There is a moment, usually about eighteen months into a promising startup's life, when a serious investor's lawyer asks for the data room. The founders, who have been heroically busy building a product, open a shared drive and find that it contains a pitch deck, a logo file, three versions of an NDA signed by nobody, and a Slack message from a co-founder that reads, "yeah we're 50/50, we'll paper it later." They never papered it later. The investor's lawyer is now reading that Slack message with the expression of a surgeon who has just found something unexpected on the X-ray.

This is the central, slightly unglamorous truth about startups: a company is, legally speaking, a stack of documents. The product is the company's soul, but the paperwork is its skeleton, and skeletons are easy to ignore right up until the body has to stand up under weight. Investors do not buy ideas. They buy clean, well-documented ownership of a clearly defined entity, with a clearly defined cap table and clearly defined intellectual property. Acquirers do the same, only with more lawyers and more zeros. The startups that raise smoothly and exit well are almost never the ones with the best lawyers at the finish line. They are the ones that did the boring documents correctly at the start.

This article is a guided tour of that document stack. We move roughly in the order a real company assembles it: formation and governance first, then the founder documents that determine who owns the equity, then the equity and option machinery, then the fundraising instruments, and finally the commercial and employment agreements that protect the operating business. For each, we explain in plain language what the document does, when you genuinely need it (and when you can wait), the legal authority that makes it work, and the specific, costly mistakes founders make. We lean on a recurring hypothetical, Acme Robotics, Inc., a fictional company that builds warehouse robots, so the abstractions stay concrete.

A word on tone before we start. None of this is meant to frighten you into hiring an army of attorneys for a company that is still a side project. Much of the early stack is standardized, cheap, and available in well-vetted templates — Thomson Reuters Practical Law, the National Venture Capital Association, and the Y Combinator accelerator all publish forms that have been pressure-tested in thousands of deals. The point is to understand what each piece is for, so you spend on the documents that protect you and skip the ones you do not yet need. Knowing the difference is the whole game.

Part One: Bringing the Company Into Existence

Choosing the entity, and why most fundable startups become Delaware C-corporations

Before any document, there is a decision: what kind of legal creature is your business? The realistic choices for a startup are a limited liability company (LLC) or a corporation, and within corporations, the choice of where to incorporate. This decision drives nearly everything that follows.

An LLC is the flexible, lightweight option. It is governed by an operating agreement rather than bylaws, it can be taxed as a pass-through entity (profits and losses flow to the owners' personal returns, avoiding the corporation's double layer of tax), and it lets you allocate profits and management rights in creative ways. For a bootstrapped consulting shop, a real-estate holding vehicle, or a lifestyle business with two owners and no outside-investment ambitions, an LLC is often exactly right.

A corporation — specifically a Delaware C-corporation — is the standard chassis for a venture-backed startup, and the reasons are worth understanding rather than memorizing. Venture funds and most institutional investors strongly prefer, and frequently require, a Delaware C-corp. Their own funds are structured to invest in corporate stock, not LLC membership interests, in part because an LLC can pass through "unrelated business taxable income" that fund investors, especially tax-exempt ones such as pensions and endowments, are organized to avoid. Stock options, the lingua franca of startup compensation, work cleanly in a corporation and awkwardly in an LLC. And Delaware itself offers a deep, predictable body of corporate law developed over more than a century by the Delaware Court of Chancery — a specialized business court whose judges actually understand fiduciary duties and merger mechanics, and whose opinions are read by every M&A lawyer in the country. The Delaware General Corporation Law (DGCL), codified at Title 8 of the Delaware Code, is the most heavily litigated and best-understood corporate statute in the United States, which is precisely why sophisticated parties want their disputes decided under it.

There is one more reason that tax advisors care about deeply: qualified small business stock (QSBS) under Section 1202 of the Internal Revenue Code. Stock in a qualifying C-corporation, held for the required period, can let founders and early investors exclude a substantial slice of gain from federal income tax on a sale — and the excluded gain is also exempt from the net investment income tax. The catch that makes this an entity-choice issue is structural: the issuer must be a domestic C-corporation, so an LLC taxed as a partnership earns no Section 1202 benefit at all. A partnership or LLC can convert to a C-corporation to start qualifying, but the five-year holding clock then runs only from the conversion date, and the company must convert before it crosses Section 1202's gross-asset ceiling. We return to QSBS below, because it changes how founders should think about the timing of their stock and their 83(b) elections; for now, file it under "another thumb on the scale toward incorporating."

The practical takeaway: if you intend to raise venture capital and grant equity to employees, you will almost certainly become a Delaware C-corporation — either now, or in a slightly painful conversion later. Converting an LLC to a corporation mid-stream is doable but costs legal fees, can create tax friction, and resets the QSBS clock, so founders with clear venture ambitions usually start as a Delaware C-corp on day one. For a deeper look at the tax tradeoffs, see our discussion of corporate structuring and running multiple businesses.

The certificate of incorporation: the company's birth certificate

The document that actually creates a corporation is the certificate of incorporation (called "articles of incorporation" in many states, and informally the "charter"). You file it with the Delaware Secretary of State under DGCL § 102, and the moment it is accepted, your company exists as a separate legal person that can sign contracts, own property, sue, and be sued — all without putting the founders' personal assets directly on the line. That limited-liability shield is the entire point of forming an entity, and it is why operating a serious business as a sole proprietorship or an informal partnership is a quiet form of financial recklessness.

The certificate is short but load-bearing. It states the corporate name, the registered agent in Delaware (a required local presence to receive lawsuits and official notices), the corporate purpose (almost always the catch-all "any lawful act or activity for which corporations may be organized under the DGCL"), and — critically — the authorized capital structure: how many shares the company may issue, and of what classes. A startup typically authorizes a generous block of common stock (say, ten million shares) so there is room to issue founder stock, carve out an option pool, and still have shares left for investors. When the company later raises a priced round, it amends the certificate to authorize a new class of preferred stock with its negotiated rights.

