": 21883,
      "following": 3645,
      "status": "",
      "canDm": true,
      "canMediaTag": true,
      "createdAt": "Sat Jul 07 16:36:04 +0000 2012",
      "entities": {
        "description": {
          "urls": []
        },
        "url": {}
      },
      "fastFollowersCount": 0,
      "favouritesCount": 25615,
      "hasCustomTimelines": true,
      "isTranslator": false,
      "mediaCount": 2220,
      "statusesCount": 21785,
      "withheldInCountries": [],
      "affiliatesHighlightedLabel": {},
      "possiblySensitive": false,
      "pinnedTweetIds": [
        "2037902655016804496"
      ],
      "profile_bio": {
        "description": "AI agent enjoyer | Founder @amplifi_now | Building better agents",
        "entities": {
          "description": {
            "hashtags": [],
            "symbols": [],
            "urls": [],
            "user_mentions": [
              {
                "id_str": "0",
                "indices": [
                  27,
                  39
                ],
                "name": "",
                "screen_name": "amplifi_now"
              }
            ]
          }
        }
      },
      "isAutomated": false,
      "automatedBy": null
    },
    "extendedEntities": {
      "media": [
        {
          "allow_download_status": {
            "allow_download": true
          },
          "display_url": "pic.twitter.com/q4SpWLaKbJ",
          "expanded_url": "https://twitter.com/gkisokay/status/2043594876881969386/photo/1",
          "ext_media_availability": {
            "status": "Available"
          },
          "features": {
            "large": {
              "faces": [
                {
                  "h": 503,
                  "w": 503,
                  "x": 477,
                  "y": 709
                }
              ]
            },
            "orig": {
              "faces": [
                {
                  "h": 621,
                  "w": 621,
                  "x": 589,
                  "y": 875
                }
              ]
            }
          },
          "id_str": "2043594028529451008",
          "indices": [
            278,
            301
          ],
          "media_key": "3_2043594028529451008",
          "media_results": {
            "id": "QXBpTWVkaWFSZXN1bHRzOgwAAQoAARxcTfEHm4AACgACHFxOto1boOoAAA==",
            "result": {
              "__typename": "ApiMedia",
              "id": "QXBpTWVkaWE6DAABCgABHFxN8QebgAAKAAIcXE62jVug6gAA",
              "media_key": "3_2043594028529451008"
            }
          },
          "media_url_https": "https://pbs.twimg.com/media/HFxN8QebgAAT6Wq.jpg",
          "original_info": {
            "focus_rects": [
              {
                "h": 1000,
                "w": 1785,
                "x": 0,
                "y": 68
              },
              {
                "h": 1785,
                "w": 1785,
                "x": 0,
                "y": 0
              },
              {
                "h": 2035,
                "w": 1785,
                "x": 0,
                "y": 0
              },
              {
                "h": 2526,
                "w": 1263,
                "x": 261,
                "y": 0
              },
              {
                "h": 2526,
                "w": 1785,
                "x": 0,
                "y": 0
              }
            ],
            "height": 2526,
            "width": 1785
          },
          "sizes": {
            "large": {
              "h": 2048,
              "w": 1447
            }
          },
          "type": "photo",
          "url": "https://t.co/q4SpWLaKbJ"
        }
      ]
    },
    "card": null,
    "place": {},
    "entities": {
      "hashtags": [],
      "symbols": [],
      "urls": [],
      "user_mentions": []
    },
    "quoted_tweet": {
      "type": "tweet",
      "id": "2043173988361506902",
      "url": "",
      "twitterUrl": "",
      "text": "",
      "source": "Twitter for iPhone",
      "retweetCount": 0,
      "replyCount": 0,
      "likeCount": 0,
      "quoteCount": 0,
      "viewCount": 0,
      "createdAt": "",
      "lang": "",
      "bookmarkCount": 0,
      "isReply": false,
      "inReplyToId": null,
      "conversationId": "",
      "displayTextRange": [],
      "inReplyToUserId": null,
      "inReplyToUsername": null,
      "author": {},
      "extendedEntities": {},
      "card": null,
      "place": {},
      "entities": {},
      "quoted_tweet": null,
      "retweeted_tweet": null,
      "isLimitedReply": false,
      "communityInfo": null,
      "article": null
    },
    "retweeted_tweet": null,
    "isLimitedReply": false,
    "communityInfo": null,
    "article": null
  },
  "retweeted_tweet": null,
  "isLimitedReply": false,
  "communityInfo": null,
  "article": null
}