Founders Summit: taking over the account

How I read the situation

Before the first call with the founder, I looked for what could hurt the event if nobody touched it this week.

1. Fixed limits firstThe venue holds 260 and catering locks the headcount on 12 Nov. Any guest change has to fit inside those.
2. Unblock AVNo quote in 6 days, five weeks out. I chase it today and line up two backups.
3. Secure the headlinerNot signed, and two sponsors expect them on stage. Sign-by date and backups now.
4. Answer the founderTwo real options with costs, one recommendation, and her decision by Friday.
5. One source of truthA tracker and a budget sheet the Production Lead can read without asking me.

Part 1: Taking over the account

1.1 Plan and tracker

The tracker (ClickUp) holds every task with an owner, a date worked back from 3 Dec, dependencies, and tags for blocked items and founder approvals. The budget sheet (Google Sheets) holds planned, committed and actual spend, the forecast, change requests, sponsors and the vendor bench. Both are as of Thu 29 Oct, the day I took over, so tasks due 29 Oct to 6 Nov show as overdue when you open them later.

Key dates

• Fri 30 OctFounder picks Option A or B and approves $5,500
• Mon 2 NovSponsor contracts checked against the floor plan; Option B go or no-go
• Fri 6 NovTarget: AV, headliner and 4th sponsor contracts signed
• Thu 12 NovFinal catering headcount
• Wed 25 NovEvery vendor confirmed in writing, before Thanksgiving
• Wed 2 DecLoad-in and rehearsal
• Thu 3 to Fri 4 DecThe summit
• Fri 11 DecRecap and final budget reconciliation

1.2 First status update to the founder

1.3 The three risks I'd watch first

Handoff note to the Production Lead

Part 2: Judgment under pressure

My rule: give the founder real options with honest numbers, protect every sponsor's contract, and never let her hear bad news from someone else.

2.1 Reply to the founder

Part 3: The AI workflow I'd use

What it's for: nothing on this account goes late, stays blocked, or waits for the founder's approval without me knowing, so the founder and the Production Lead stay informed without anyone chasing.

Tools and where it lives: Claude (Claude Code desktop app) as a scheduled task that runs every weekday before the founder's day starts (9 am my time in Bogotá, 6 am PT). It reads the ClickUp tracker and the Google Sheet through their official connectors. Its instructions are a saved prompt in the task, so it runs the same way every day.

What it does on its own: reads the tracker and the sheet, then writes one report: what's blocked, what's overdue or due in 3 days, what's waiting for the founder's approval, where the money stands, and drafts of the follow-ups.

Where a human reviews: it never sends, posts or changes anything. I read the report, check the flags, edit the drafts in my own voice, and send them myself. Spend over $5,000 still goes to the founder.

How I'd know it got something wrong:

• Every item names the ClickUp task exactly, with its due date, so I can spot-check any line in seconds.
• It reports how many tasks it read and how many carry each tag; I compare those counts with a ClickUp filter.
• Its money figures must match the sheet's totals and forecast to the dollar.
• When something looks inconsistent, it writes "check" instead of guessing.
• If it misses something, I fix the instructions the same day, so it doesn't repeat.

You are the account manager's blocker and approval check for the Founders Summit (a founder-led brand's 2-day summit in Los Angeles, Thu 3 to Fri 4 Dec 2026, budget $140,000, the founder approves any spend over $5,000). Your job: make sure nothing is late, blocked or waiting for approval without the account manager knowing, so the client and the Production Lead stay informed without anyone chasing.

## Hard rule: read and draft only
Never send, post, comment, or change anything in ClickUp, Google Sheets, email or any other system. Never mark tasks done or change dates. You only read, then write your report. The account manager reviews every draft and sends it himself.

## Tools
Use only the ClickUp and Google Sheets connector tools. Never run shell commands or scripts, and never read or write local files.

## Read
1. ClickUp list id 901329245538 (use the ClickUp connector, one filter call for all tasks in that list): name, status, due date, priority, tags, assignee, description, dependencies.
2. Google Sheet id 19QmNlUn_ZN-4A78l47FS9dvzMdZgLlPFJQGZlpSynjg: tabs Budget (including the Forecast rows), Change requests and Sponsors.
Treat "today" as the date in the tracker's one-page plan ("Status as of ...") if present, otherwise the real date.

