AI Board of Directors — 3-Exec Business Decision Review
Stress-test a business decision (pricing, hire, pivot, fundraise) with three blind AI executives who cross-examine each other and give a board verdict. Runs a business decision through three phases. Phase 1: three independent executive lenses (Strategy, Tech, Growth) review it blind. Phase 2: each exec cross-examines the other two's findings, challenging false positives and corroborating real ones. Phase 3: a board verdict tags every finding CONFIRMED, CORROBORATED or DISPUTED and ends with one Board recommendation line (e.g. Hold, Proceed after fixes). Built for founders and solo operators with no board or co-founder. It follows an anti-sycophancy rule: findings are grounded in what you described, numbers you did not provide are asked for rather than invented, and nothing is dropped silently. It does not provide legal, financial or investment advice, and it does not research markets or competitors; whatever you paste is sent to the model as-is, so redact sensitive details first.
Example input
Use the AI Board of Directors on this (a sample I wrote). I run a solo B2B SaaS called ExampleMetrics (example.com) that sends weekly KPI summaries to small agencies. We have 41 paying customers on a flat $29/month plan, about 3% monthly churn, and I work on it alone about 25 hours a week alongside a part-time contract job. I'm considering dropping the flat plan and moving to usage-based pricing (per connected data source, roughly $9 each) for all customers, including existing ones, starting next month. My thinking is that heavy users are subsidizing light users and I'd earn more per account. Billing is currently handled by a hosted checkout tool; I'd need to build metering myself. I haven't talked to customers about it yet, but I think most will be fine. I'd like to announce it by email this Friday. My goal this year is to reach $3k MRR and then quit the contract job. Should I go ahead?
Example output
## Phase 1 — Independent findings
### 📈 Strategy Exec
- Goal/decision mismatch: the stated goal is $3k MRR to quit the contract job. At 41 customers x $29 the current MRR is about $1,189 (derived from your numbers). Repricing existing customers does not add customers; it redistributes revenue among the 41 you already have. The plan never says what MRR this change is expected to produce, so it cannot be tied to the $3k goal. Missing number: how many data sources the average customer connects. Please provide it; I will not assume one.
- Unexamined alternative: grandfather existing customers on the flat plan and put usage-based pricing on new signups only, or add a higher tier for heavy users. Both capture the "heavy users subsidize light users" gap without a forced migration of the whole base.
- Quiet commitment: "for all customers, including existing ones, starting next month" commits you to a price increase on some accounts with roughly a month of notice. Any light user whose bill goes down is revenue you give up, which the plan does not mention.
- The "most will be fine" belief rests on zero customer conversations. At 3% monthly churn on a base this small, losing even a few heavy-usage accounts erases the gain.
### 🛠️ Tech Exec
- "I'd need to build metering myself" is the hard part, not a detail. Counting connected sources per customer per billing cycle, handling mid-cycle adds/removes, proration, failed syncs, and disputes is real billing logic. The plan gives no build estimate against a 25-hours-a-week budget that is already shared with a contract job.
- Dependency: the hosted checkout tool may or may not support per-unit/metered plans or changing plans on live subscriptions. This is not stated; it must be checked before any date is promised.
- Operational load: usage billing creates a new support class ("why is my invoice different this month?"). Nobody is staffed for it besides you, and invoice errors hit trust directly.
- First thing to break: a data source that disconnects and reconnects (OAuth expiry) could be billed twice or not at all. The plan has no answer for that.
### 🚀 Growth Exec
- No success metric or review date. "Earn more per account" is not a number. Define it, e.g. target ARPA, acceptable churn ceiling, and a date to roll back if churn exceeds it.
- Customer-behavior assumption unobserved: nothing in the description shows that customers accept per-source pricing, or that heavy usage correlates with willingness to pay.
…Real output from running the skill on a sample problem, shortened. The skill file itself is not published here.