Loom Guide — the two recordings and the five projects

Word-for-word guide for the two Loom recordings Akash makes for Collins (the daily loop, the weekly ad pack), plus one-page briefs for Collins's five projects this week. What to show on screen, what to type into Claude, what to say.

Updated 2026-09-07

1 How to use this page

Two recordings and five projects. Record Loom A tomorrow morning while you do the real work. Record Loom B on Thursday when you build the pack. Each recording is laid out the same way: what to have open before you press record, then a table of segments. In every segment: what is on your screen, what you type into Claude (in bold), and what you say (in the quote box). Say it in your own words; the quotes are the beats, not a script to read.

Rules for both recordings: no jargon without a one-line meaning. Say the name of the page before you click it. When you type into Claude, read what you typed aloud. When Claude answers, point at the one line that matters. Keep each segment under five minutes.

2 Loom A · The daily loop on AlphaLift (about 40 minutes)

What this recording is for. This is his every-morning job. Five screens, in order, every day. The recording shows one real morning, with real leads, so he sees what "done by 9:15" looks like.

Have open before you press record, in this tab order

1Command Home
2The nightly checks, logged in
3The list view of leads
4The lead board
5The client lane
6Austin's DripJobs board (app.dripjobs.com, signed in as AlphaLift)
7The problems board
8Claude, open in the workspace folder, a fresh session named "daily loop 8 Sep"

Opening (1 minute)

"This is my morning, every morning, and from next week it is yours. Five screens, in this order, done by 9:15. The point of each one is a question. I'll say the question before I open the screen."

Segment 1 · Command Home (3 minutes)

On screen	Command Home, top of page
Type into Claude	nothing yet

"Question one: is anything red? These six numbers are the week against the target. The band underneath is every automated system, worst first. If anything here is red, it becomes a problem card today. Not a note, not a Slack message, a card."

Point at one number and one rail. Say what green and red mean for that one.

Segment 2 · The nightly checks (4 minutes)

On screen	The nightly checks page, sorted by "nights running"
Type into Claude	Explain the top three findings on the nightly checks by streak, and tell me which ones are about AlphaLift.

"Every night at 5:30, the database asks itself 35 questions. Things like: is there a lead with no next step, is there a client card nobody touched in two days. Each answer is a finding. A finding that shows up two nights in a row is the signal. One night can be noise. Two nights is a pattern."

Read Claude's answer. Pick one finding.

"This one is ours. Watch where it goes: it becomes a card on the problems board at the end of this loop, with a name on it."

Segment 3 · The list view: prove yesterday's leads landed (10 minutes)

On screen	The list view, filtered to yesterday
Type into Claude	For every lead since yesterday with a name and a verified phone, show me the delivery rows: client, channel, status, last error, sent time.

"Question three, the most important one: did every lead from yesterday reach its client? A lead is a name plus a phone number we verified. The moment we have both, it goes to the client. Not after a zip code, not after photos. Name and phone."

Wait for Claude's table.

"This table is what our system believes. Now we check what the client sees. Those are two different things, and the second one is the truth."

On screen	Austin's DripJobs board, global search

Pick one lead from Claude's table with status "sent". Search the name on Austin's board.

"There it is. Name, phone in the phone field, source St. Pierre. That lead is proven. 'Sent' on our side is a timestamp. A card on the client's board is proof."

Pick one lead with an error or no row.

"This one is stuck. Here is the order you check, and you never skip a step or guess: first the delivery rows, which we just saw. Second, the send guard's log, which says whether each message was allowed and why."

| Type into Claude | Why did lead [name] not get delivered? Read the send-guard log first, then the GoHighLevel execution log, then the contact. One sentence on the cause. |

"Third is the execution log inside GoHighLevel, our messaging system. Fourth is the contact record. Only after those four do you form a theory. The reason this order matters: for a month we guessed, and the guesses were wrong."

Read the cause aloud.

"Every lead from yesterday ends in one of three buckets: proven, sent but not proven, stuck. Stuck ones get an owner today."

| Type into Claude | Save yesterday's leads as a three-column list, proven, unproven, stuck, to the ops log for today. |

Segment 4 · The lead board: who are we ignoring (8 minutes)

On screen	The lead board, "Needs reply" column

"Question four: is anyone waiting on us? This board is every homeowner conversation, by state. 'Needs reply' means their message is the last one in the thread and nothing answered it. By 9:15 this column should be empty. If it is not, we are the bottleneck."

Open one card. Read the thread on screen.

