Jake (@JakeDSBD24) header
Jake (@JakeDSBD24) profile picture

MarTech + AI systems architect. Building with AI agents, Workato. Publishing books & Apps w/AI. CLE sports. ➡️ https://teapaw.org/t.co/CH9q89or5U

Fishing on a Lake
Joined February 2010
Photo by Jake (@JakeDSBD24)Photo by Jake (@JakeDSBD24)Photo by Jake (@JakeDSBD24)Photo by Jake (@JakeDSBD24)Photo by Jake (@JakeDSBD24)Photo by Jake (@JakeDSBD24)Photo by Jake (@JakeDSBD24)Photo by Jake (@JakeDSBD24)Photo by Jake (@JakeDSBD24)Photo by Jake (@JakeDSBD24)
What’s up, @X ? I am 37 living in Alabama, dad of 2, into GTM Tech & Agent AI. Now I’m trying to vibe code my own apps for Apple using @rerun_build @ChatGPT @grok if you are into tech or building in public, let's connect 👍🏼
28
1
22
802
House Foreign Affairs says Commerce approved more than 879,000 AI chips for Chinese firms, Dec 2025 to May 2026. The paperwork got released. The shipping numbers did not. Same day, California's AG subpoenaed Nvidia. #AI
2
59
Guardians fans in November: patiently waiting for news scheduled to arrive around February. We are trained for this. Cleveland fans are trained for every kind of waiting.
1
36
A buying signal is not permission to send. Before an agent touches a prospect, I want to know where the signal came from, how fresh it is, and what happens if it is wrong. That is the GTM systems work under the demo.
18
Sup, @Muse what’s the latest update?
19
HubSpot published how it plans to price headless access, and the part I would circle is the boundary, not the welcome mat. Every seat includes everyday work from Claude or ChatGPT at no extra charge. Third-party agents can use the APIs already on the plan. Those calls count toward included API capacity and do not consume credits. If a team wants the governed environment, with traceability and purpose-built agent tools, that usage runs on HubSpot Credits. Three meters. The page does not define what everyday means. Anyone running a house with two young kids already knows included and a lot are not the same word. Write down which jobs stay on the seat, which ones ride API capacity you already pay for, and which ones move to credits because the agent got more responsibility. HubSpot says it is the number 1 CRM connector in Claude and ChatGPT. That is their line. I have not checked it. Open the boundary before the bill explains it for you. #MarTech #HubSpot #GTM
53
Anthropic priced its IPO on Nov 18 at $83 a share. The offering is 980,000,000 Class A shares. Anthropic is selling 956,351,354 of them. Selling stockholders are selling 23,648,646, and Anthropic receives no proceeds from those shares. The shares are expected to start trading today on Nasdaq under ANT. The offering is expected to close on Nov 20, if customary conditions are met. Expected is not closed. Underwriters also have a 30-day option for up to 147,000,000 more shares. I would not treat a ticker as a procurement plan. A public model vendor changes the operator question from which model is good to what happens when price, terms, or the exit ramp move. Gross proceeds are not stated on Anthropic's page. I will not invent a raised total. This is not a buy or sell note. If the invoice moves, the workflow still needs a second path. #AI #GTM
35
Final from Wednesday. Grizzlies 124, Cavs 121. Cleveland is 8-8. I was not in the building. I do not need to have been. Three points is still three points, and the standings do not take vibes as a receipt. #Cavs
1
55
People hear "AI builds my app" and picture a prompt and a prayer. The reality is branches, pull requests, code review, and merge rules. Dot and Codex write code inside the same guardrails a human team uses.
14
OpenAI introduced GPT-6.2 Astra and Sol on Nov 17. Astra is the frontier model for harder projects. Sol is the everyday one. OpenAI says Astra's API price is $8 per million input tokens, $0.40 per million cached input tokens, and $40 per million output tokens, 20% below GPT-6 Astra on input and output, with a 60% discount on cached input. Sol keeps the prior Sol rates, $2, $0.10, and $10. Those are OpenAI's prices and OpenAI's benchmark claims. I have not run the evals. A lower posted price is still a routing question. Which jobs can move to the cheaper model, what still needs the bigger one because a wrong step costs more than the tokens, and which score stays on their page until you check the work. OpenAI's LifeVal test includes shortlisting kindergartens after a move. That is their scenario, not my Tuesday. The house version of the same test is simpler. Did the finished task survive contact with two young kids and a morning that does not sit still. Announced availability is for Plus, Pro, Enterprise, and Edu in ChatGPT Work and Codex, plus the API. I have not turned it on. #AI #GTM
1
43
HubSpot says everyday Claude and ChatGPT work is included with the seat. Agent API calls ride the plan's API capacity and do not use credits. Scaled work in the governed environment uses credits. The page never defines everyday. Included and a lot are not the same word in a house with two kids. #MarTech
1
33
Grok has its own corner of my setup. I keep a shared brain for my bots so the knowledge does not live in five separate chat windows. Assistants are better when they can read the same notebook.
21
Anthropic priced the IPO at $83 a share. Trading is expected today under ANT, and the close is expected Nov 20, not done. A ticker is not a second path. If the invoice moves, the job still needs somewhere else to run. Not a buy note. #AI
40
A lower posted price is a routing decision, not a rank. Google lists Gemini 4 Neon at $3 per million input tokens and $15 per million output tokens, and puts it between Carbon and Hydrogen. I would still judge it by cost per finished task, and by the checks I run myself. #AI
34
A buying signal is not permission to send. Before an agent touches a prospect, I want to know where the signal came from, how fresh it is, and what happens if it is wrong. That is the GTM systems work under the demo.
1
56
Google’s Genie 4 makes interactive worlds feel closer. But the real test isn’t a convincing moment. It’s whether a scene stays coherent as you change it. Google itself flags latency and limits on long-term scene permanence. That’s the part I’d watch as an AI builder.
23
On a video call I look like a reasonably put-together professional. The second it ends, I am back in dad mode. There is no transition. There is just a laptop closing.
1
1
88
AI can help find a bug. The harder part is getting the fix into production. Wiz says AI-assisted scanning surfaced two NGINX HTTP/2 vulnerabilities. The fixes are out, but operators still need to identify systems with HTTP/2 enabled, upgrade to 1.30.7 or 1.31.9, and verify the fix in production. Wiz rates both vulnerabilities critical, while NGINX labels them major and medium. The useful question is whether your patch process confirms the upgrade actually reached production.
33
For agent and GTM workflows, I’d judge specialized inference by the whole job, not a chip headline. Does it improve latency, cost per completed task, reliability and portability? AMD plans to integrate Taalas tech into its accelerator roadmap. That’s not a shipped integration, and an acquisition isn’t performance data.
33
Oleria is extending identity governance to AI agents and previewing a gateway for MCP and REST actions. The useful question is how delegated authority gets scoped, logged and revoked. "The agent had access" is not a satisfying incident review.
1
1
55