The certificate is also where a corporation adopts protective provisions that matter later. A clause exculpating directors from personal monetary liability for certain breaches of the duty of care is permitted under DGCL § 102(b)(7); after a 2022 amendment to that section, Delaware corporations may extend comparable protection to certain senior officers as well. These provisions are not glamorous, but they are part of why talented people are willing to serve on early-stage boards rather than treat a board seat as an uninsured liability.

A common founder mistake is authorizing too few shares — which forces an early amendment and another filing fee — or, worse, treating the certificate as a fill-in-the-blank afterthought when it should be coordinated with the founders' stock issuances and the option pool from the start. When Acme Robotics incorporates, it authorizes ten million shares of common stock and issues eight million to its two founders, leaving a comfortable reserve. That single coordinated decision makes every later equity conversation easier.

Bylaws and operating agreements: the operating manual

If the certificate is the birth certificate, the bylaws are the operating manual. Corporate bylaws are an internal governance document (not filed with the state) that set out how the corporation runs: how directors are elected and removed, how board and stockholder meetings are called and conducted, what officer roles exist and what they do, how quorums and voting work, and the indemnification rights of directors and officers. Bylaws fill the gaps left by the statute and the certificate, and they prevent the kind of procedural confusion that turns a routine board action into a contested mess.

The LLC equivalent is the operating agreement (also called a company agreement or LLC agreement). Because LLCs are creatures of contract, the operating agreement does even more heavy lifting than corporate bylaws: it defines ownership percentages, capital contributions, how profits and losses are allocated, whether the LLC is member-managed or manager-managed, voting thresholds, transfer restrictions, and the terms for admitting or buying out members. A state's LLC act supplies default rules only where the operating agreement is silent — and the defaults are frequently not what the owners would have chosen if anyone had asked them.

The founder mistake here is perennial and emotionally understandable: skipping these documents because "we're all friends and we'll figure it out." Disagreements are not a risk in a growing business; they are a certainty. Without bylaws or an operating agreement, two friends who fall out over the direction of the company have no agreed mechanism for breaking a tie, and they end up litigating questions that a one-paragraph clause would have answered for free. It is almost always cheaper, faster, and friendlier to agree at the outset than to resolve differences after the relationship has soured — which is also the logic that animates the founder documents in Part Two.

Organizational consents and the founders' first board actions

Filing the certificate creates the shell; the company is not actually operational until the first round of internal housekeeping is done by written consent. The initial board consent (typically preceded by an incorporator's consent appointing the first directors) adopts the bylaws, appoints the officers, authorizes a corporate bank account, approves the issuance of founder stock, adopts the equity incentive plan, and ratifies any pre-incorporation contracts the founders signed on the company's behalf. This sounds clerical, but it is the documentation that proves the company actually did the things it claims to have done. When an acquirer's lawyer later asks, "where is the board approval for this stock issuance?", a clean set of organizational consents is the difference between a smooth diligence and a fire drill — and, as we will see with 83(b) elections, a missing board approval can quietly invalidate a stock grant that everyone assumed was effective.

Part Two: Founder Documents — Pinning Down Who Owns What

This is where the most expensive mistakes live, because this is where equity, intellectual property, and human relationships intersect. Everything in this section exists to answer one question with paper instead of memory: who owns what, and what happens if someone leaves?

The founders' agreement: the prenuptial agreement of the startup world

A founders' agreement records the deal the co-founders made with each other before romance turned into a balance sheet. It addresses equity splits (who gets what percentage and why), roles and responsibilities, decision-making authority, what happens to a founder's equity if they leave or are removed, how to handle a founder who simply stops contributing, intellectual property each founder brings to the table, and confidentiality. In a corporation, much of this lands in a combination of the founders' stock purchase agreements, the bylaws, and a separate founders' or stockholders' arrangement; in an LLC, it folds into the operating agreement.

The reason this document deserves the "prenup" comparison is that founder breakups are common, painful, and far more damaging when there is no agreement. The cautionary tale every startup lawyer can recite is the co-founder who leaves three months in, contributes nothing further, and yet walks away holding a large, fully owned slice of a company that the remaining founders then spend the next decade building. Investors call that "dead equity," and it can make a company effectively unfundable, because no rational investor wants to back a business whose departed co-founder owns a third of it and answers no emails. The cure is vesting, which we turn to next — but the founders' agreement is where the parties commit to it before anyone has a reason to resist.

Founder stock purchase agreements and the discipline of vesting

When founders receive their shares, they do so through a founders' stock purchase agreement (often a "restricted stock purchase agreement"). The word restricted is the key. The shares are subject to a vesting schedule and, crucially, to a repurchase right: if the founder leaves before the shares vest, the company can buy back the unvested shares, usually at the original (tiny) purchase price. In the language of the tax code, this repurchase right is a substantial risk of forfeiture — a term that will matter enormously when we reach the 83(b) election in a moment. Vesting is the mechanism that ties equity to continued contribution, and it is the single most important protection against dead equity.

The market-standard schedule is four-year vesting with a one-year cliff. Nothing vests for the first twelve months; a founder who leaves before the one-year mark keeps nothing. At the one-year anniversary, a quarter vests in a single lump (the "cliff"), and the remainder vests in monthly increments over the following three years. The logic is humane and commercial at once: the cliff filters out people who leave very early, and the steady monthly vesting rewards staying.

It feels strange to founders to put their own equity on a vesting schedule. The answer is that vesting protects the founders from each other. Suppose Acme Robotics has two founders, Dana and Priya, each holding four million shares vesting over four years, and Dana leaves after fourteen months. Dana keeps a little over a quarter of her shares (the one-year cliff plus two months of monthly vesting), and the company repurchases the rest at the price she paid. Priya and the company are spared the burden of carrying a non-contributing co-founder's full stake for a decade. Without vesting, Dana would have walked away with four million fully owned shares and a permanent claim on the company's success. Investors know this, which is why a founder stock arrangement without vesting is a red flag that usually gets fixed — sometimes painfully, by imposing fresh vesting at the financing — before any institutional money arrives.