"Read the last three messages. Ask one thing: what should the bot have said here? Usually the answer is: ask for the phone number, or answer the question and then ask for the phone number. Two questions, then the phone. Every extra question loses people."

| Type into Claude | Show me the last ten send-guard log rows for lead [id]: when, kind, allowed, reason. |

"The guard is the gatekeeper on every automated text. If it said no, this log says why. Quiet hours, the person already replied, a duplicate, a hold. If it said yes and nothing went out, that is a different problem, and it is a card."

| Type into Claude | Add a note to that card: cause in one sentence, next action, due tomorrow. |

"Every card you touch leaves with a next action and a date. A lead with no next action is a lead nobody owns."

Segment 5 · The client lane: nudge the client (8 minutes)

On screen	The client lane

"Question five: what is the client sitting on? These are leads we handed over. Green is under a day, amber under two, red past two. The client's job is to call, quote, book. Our job is to make sure they do, politely, every day."

Click the coaching draft for AlphaLift.

"Every night the system writes one draft per client from what is red. Nobody has been sending these. That is the gap you are closing. Before you send anything, check the board, because our labels lie sometimes."

On screen	Austin's DripJobs board, search one red name, then the Appointments calendar

"Board first, ask second. If the card is already booked on his calendar, we do not nudge. If it is sitting there untouched, we do."

| Type into Claude | Rewrite the AlphaLift coaching draft as one message from me to Austin. Plain, short, end with: mark it in your portal, or reply with one word. |

"Portal first, always. When Austin taps a button in his portal, our records update themselves. When he texts a word, a human has to transcribe it. So every nudge points him to the portal and offers the text as the fallback."

Send it, or say it goes out at 3.

| Type into Claude | Log one client send in the daily touch log: AlphaLift, now, nudge on [name]. |

Segment 6 · The problems board (5 minutes)

On screen	The problems board, filtered to unowned

"Last question: what broke today, and who owns fixing it? Every problem from the last forty minutes goes here once. If it is already on the board, it gets a name. If it is new, it gets a card. Anything that shows up three times in a week stops being a patch and becomes a build."

| Type into Claude | Log a clunk: key [short-key], what broke: [one line from this morning]. |

| Type into Claude | Set owner to me on the card for the finding we picked in segment 2, and read it back. |

"That finding from the nightly checks now has an owner. That is the whole loop: the system finds it, a person owns it, the ledger records the fix. If you cannot say how you would undo a fix, it is not designed yet."

Closing (1 minute)

"That is the morning. Five questions: anything red, what repeated overnight, did yesterday's leads land, is anyone waiting on us, what is the client sitting on. Then one more: what broke. Tomorrow you run it and I watch."

3 Loom B · The weekly ad pack (about 30 minutes, Thursday)

What this recording is for. Once a week a new set of foundation repair videos gets built and staged for my approval. He has never seen this. The recording shows one video going from a winning hook to a finished, captioned ad on a contact sheet, entirely by talking to Claude.

Have open before you press record

1The scoreboard, Creatives tab, set to All time
2Google Drive, the folder "Proven B2C Ads by Niche" → foundation repair
3Voicebox, open and finished loading
4Claude, open in the workspace folder, a fresh session named "FR pack week 37"

Opening (1 minute)

"Every week we make a new set of ads for foundation repair. Even while ads are off, the pack gets built and staged, because the day a client signs we relight the same day. Three parts to every ad: the voice, the picture, the words on screen. Claude makes all three. My job is choosing and approving."

Segment 1 · Read what won (5 minutes)

On screen	The Creatives tab, All time, sorted by cost per qualified
Type into Claude	Open the Creatives tab of the scoreboard, all time, and rank the top three foundation repair videos by cost per qualified lead. Give me the hook words, the surface on screen, spend, qualified leads, and cost per qualified.

"Cost per qualified lead is the only number I sort by. Views and clicks are noise. Note two things on each winner: the first words of the voiceover, and what is on screen in the first second. Those are the hook and the surface. Everything we build this week is a variation of those."

Segment 2 · Pick the footage (3 minutes)

On screen	Drive, "Proven B2C Ads by Niche" → foundation repair

"These are finished ads that already made money. Before we generate anything new, we re-voice a winner. New footage costs real money and real time; a proven winner with a new read is the cheapest test we have."

| Type into Claude | Write the brief for this week's FR pack: the three hooks with their numbers, which proven ad to re-voice first, and the market it is for. Save it in ready-to-deploy as this week's brief. |

Segment 3 · The voice (7 minutes)

On screen	Claude
Type into Claude	Generate an Eric voiceover for this script with the standard treatment: We are a local company that specializes in foundation stabilization. We fix sagging floors, wall cracks and sticking doors, without the cost or mess of a rebuild. Send us a message. We can come out and get you a free estimate right away.

