Cap Table

BLUEPRINT — every business spun up from the factory inherits this template. Instance data lives in .invisible/ventures/{slug}/finance/. Never hardcode a business name, currency, or specific investor in this file.

The cap table is the ownership truth. Get it wrong and every raise, exit, and hire breaks downstream. This template covers founder equity, SAFE mechanics, priced-round dilution, and option pool sizing — the four decisions that compound into the final ownership map.

Business Alignment

Question	Answer
Venture name	[instance: name]
Raise type	SAFE / priced seed / Series A / token / no raise yet
Conviction level	HIGH / MEDIUM / LOW / NONE
Who owns this cap table?	[founder / fractional CFO / cap-table-architect agent]

The Four Decisions

Each decision compounds. Get them in order.

#	Decision	Default healthy range	Kill signal
1	Founder equity split	Balanced by contribution, vested over 48 months with 12-month cliff	Unvested founder equity, equal split without rationale
2	Option pool size at seed	10–15% pre-money	Zero pool (dilutes founders at first hire) or 25%+ (over-dilutive)
3	Convertible instrument type	SAFE (post-money, valuation cap + discount) or priced round	Note with interest + maturity date (treats equity like debt)
4	Investor rights	Pro-rata, information rights, standard anti-dilution	Full ratchet anti-dilution, unusual liquidation preferences

Founder Vesting

Founder	Starting equity %	Vesting schedule	Cliff	Rationale
[Founder A]	%	48 months	12	[contribution basis]
[Founder B]	%	48 months	12	[contribution basis]

Rule: All founder equity vests. Unvested equity is a time bomb — co-founder departs early, walks away with unearned ownership, wrecks the next raise.

SAFE Mechanics

Parameter	Range	Notes
Valuation cap	Depends on stage	The price ceiling for the SAFE holder
Discount	15–25% typical	Applies if priced round comes in under cap
Post-money vs pre-money	Post-money default (YC 2018+)	Post-money is cleaner dilution math
MFN clause	Include	Most Favored Nation — if a later SAFE gets better terms, earlier holder gets them too

The SAFE trap: stacking SAFEs without modeling conversion dilution. Each SAFE converts at the priced round. Founders often discover they've sold 40% of the company when they thought they'd sold 20%.

Priced Round Dilution

Every raise dilutes. Model the before/after.

Line	Pre-money	New investment	Post-money	% ownership
Pre-money valuation	$X
Investment amount		$Y
Post-money valuation			$X + $Y
Existing shareholders				X / (X+Y)
New investors				Y / (X+Y)

Option pool shuffle: If the option pool is expanded as part of the round, the expansion dilutes pre-money shareholders, not the new investor. Always clarify which side of the round the pool sits on.

Scenario Branches

Three scenarios every cap table must model.

Scenario	What it answers
Base	Current structure, no changes
Next raise (proposed terms)	What's founder ownership after the next round?
Dilution stress	If we raise 2x more at same valuation, where do we land?

Common Killers

• Equal founder splits with no vesting
• SAFEs with no cap OR no discount
• Option pool carved post-money (founder takes the hit)
• Missing pro-rata for early investors (they'll ask for it in the next round anyway)
• Full ratchet anti-dilution (poisons any down round)

Receipt

On completion, cap-table-architect writes a receipt to .invisible/context/receipts/ with:

• factory_instance_id = parent factory
• business_id = this venture slug
• phase = finance
• task_id = finance.4
• Primary artefact = .invisible/ventures/{slug}/finance/cap-table.md

Context

• Unit Economics · Cash Flow · ROI Analysis
• Valuation Comps — setting the cap
• Investor Map — who prices the round
• Pitch Deck — where cap table meets narrative

Questions

• If the cap table forces you to choose between founder control and market-standard terms, which do you preserve?
• Which clause in a typical SAFE would you refuse — and why would an investor walk rather than accept your refusal?
• What does the cap table look like at exit if everything goes right? What if everything goes wrong?