The 83(b) election: thirty days that can save (or cost) a fortune

Here is the part of this article to read twice. When a founder buys restricted stock subject to vesting, the tax code presents a fork in the road, and the founder has thirty days from the date of the stock transfer to choose the right path (26 C.F.R. § 1.83-2(b)). Choose correctly and the tax consequences are trivial. Miss the window and the founder can face a tax bill that grows with the company's success — on equity they have not sold and cannot spend.

The mechanics flow from Section 83 of the Internal Revenue Code (26 U.S.C. § 83). The default rule of Section 83(a) is that when you receive property — here, stock — in connection with the performance of services, you are taxed on it as ordinary income when it vests (more precisely, when your rights become transferable or are no longer subject to a substantial risk of forfeiture), on the excess of the stock's fair market value at vesting over what you paid. For a founder who paid essentially nothing for stock the company hopes will become extremely valuable, this default is a trap: each tranche that vests over four years is taxed as ordinary income at the (hopefully rising) value on each vesting date, and the founder owes cash tax having sold nothing.

Section 83(b) offers an escape. By filing an election with the IRS within thirty days of receiving the restricted stock, the founder chooses to be taxed now, at grant, on the spread between fair market value and purchase price. For a founder who buys stock at its current fair value at the very beginning, when the company is worth almost nothing, that spread is zero or near zero — so the tax due is zero or near zero. Just as important, the election starts the clock for long-term capital gains treatment and for the QSBS five-year holding period, so future appreciation is taxed at favorable capital-gains rates (or excluded entirely under Section 1202) on a sale, rather than as ordinary income on each vesting date. Practical Law maintains a standard "Section 83(b) Election: Restricted Stock" form precisely because the filing is routine but unforgiving.

To make the stakes concrete, consider a worked example adapted from Practical Law's own startup materials (this is a hypothetical). Two founders are each granted one million shares of restricted stock with the usual four-year, one-year-cliff schedule, valued at issuance at a par value of $0.00001 per share — so each pays $10 for a million shares. Founder A files no election. She owes nothing at grant, but as the company grows, the default rule of Section 83(a) catches up with her. If the stock is worth $0.10 a share when 25% vests at the one-year mark, she recognizes roughly $25,000 of ordinary income that year (the $25,000 vesting value minus the trivial price she paid) — and she keeps recognizing income, at rising valuations, on every monthly vesting event for the next three years. Because she cannot sell the illiquid shares to raise cash, she is paying real tax on paper gains. Founder B files an 83(b) election within thirty days. She owes essentially nothing at grant (she paid fair value when fair value was almost nothing), owes nothing as the shares vest, and converts all future appreciation into capital gain. When both founders eventually sell in an acquisition, Founder B walks away with the larger after-tax gain. Same company, same stock, same vesting — one piece of paper, filed on time, makes the difference.

The cruelty of the rule is its rigidity. The thirty-day deadline is regulatory and the IRS does not grant extensions for forgetting. A founder who misses it can owe escalating ordinary-income tax over the entire vesting period on stock they cannot sell, and the company picks up a parallel problem: it must value the stock at every vesting date and report the amount as compensation, with attendant withholding exposure. There are two narrow lifelines, and neither is appealing. First, counsel can confirm the grant date — occasionally a grant was conditioned on a board approval that was never obtained, or on a purchase price the founder never paid, in which case the stock was arguably never issued and can be re-granted (starting a fresh thirty-day window, but at the company's current, higher valuation). Second, the company and founder can amend the stock purchase agreement to accelerate vesting or grant a transfer right, which accelerates the tax event to the amendment date and the then-current value — better than nothing, but only if the value has not yet climbed.

A few practical points round out the picture. The founder — never company counsel — must file; a well-drafted founders' stock purchase agreement says exactly this and requires the founder to deliver the company a copy of the election, because the tax burden of a botched filing falls on the individual. For elections, the founder files with the IRS office where they file their return and provides a copy to the company; since final regulations took effect, taxpayers no longer must attach the election to the year's return, but keeping proof of timely filing — a certified-mail receipt is the classic belt-and-suspenders move — is non-negotiable, because the taxpayer bears the burden of proving the election was timely. (The IRS also temporarily permitted electronic signatures on 83(b) elections during the pandemic; confirm the current signature rules before filing.) And the election is exceedingly hard to revoke (26 C.F.R. § 1.83-2(f)) — so if the company fails and the shares are forfeited, the founder gets no refund of tax paid. That risk is mostly theoretical for early-stage founders paying near-zero value, which is exactly why the election is close to a free option for them. Acme's founders each file within a week of buying their stock and keep the receipts forever. Dollar for dollar, the thirty-day 83(b) election is one of the highest-stakes pieces of paper a founder will ever sign — and it is one paragraph long.

A subtle bonus, easy to miss, lives at the intersection of Section 83(b) and Section 1202. Because a service provider who makes an 83(b) election is treated as acquiring the stock on the election date for QSBS purposes, filing the election does not just start the capital-gains clock — it starts the five-year QSBS holding clock too, locking in eligibility for the gain exclusion from the earliest possible moment. A founder who files 83(b), holds for more than five years, and meets the other Section 1202 requirements may exclude a large portion (and, for stock acquired in recent years, potentially all up to the statutory cap) of the gain on an eventual sale. The Section 1202 exclusion has historically been capped per issuer at the greater of $10 million or ten times the taxpayer's basis in the stock, and Congress periodically adjusts the percentages, holding periods, and ceilings — most recently in 2025 legislation that expanded the benefit — so the exact figures are fast-moving and worth confirming with a tax advisor, because the dollars at stake can be enormous. The headline for founders: the same humble election that solves your ordinary-income problem may also be the on-ramp to a multimillion-dollar tax exclusion.