"Eric is our voice. It is a free local voice engine on this computer, no account. The script is always the same four beats: who we are, what we fix with the thing on screen named first, no cost or mess, send us a message. No adjectives. 'Standard treatment' means a slight speed-up and the pauses cut short, with one longer pause before 'send us a message'. Raw computer voice never ships; the treatment is what makes it sound like a person."

Play the file. Point at the gate result.

"That check compares the pace against the read that converted. If it fails, we regenerate. We never stretch the voice to fit a clip. The picture gets retimed to the voice, never the other way round."

Segment 4 · The picture (5 minutes)

| Type into Claude | Generate one five-second foundation-crack b-roll clip with Higgsfield: vertical, no text, no people, a real crack in a concrete slab. |

"Higgsfield is the video generator. It shows the credit cost before it renders, and I want you to always look at that number. Real footage beats generated footage on cost per lead, so generated clips are for gaps, not the default."

Segment 5 · Words on screen and the contact sheet (6 minutes)

| Type into Claude | Burn the red karaoke captions from the script onto that clip, loop the clip to the length of the read, and build a contact sheet for this one ad. |

"Captions come from the script, not from listening to the audio; the listener mishears 'lift' as 'live'. White words on solid red, the spoken word turns yellow. The clip loops until the voice ends."

On screen	The contact sheet Claude opens

"This sheet is the approval. Every ad we ship goes on one of these, I look at it, I say yes. Nothing goes to Meta before that. From here, the steps you will not do this week are: create the ads paused, run the launch check, activate on my word."

Closing (1 minute)

"That is one ad. A pack is many. Thursday you build, you send me the sheet, Friday I approve, and the day a client signs we are live in an hour."

4 Collins's five projects this week

One per day. Each is one sentence to Claude to start, and one thing that proves it is done. Post the proof in Slack by end of day.

Monday · The AlphaLift autopsy

Start with: "Trace every qualified AlphaLift conversation. For each one: the stage it reached, where it stopped and why, whether a transfer row exists, and whether the card is on Austin's board. One row per conversation."

Then: for every one marked recoverable, "create an issue row owned by me."

Done when: a sheet with one row per conversation, saved to the ops log, and an issue row for each recoverable lead.

Tuesday · Kansas, created to launch-ready

Start with: "Run the page readiness sweep and show me everything Kansas is missing."

Then, in this order, one sentence each: rename the page to "KS Foundation Repair & Waterproofing" (Akash clicks); attach both service accounts; create the tracking dataset in Events Manager (a browser step, Claude drives Chrome); "fire the dataset test event for the Kansas page and show me the result"; accept the Lead Ads terms (a browser click); connect the page in GoHighLevel (Akash's login); "stand up Kansas's routing rows, dry run first, then apply"; "add the Kansas page to the router and deploy through the gate."

Done when: the readiness sweep shows Kansas green on every check. No ads.

Wednesday · One post on every live page

Start with: "List the live state pages and which of them our page tokens can post to."

Then: "Take the top proven foundation repair video, write two captions, one for northern states about freeze and thaw, one for southern states about soil and heat, and post it to every live page with the right caption. Log each post."

Done when: one post per live page, a log of page, caption and post id, and a written note titled "what broke" saved beside the log. That note is the specification for the posting system.

Thursday · Comment to message, one page

Start with: "Show me how the comment-to-message automation is set up for Level King and what it needs to run on the Kansas page."

Then: turn it on for Kansas only; comment on Wednesday's post from the test account; "show me the DM that went out and the board card it created."

Done when: one test comment produced one message, one captured phone, one card on the board. Then turn it back off until Akash rules on fleet-wide.

Friday · Own the problems board

Start with: "Show me every unowned card on the problems board."

Then: "Set owner to me on all of them." Then one clunk per problem hit Monday to Thursday: "Log a clunk: key …, what broke …". Then "Rebuild the problems board and deploy it through the gate."

Done when: zero unowned cards, five or more clunks logged, one ledger row per fix with how to undo it, and the board redeployed.

5 The words, one line each

Word	Means
The brain	Our database. Every lead, message and change lives there. Claude reads it.
The vault	The part of the brain that holds passwords and keys.
The ledger	The log of every change: what, why, how to undo it.
The send guard	The gatekeeper on every automated text. Says yes or no and writes down why.
The nightly checks	35 questions the database asks itself every night. A repeat is the signal.
The problems board	One card per thing that keeps breaking.
A clunk	One line about something that broke today. Three in a week becomes a build.
DripJobs	The clients' own job boards. The only proof a lead reached them.
GoHighLevel	Our messaging system. The bot and the texts live there.
The contact sheet	One page of every ad in a pack. Akash approves it before anything ships.