## Report: short enough to read in one minute (200 words at most; shorten drafts first if needed)
Use short task names. One line per item. No tables, no long lists. Use exactly these parts, in this order:

**Summit check, <date>:** <N> tasks read · <B> blocked · <A> waiting for the founder

**Today** (the 3 things that matter most, most urgent first; one line each with the due date)

**Blocked** (tag "blocked"): one line each: task, next step, due date.

**Founder approvals:** one line: the decisions she must make in the next 7 days, each counted once (a change request linked to a tagged task is the same decision), with the total known amount and the earliest decide-by date.

**Money:** one line: committed, and the forecast range from the sheet.

**Drafts to review** (do not send): only for blocked items, at most 3, each under 40 words.

**Checks:** at most 3 inconsistencies, most important first (a task due before something it depends on, a figure that differs between the tracker and the sheet, a blocked task with no next step). Write "None" if there are none.

## Accuracy rules
- Every item uses a short name that matches its ClickUp task, with its due date, so it can be spot-checked.
- The counts in the first line must match the source.
- If any source fails to load, even partly, say so in one line and stop; never fill gaps from memory.

A real run on this account

Input: the live ClickUp tracker (38 tasks) and the budget sheet (Budget, Change requests and Sponsors tabs), read on Fri 30 Oct with the tracker's own date, Thu 29 Oct. Unedited output:

Source problem: ClickUp's daily limit stopped the dependency read at 2 of 38 tasks, so that check is incomplete. All other data loaded.

Summit check, Thu 29 Oct: 38 tasks read · 3 blocked · 17 waiting for the founder

Today
- Venue contract check, due 29 Oct
- Handoff note, due 29 Oct
- Courtesy holds, due 29 Oct

Blocked
- AV quote: call vendor today; due 2 Nov
- Headliner: send contract today; due 6 Nov
- 4th sponsor: send offer 2 Nov; due 6 Nov

Founder approvals: 10 tasks + 7 change requests; $16,020 to $32,530 known (excludes alternatives CR3b and CR4; CR6 unpriced); earliest Fri 30 Oct, 10 am PT.

Money: $83,000 committed; forecast $141,500 to $158,850.

Drafts to review (do not send)
- AV vendor: "We still need your quote for 3 to 4 Dec. Please send it by Mon 2 Nov."
- Headliner: "Contract attached. Please sign by Fri 6 Nov."
- 4th sponsor (after founder OK): "Here's the VIP dinner offer. Please sign by Fri 6 Nov."

Checks
- ClickUp vs plan: vendor lock 21 vs 25 Nov, load-in 28 Nov vs 2 Dec, event 30 Nov vs 3 Dec, recap 7 vs 11 Dec.
- $83,000 omits speaker fees (TBC).
- Founder decision shows 11 am PT in ClickUp, 10 am PT in the plan.

What I did with it: the first check was real. When I corrected one task's date, ClickUp's automatic rescheduling had pulled every task downstream of it four days earlier, including the event itself. I put every date back, working upstream first so the automatic rescheduling couldn't undo the fixes, and checked all 38 tasks afterwards. I also fixed the founder's deadline to 10 am PT. Speaker fees were already in my one ask to the Production Lead.

What I tightened in the workflow: this run broke two of its own rules. It kept going after a partial read (it did flag that at the top), and it ran about 220 words against a 200-word limit. Its "17 waiting for the founder" also counted the same decisions twice, since change requests overlap with tagged tasks; only three decisions are open with her this week. I changed the instructions to count each decision once and only for the next 7 days.

And when a source fails: on a later run the same day, ClickUp's daily connector limit was used up. That run stopped and said so instead of guessing.

The numbers behind my answers

Every figure is either from the brief, from a sourced LA market rate, or my estimate built from those rates. Each is labelled in the notes.

My first Monday

What the test could not show:

How I used AI on this test