The IP assignment: making sure the company actually owns its company

A startup is often, at its core, a bundle of intellectual property: source code, designs, algorithms, brand, and trade secrets. Here is the uncomfortable surprise buried in copyright and patent law: the person who creates IP owns it by default, not the company they create it for. Under the Copyright Act, copyright vests initially in the author (17 U.S.C. § 201(a)). Under patent law, an invention belongs to the inventor unless and until it is assigned. A corporation is not automatically the owner of what its founders, employees, and contractors create. The bridge from creator to company is a written assignment of intellectual property, and without it the company may not own the very thing investors are trying to buy.

The "work made for hire" doctrine is where founders most often go wrong, because it is half-true in a way that breeds false confidence. Under 17 U.S.C. §§ 101 and 201(b), a work prepared by an employee within the scope of employment is automatically owned by the employer as a work made for hire — but that rule reaches copyright only, and only for genuine employees. It does not transfer patent rights, which always require a written assignment, and it does not capture work by independent contractors except for a handful of narrow statutory categories and only when there is a signed writing saying so. So the freelancer who builds your app, the contractor who designs your logo, and the consultant who writes your firmware each owns what they made until they assign it — even though you paid them. The safe and universal practice is a written assignment from everyone who touches the company's IP, full stop.

The classic disaster scenario is the founder who wrote the original code before the company existed. That founder personally owns the copyright in that code unless and until they assign it to the company in writing. If they later leave on bad terms, the company can discover that its core product rests on IP it does not legally own, and the departed founder holds the leverage. This is why every founder signs an IP assignment agreement — often folded into the broader confidential-information-and-invention-assignment agreement discussed in Part Five — at formation, transferring to the company all relevant pre-existing and future work product. Investors demand evidence of clean IP ownership during diligence, and a missing founder assignment is one of the most common reasons a financing stalls. For the mechanics of making these assignments enforceable across different states, see our guide to employee invention assignment agreements; for how confidentiality and assignment fit together in tech deals, drafting enforceable non-disclosure agreements for technology transactions; and for the broader landscape of how software is protected, legal protection of software.

Part Three: The Equity Machinery — Cap Tables, Option Plans, and 409A

Once founders own their stock, the company needs infrastructure to track ownership, compensate employees with equity, and value that equity defensibly. Get this machinery right and fundraising and hiring become routine; get it wrong and you create problems that compound silently for years.

The capitalization table: the company's financial spine

A capitalization table, or cap table, is the master ledger of who owns what. It lists every stockholder, every option holder, every convertible security (SAFEs and notes), and the number and percentage of shares each holds — both as issued today and on a "fully diluted" basis that assumes every option and convertible instrument has been exercised or converted. The cap table is not a contract in the sense of being signed and binding; it is a record. But it is the record that every investor, lawyer, and acquirer reads first, and an inaccurate cap table is a screaming signal that the company's affairs are disorganized.

The discipline a cap table imposes is precision. Every share issuance, every option grant, every SAFE, and every conversion must be reflected, with the board approvals to back each entry. Early-stage founders often keep the cap table in a spreadsheet; as complexity grows, dedicated software earns its keep. The cardinal sin is letting the cap table drift out of sync with reality — promising equity in conversations or emails without documenting it, so that the "real" ownership lives partly in people's memories. When Acme Robotics raises its seed round, the investor's very first request is the cap table, and because Acme has kept it precise from the first share issued, the diligence on ownership takes an afternoon rather than a month.

The equity incentive plan: how startups pay people they cannot afford

Startups cannot pay top-of-market salaries, so they pay in equity, and the vehicle is an equity incentive plan (commonly a "stock option plan" or "stock incentive plan"). The plan is a master document, approved by the board and the stockholders, that reserves a pool of shares (the "option pool") and sets the rules under which the company grants options and other awards. Each individual grant is documented by an option grant agreement specifying the number of shares, the exercise price, the vesting schedule (again, typically four years with a one-year cliff), and the type of option.

That last point matters for taxes. Options come in two flavors. Incentive stock options (ISOs), governed by Section 422 of the Internal Revenue Code, can be granted only to employees and offer potentially favorable tax treatment — no regular income tax at exercise if the holding requirements are met, though the alternative minimum tax can intrude. Non-qualified stock options (NSOs) can be granted to anyone, including advisors and contractors, but generate ordinary income at exercise on the spread between fair market value and the exercise price. A well-drafted plan supports both and lets the company grant the right type to the right person. (Some startups also permit "early exercise" of unvested options, which loops back to Part Two: an early exercise of unvested stock is itself a Section 83 event for which the holder may file a 30-day 83(b) election.)

The size of the option pool is itself a negotiation point in fundraising. Investors typically insist that the pool be expanded before their investment, so that the dilution from future hires falls on the founders rather than on the new investors — the so-called "option pool shuffle." Understanding this dynamic before you negotiate a term sheet can be worth real percentage points of the company.

The 409A valuation: setting the strike price without inviting the IRS

Here is the quiet rule that ties the option machinery together. When a company grants stock options, the exercise price (the strike price) must be at least the fair market value of the common stock on the grant date. If the strike is set at or above fair market value, the option is generally exempt from Section 409A of the Internal Revenue Code (26 C.F.R. § 1.409A-1(b)(5)). Set it below fair market value, and the option becomes deferred compensation that violates Section 409A, with severe consequences for the holder: immediate income inclusion plus an additional 20% federal penalty tax, plus interest. Section 409A was enacted after the Enron-era compensation scandals to police deferred compensation, and it swept discounted stock options into its orbit. Most companies sidestep the whole problem by the simple expedient of never granting below fair market value — which only relocates the question to "what is fair market value?"

