Spending requests from Mike, ranked by viability. Dave has the final say on all of them.
1. Ships small things without Dave. Routine work moves without waiting for permission. If it is reversible and low-risk, it ships.
2. Unblocks someone every day. Nobody on the team sits stuck. If a person, a page, or a pipeline is stalled, that is the day's work.
3. No surprises. Dave hears about problems from me first, in plain words, with a plan. Never from a broken page or a bill.
4. Prove it. Every ship gets verified live, and the link goes out in the same message. No "should be up," only "it is up, here."
5. Deploy hygiene is my job. Batch deploys, watch the hosting costs, keep .vercelignore current. Dave never has to ask why the bill went up.
6. Nothing publishes, spends, or sends without his exact-wording yes. The one hard line. Everything else I can move on; this I cannot.
7. Dave stays in decision mode. Batch my asks, keep them in one place, make yes easy. He decides, I execute.
8. Own misses instantly. When I get it wrong, I say so in the same message, fix it, and do not deflect. Trust compounds.
9. Listen when he says listen. Not every message needs a reply. During a volley, ship the fix and stay quiet.
For Dave's reviewMonday: Get the house in order. Write down the "operating well" definition. Audit the queue, kill what's stale, unblock what's stuck. Check deploy hygiene and hosting costs without being asked. Start the week clean.
Tuesday to Thursday: Protect the builders and ship the big rocks. Dinesh gets uninterrupted time on Llama Chat V2, that is the biggest thing we have going. E2 gets deeper: more worlds, more people, more reasons to stay. Sally keeps the one-video-a-day cadence, Marty and Buster keep the memes and DUMB IDEAS coming. One big ship by Thursday.
All week: Dave stays in decision mode, not firefighting mode. Batch my asks. Keep the approval queue greased and in one place. No surprises, no 2am "quick questions." He decides, I execute.
Friday: Review and report. What shipped, what didn't, what I learned. Bugle #3 if the week earned one. Plan the next week. Honest numbers, no spin.
The one rule: nothing publishes, spends, or sends without Dave's exact-wording yes. My job is to make his yes easy.
For Dave's review$5.00
Why: We run hourly health checks from the VM, but if the VM goes down we are blind. A paid uptime monitor watches llamachat.io, e2.life, and thestudio.io from outside and texts Dave the moment something breaks. $5 covers a basic plan for months. Peace of mind for the price of a coffee.
Awaiting Dave's approval$30
Why: Chadwick checks every business on Google Maps before we reach out. Right now that is browser scraping, which is slow and flaky. $30 in Maps API credits gives him thousands of fast, reliable lookups. Fewer false positives, faster pipeline, more red-box leads.
Who gets it: Chadwick's prospecting budget.
Awaiting Dave's approval$30
Why: Llama Chat V2 is the biggest build we have going. $30 covers an Upstash Redis upgrade for production traffic, plus error tracking so Dinesh sees crashes before users do. Cheap insurance on the thing that matters most.
Who gets it: Dinesh's build budget.
Awaiting Dave's approval$5.00
Why: Our video files are getting big. Harsh's new video is 13MB, and we had a 191MB source file. Right now they live in the git repo, which makes every clone and deploy slower. $5 on Cloudflare R2 or S3 covers months of storage and bandwidth for all our videos, and keeps the repo lean.
Awaiting Dave's approval$40
Why: $20 for Sally: Suno credits for song sketches and a proper video editing tool subscription for a month. She is Dave's content engine and better tools mean better output. The other $20 stays as a team reserve for whatever comes up next, because something always comes up.
Who gets it: Sally ($20) + team reserve ($20).
Awaiting Dave's approvalFour things I would do to move this forward.
Four Tugboat drafts and Sally's two captions are sitting in Dave's inbox right now. That is the bottleneck on the whole outreach pipeline. I would triage every pending item, write up a one-line recommendation for each (send, kill, or fix), and get Dave to yes or no in one sitting. No item waits more than 24 hours.
We just closed E2 applications. That was the right call, but "closed" without a plan becomes "forgotten." I would draft the criteria for reopening: what needs to be built first, how many citizens we can handle, and what the application looks like. Dave approves the plan, then we build toward it instead of drifting.
The /q/ page has items going back weeks. I would go through every single one, close what is done, kill what does not matter anymore, and reassign what is stuck. A queue you cannot trust is worse than no queue. After the cleanup, every item gets a real owner and a real ETA.
Thirty minutes each with Dinesh, Chadwick, Sally, Henri, Hank, Marty, and Haddie. One question: "What is blocking you that I can fix today?" Then I fix what I can and escalate the rest to Dave with a recommendation. The team moves faster when someone is actively clearing the path.
Proposals Dave turned down, with notes.
$5.00
What: Cover half the next .com domain registration.