For a private startup, that question has no obvious answer, because the common stock has no market price. Section 409A's regulations supply the framework. A non-public company must determine value by "the reasonable application of a reasonable valuation method" that takes into account all material information, and a stale valuation becomes unreasonable once new information (a closed financing, an issued patent, resolved litigation) materially changes the picture — which is why valuations generally expire after twelve months. To give companies certainty, the regulations offer safe-harbor methods, the most common of which is an independent appraisal meeting the standards used for employee stock ownership plans, performed no more than twelve months before the grant. Using a safe-harbor method creates a rebuttable presumption of reasonableness, shifting the burden to the IRS to show that the method or its application was "grossly unreasonable." This appraisal is what founders mean when they say "409A valuation," and a competent appraiser typically blends three standard approaches — an asset/book-value approach, a market approach comparing the company to similar public companies (adjusted downward for illiquidity), and an income approach discounting projected cash flows to present value.

One regulatory wrinkle is worth internalizing: the valuation method must be used consistently for both compensatory and non-compensatory purposes. You cannot tell the IRS your common stock is worth a dollar a share when you grant options and then tell an acquirer it is worth ten — the law expects one honest number, applied across the board. For the founder, the practical rule is simple: do not grant options off a number you invented. Obtain a 409A valuation, grant at or above that price, refresh it at least every twelve months and after any material event such as a financing round (each priced round resets the company's value), and you protect both the company and every employee who holds an option. Skipping it to save a few thousand dollars can expose your employees to penalty taxes and your company to a diligence problem that haunts every future deal.

Part Four: Fundraising Instruments — From SAFEs to the Priced Round

When a startup raises money, the form the investment takes is itself a major legal decision. The instruments range from simple and founder-friendly at the early seed stage to comprehensive and investor-protective at the priced venture round. Understanding the spectrum lets founders choose the right tool for the moment. As background, almost every seed instrument discussed here is sold in a private placement to accredited investors under a Regulation D exemption, not in a registered public offering — the streamlined form of a SAFE or note is never a substitute for securities-law compliance. For how to get ready for all of this, see preparing your startup for capital raising and the broader capital raising maze guide.

The term sheet: a handshake with footnotes

Before any financing closes, the parties sign a term sheet: a short document laying out the key economic and control terms of the proposed deal. Most of a term sheet is explicitly non-binding — an outline of intentions rather than a contract — but a few provisions usually are binding, typically confidentiality and an exclusivity or "no-shop" clause that stops the company from courting other investors for a set period. The term sheet is where the real negotiation happens: valuation, the size of the investment, the type of security, board composition, protective provisions, and the all-important option pool. Once it is signed, the long-form definitive documents largely follow its outline, which is why experienced founders treat the term sheet — short as it is — as the document that decides the deal.

Convertible notes: debt that wants to be equity

The earliest outside money — often from friends and family, angels, or full-time "super-angel" investors — frequently arrives as a convertible note: a loan designed not to be repaid in cash but to convert into equity at the company's next priced financing. Formally it is debt, with a principal amount, an interest rate, a maturity date, and a claim on the company's assets senior to equity. In substance, both sides usually treat it as deferred or unpriced equity; the noteholder's goal is to convert into the same preferred stock the company sells to its first institutional investor in the Series A, not to collect principal and interest at maturity.

The conversion mechanics are the heart of the instrument, and they come in three triggers. The ordinary path is a Next Equity Financing conversion: when the company raises its priced round, the note's principal and accrued interest convert into shares of that new preferred series. A Corporate Transaction conversion governs what happens if the company is sold while the note is still outstanding — typically the investor may elect either to be repaid (sometimes with a multiple of principal) or to convert into common stock at a discount to the acquisition price. And a Maturity conversion addresses the awkward case where the note reaches maturity without either of the above, usually letting the investor convert at a formula price or leave the note outstanding to keep accruing interest. To reward early risk, notes convert at a price below the next round's price, set by a discount rate (commonly around 20% off the new investors' price), a valuation cap (a ceiling on the conversion valuation, protecting the early investor if the company's value has soared), or — most commonly — the more favorable of the two.

Convertible notes have real virtues: they are familiar to investors, the documents are short, and they avoid the cost of negotiating a full priced round. There is even a subtle benefit for employee equity — because a note is debt rather than priced equity, it tends to keep the common stock's fair market value low, which keeps option strike prices (and the 409A valuation) attractive for new hires. But notes carry baggage. Because they are debt, they mature; if they have not converted by maturity, the noteholders can in theory demand repayment, giving them leverage founders dislike and often forcing a renegotiation in which the noteholders extract better terms for an extension. They accrue interest, which sits uneasily with the idea that the note is really deferred equity. The maturity-and-interest problem is precisely what the SAFE was invented to solve.

SAFEs: simple agreements for future equity

The Simple Agreement for Future Equity, or SAFE, is a creation of the Y Combinator accelerator, first released in 2013 and now one of the most common seed instruments in the United States. A SAFE is, in essence, a convertible note with the debt stripped out: it is not a loan, so it has no interest rate and no maturity date. The investor gives the company money now in exchange for a contractual right to receive equity automatically when the company next raises a priced round, on terms — a valuation cap, a discount, or both — fixed in the SAFE. There is no looming maturity date to renegotiate and no interest accruing, which is exactly why founders love it. Mechanically, a SAFE converts on substantially the same terms as a convertible note and sits in the same place in a liquidation; it simply lacks the hallmarks of debt.

A crucial wrinkle is the pre-money versus post-money distinction. The original 2013 SAFE was a "pre-money" instrument. In 2018, Y Combinator released a revised post-money SAFE, and the change is not cosmetic. In a post-money SAFE, the investor's ownership percentage is calculated after accounting for all the SAFE money raised, which means the investor's stake is locked in and protected from dilution by other SAFEs — but it pushes more of the dilution onto the founders. This matters enormously when a company raises through many SAFEs over time, the so-called "SAFE pileup" or "stacking" problem: founders who sell a series of post-money SAFEs without modeling the cumulative dilution can be genuinely shocked at how little of their company they own once everything converts at the priced round. The lesson is to model conversion on the cap table before signing, not after.

SAFEs are securities, and selling them is the sale of a security subject to federal and state securities laws. They are not a magic way to avoid compliance; they are a streamlined form, not a regulatory exemption. For the full treatment — the pre-money/post-money story, the stacking-dilution math with a worked cap-table example, and the founder and investor risks — see our dedicated guide, SAFEs: an overview of simple agreements for future equity.

Choosing among the seed instruments

There is no universally correct seed instrument — only the right one for a particular company, stage, and investor base. Three factors usually drive the choice. The first is investor sophistication: friends and family rarely want to negotiate complex terms, and educating unsophisticated investors on venture conventions can balloon a company's legal bill, which pushes small rounds toward the simplest instruments (SAFEs and short-form notes). The second is investor preference: founders want to minimize friction, so if the lead investors are most comfortable with notes, the company sells notes, and if they prefer priced preferred stock, the company may issue a streamlined "Series Seed" preferred. The third is cost versus amount raised: legal fees can devour a small round, so heavier instruments make sense only when the round is large enough that the friction is a small percentage of the proceeds. As a rough rule, larger seed rounds (around $1 million and up, often anchored by angels and super-angels) can justify priced Series Seed preferred, while smaller friends-and-family rounds lean on SAFEs and notes. Every feature that is a "pro" for the founder tends to be a "con" for the investor and vice versa; the art is matching the tool to the moment.

The priced round and the NVCA documents

Eventually a company graduates from convertibles to a priced round: a financing in which investors buy a newly created class of preferred stock at a negotiated price per share, fixing the company's valuation today rather than deferring it. This is where startup fundraising becomes genuinely heavy machinery, and the industry has standardized that machinery through the National Venture Capital Association (NVCA) model documents — a suite of model agreements maintained by the NVCA in collaboration with leading venture lawyers. Starting from the NVCA forms saves enormous time and legal cost, because the parties argue over the negotiated terms rather than reinventing boilerplate.

The core NVCA suite is several documents working in concert. The Stock Purchase Agreement is the master contract: the investors agree to buy preferred stock, the company makes representations and warranties about its business and capitalization (this is where those clean 83(b) and IP-assignment records pay off, because investors demand reps that elections were timely filed and IP is fully owned), and the closing conditions are set. The Amended and Restated Certificate of Incorporation creates the new preferred class and spells out its rights: the liquidation preference (preferred gets paid first, up to a stated amount, if the company is sold or wound down), dividend rights, conversion rights into common stock, anti-dilution protection (adjusting the preferred's conversion ratio if the company later sells stock at a lower price — "down round" protection), and protective voting provisions. The Investors' Rights Agreement grants registration rights, information rights, and pro-rata rights to participate in future rounds. The Voting Agreement sets board composition and drag-along rights (forcing minority holders to go along with a sale approved by the majority). And the Right of First Refusal and Co-Sale Agreement governs what happens when a founder tries to sell shares.

This is a lot of paper, and it is appropriate for a real institutional financing where millions of dollars and meaningful control rights are at stake. Negotiating these terms — especially the liquidation preference, board seats, and protective provisions — is where a startup lawyer earns the fee. For what happens after these documents are in place and new investors keep arriving, see the complete guide to adding new investors after your seed round.

Part Five: Commercial and Employment Documents — Protecting the Operating Business

The first four parts built and capitalized the company. This part protects it as it operates — as it hires, sells, shares secrets, and goes online. These are the documents the company signs over and over, and standardizing them well is a quiet competitive advantage.

Nondisclosure agreements: guarding the secrets

A nondisclosure agreement (NDA), also called a confidentiality agreement, is a contract in which one or both parties promise to keep designated information secret and use it only for an agreed purpose. NDAs protect everything from product roadmaps and source code to financial data and customer lists. A startup uses them constantly — with potential business partners, vendors, contractors, and employees — though note that most institutional venture investors will refuse to sign an NDA to see a pitch, both because they review too many similar deals to risk the conflict and because industry norms assume good faith.

An NDA's value is only as good as its drafting. The agreement should define confidential information precisely, carve out the standard exceptions (information that is already public, independently developed, or rightfully received from a third party), specify the permitted purpose, set a duration, and address return or destruction of materials. A vague or overbroad NDA is hard to enforce; a tight, purpose-built one is a genuine deterrent and a real remedy. NDAs also interact with trade secret law: under the federal Defend Trade Secrets Act and state law, information qualifies as a protectable trade secret only if the owner took reasonable measures to keep it secret — and a well-administered NDA program is exactly the kind of reasonable measure courts look for. For the deeper craft, see drafting enforceable non-disclosure agreements for technology transactions, and for building a broader secrecy program, building a trade secret protection program from scratch.

Offer letters and the PIIA: hiring people the right way

When a startup hires an employee, the relationship usually rests on two documents. The first is the offer letter — a relatively short document stating the position, start date, compensation, benefits, any equity grant, and (importantly) that the employment is at-will, meaning either party can end it at any time for any lawful reason. Most U.S. employment is at-will by default, and the offer letter's at-will language exists to avoid accidentally creating an implied promise of continued employment. Founders should resist the temptation to dress up an offer letter with vague assurances ("we see a long future here") that a disgruntled departing employee could later argue created a contract.

The second — and from the company's perspective the more important — document is the Proprietary Information and Inventions Assignment Agreement (PIIA), sometimes styled a CIIAA. This single agreement does three jobs at once: it imposes confidentiality obligations, it assigns to the company the intellectual property the employee creates in the course of their work, and it often includes non-solicitation and (where enforceable) restrictive covenants. The IP-assignment function is the crown jewel — and here the doctrine matters precisely. As noted in Part Two, copyrights created by an employee within the scope of employment are owned by the employer automatically as works made for hire (17 U.S.C. §§ 101, 201(b)), but that automatic rule does not extend to patents and other IP rights, which require an express assignment. The PIIA supplies that missing assignment, which is why every employee and contractor should sign one before doing any work, and why investors check for exactly this in diligence. To be enforceable, the agreement also needs adequate consideration: signing at the start of new employment usually suffices because the job offer itself is the consideration, but asking an existing employee to sign in a state where continued at-will employment is not enough consideration may require something extra — a raise, a bonus, or a promotion.

A few legal wrinkles deserve flagging. Several states limit how broadly an employer may claim an employee's inventions: California Labor Code § 2870, and similar statutes in Washington, Illinois, and elsewhere, carve out inventions the employee develops entirely on their own time, without company resources, and unrelated to the company's business, and a PIIA in those states must include the required statutory notice. Separately, the federal Defend Trade Secrets Act requires employers to include a specific whistleblower immunity notice in any agreement governing trade secrets in order to preserve the ability to recover exemplary damages and attorney's fees from an employee — so a well-drafted PIIA includes that notice too. And the enforceability of non-compete covenants varies enormously by state, with California and a growing list of jurisdictions banning or sharply limiting them; the FTC's attempt at a nationwide non-compete ban has been the subject of intense litigation, leaving the field unsettled and state-specific. For where that fight stands, see non-compete agreements under siege.

As the company grows past a handful of employees, it will also want an employee handbook to set out its policies consistently — at-will disclaimers, anti-harassment policies, leave policies, and conduct rules — drafted carefully so the handbook does not accidentally become a contract. See how to write an employee handbook.

Commercial contracts: MSAs, SOWs, and the deals that make money

As the startup begins selling, it relies on a family of commercial contracts. A Master Services Agreement (MSA) sets the general terms governing an ongoing relationship with a customer or vendor — liability limits, indemnification, confidentiality, payment terms, intellectual-property ownership, and warranties — while individual projects are documented by short Statements of Work (SOWs) that reference the MSA and specify the particular deliverables, timeline, and price. This structure lets the parties negotiate the heavy legal terms once and then transact quickly thereafter.

Vendor agreements, service agreements, and reseller or distribution agreements round out the commercial stack. A recurring theme across all of them is IP ownership: when your company hires a contractor to build something, or builds something for a customer, the contract must say clearly who owns the resulting work product — because, as we saw with the IP assignment, the default rules rarely match expectations. Limitation-of-liability clauses, indemnities, and warranty disclaimers are the other terms that determine who bears the risk when something goes wrong, and they are worth understanding rather than skimming. For software-specific terms, see software licensing agreements.

EULAs, terms of service, and privacy policies: the documents your users never read

If the startup has a product, app, or website, it needs the user-facing legal documents. An End User License Agreement (EULA) or Terms of Service (TOS) is the contract between the company and its users: it grants a license to use the product, disclaims warranties, limits liability, sets dispute-resolution terms (often arbitration), and reserves the company's rights. Courts will enforce these "clickwrap" and "browsewrap" agreements, but only if they are presented in a way that gives users genuine notice and a meaningful opportunity to assent — so the user-experience design of how the terms appear has become a legal question, not just a product one.

A privacy policy is, for many businesses, not optional but legally required. A web of laws — the California Consumer Privacy Act as amended by the CPRA, the growing roster of other state privacy statutes, the EU's GDPR for users in Europe, COPPA for services directed at children, and the FTC's authority over deceptive practices — requires businesses that collect personal data to disclose what they collect and how they use it, and increasingly to honor consumer rights to access and delete data. A privacy policy that does not match what the company actually does is worse than no policy at all, because it becomes the basis for an FTC enforcement action or a state attorney general's inquiry. Building a real compliance program early is far cheaper than retrofitting one after a breach; see our guide to developing a privacy compliance program and to data minimization.

A Worked Timeline: Acme Robotics Assembles Its Stack

It helps to see the documents arrive in order, because in real life they do not all happen at once. Here is how our hypothetical, Acme Robotics, Inc., builds its legal foundation.

On day one, Dana and Priya decide to go all in. They incorporate as a Delaware C-corporation, filing a certificate of incorporation that authorizes ten million shares of common stock and names a Delaware registered agent. The same week, an initial board consent adopts bylaws, appoints Dana as CEO and Priya as CTO, opens a corporate bank account, and approves the issuance of four million shares of restricted common stock to each founder under founders' stock purchase agreements with four-year vesting and a one-year cliff. Within thirty days — actually within a week — both founders file their 83(b) elections and keep the certified-mail receipts, which not only solves their ordinary-income exposure but starts their QSBS holding clocks. Both sign IP assignment agreements transferring to Acme the prototype code they wrote before incorporating. The dead-equity risk is now controlled, the IP is clean, and the tax clocks are set correctly. Total legal cost: modest, and entirely standardized.

Six months later, Acme adopts an equity incentive plan, obtains a safe-harbor 409A valuation, and grants ISOs to its first three engineers at a strike price equal to that appraised fair market value. Each engineer signs a PIIA with the California § 2870 notice and the DTSA whistleblower language, signed on the first day of work so the job offer itself supplies the consideration. The cap table, maintained from the first share, reflects every grant precisely.

A year in, an angel investor offers $500,000. Acme raises it on post-money SAFEs with a valuation cap, and Dana models the conversion on the cap table before signing, so the founders understand exactly how much they will own when the SAFEs convert. Eighteen months in, a venture fund offers a $4 million Series A. The fund's lawyer requests the data room. Because Acme has every document — clean and signed, with timely 83(b) elections and full IP assignments to back the representations the fund demands — the diligence on corporate and IP matters takes days rather than weeks, and the round closes on the NVCA forms with a liquidation preference, a board seat for the investor, and an expanded option pool negotiated into the term sheet. The founders own a smaller percentage than they did at incorporation, as they always will after raising, but they own it cleanly — and no one is reading a Slack message with a surgeon's expression.

That is the whole point of the document stack. None of it is glamorous. All of it is the difference between a company that can stand up under weight and one that collapses the first time someone leans on it.

Key Takeaways

The documents a startup needs arrive in a logical sequence, and each one answers a question that memory and good intentions cannot. Formation documents (certificate of incorporation, bylaws or operating agreement, organizational consents) bring the company into existence and erect the limited-liability shield. Founder documents (the founders' agreement, vesting stock purchase agreements, the thirty-day 83(b) election, and IP assignments) determine who owns the equity and ensure the company owns its own intellectual property. Equity machinery (the cap table, the equity incentive plan, and the 409A valuation) lets the company compensate talent it cannot afford to pay in cash, without inviting tax penalties. Fundraising instruments (term sheets, convertible notes, SAFEs, and the priced-round NVCA suite) shape how outside money comes in and on what terms. And commercial and employment documents (NDAs, offer letters and PIIAs, MSAs and SOWs, EULAs and privacy policies) protect the operating business as it hires and sells.

The recurring lesson is that the cheap, standardized, boring version of each document, done early, prevents the expensive, custom, painful version later. The 83(b) election is one paragraph and thirty days; missing it can cost a fortune and even forfeit a QSBS exclusion. The IP assignment is a routine form; missing it can make a company unfundable, because copyright's work-made-for-hire rule does not reach patents or contractors. Vesting feels strange to impose on yourself; without it, a departed co-founder can sink the ship. The startups that win the paperwork game are not the ones with the fanciest lawyers. They are the ones that understood, from the beginning, that the company is the paperwork.

Frequently Asked Questions

Do I really need a Delaware C-corporation, or can I start as an LLC? It depends entirely on your ambitions. If you plan to raise venture capital and grant stock options, the Delaware C-corporation is the standard chassis institutional investors expect, and it unlocks the QSBS exclusion under Section 1202 of the Internal Revenue Code that LLCs taxed as partnerships cannot earn. If you are building a bootstrapped business with no outside-investment plans, an LLC's flexibility and pass-through taxation may serve you better. You can convert an LLC to a corporation later, but it costs legal fees, can create tax friction, and resets the QSBS holding clock — so founders with clear venture ambitions usually start as a Delaware C-corp.

What is the single most time-sensitive document a founder will sign? The Section 83(b) election. When a founder buys restricted stock subject to vesting, they have only thirty days from the date of transfer to file the election with the IRS (26 C.F.R. § 1.83-2(b)). File it and a founder who bought stock when the company was worth nearly nothing owes nearly nothing in tax, converts future appreciation into capital gain, and starts the QSBS clock. Miss the window and the founder can face escalating ordinary-income tax over the entire vesting period on stock they cannot sell. Extensions are essentially never granted, so the founder — not company counsel — must calendar it and keep proof of timely filing.

My co-founder and I trust each other completely. Do we still need vesting and a founders' agreement? Yes, and partly because you trust each other. Vesting and a founders' agreement protect each of you from the situations no one anticipates: a co-founder who leaves early, who stops contributing, or whose life circumstances change. Without vesting, a departed co-founder keeps their full equity stake forever, creating "dead equity" that can make the company unfundable. Investors will insist on vesting before they invest, so it is far easier to agree to it at the start than to renegotiate it under pressure later.

Why does my company need an IP assignment if I'm the one who wrote the code? Because under copyright and patent law, the creator owns the IP by default, not the company. Copyright vests initially in the author (17 U.S.C. § 201(a)), and patents belong to the inventor until assigned. The "work made for hire" rule (17 U.S.C. §§ 101, 201(b)) makes the employer the automatic owner of copyrights created by employees within the scope of employment — but it does not transfer patents, and it does not reach independent contractors except in narrow categories with a signed writing. If you wrote your product's original code before the company existed, you personally own that copyright until you assign it. A missing founder IP assignment is one of the most common reasons a financing stalls.

What's the difference between a SAFE and a convertible note? Both are ways to raise early money that converts to equity at a later priced round, but a convertible note is technically debt: it carries an interest rate and a maturity date, and if it has not converted by maturity, the investor can in theory demand repayment. A SAFE (Simple Agreement for Future Equity), created by Y Combinator, strips out the debt features: no interest, no maturity date, just a contractual right to future equity. Both typically use a valuation cap and/or a discount and convert on similar mechanics. SAFEs are simpler and more founder-friendly, but watch the cumulative dilution when you stack several — especially the post-2018 post-money version. See our dedicated SAFEs guide.

Do I need a 409A valuation before granting stock options? Yes, if you want to protect your employees and your company. Options must be granted with a strike price at least equal to the fair market value of the common stock on the grant date; doing so generally exempts the option from Section 409A (26 C.F.R. § 1.409A-1(b)(5)). A private company's stock has no market price, so you obtain a safe-harbor independent appraisal to establish that value, which creates a rebuttable presumption of reasonableness. Set the strike too low without a valuation and you risk a Section 409A violation, which subjects the option holder to immediate income tax plus a 20% federal penalty. Refresh the valuation at least every twelve months and after any financing or other material event.

Will investors sign my NDA before I pitch them? Usually not, and that is normal. Most institutional venture investors decline to sign NDAs to review a pitch, both because they evaluate too many similar companies to risk the conflict and because industry norms assume good faith. Save your NDAs for the contexts where they genuinely belong and are honored: vendors, contractors, employees, and prospective business or acquisition partners. When you do use one, draft it tightly, define confidential information precisely, include the standard exceptions, and remember that a well-run NDA program also helps establish the "reasonable measures" required for trade-secret protection.

How much of this do I need on day one versus later? A surprising amount of the foundational stack belongs on day one, because it is time-sensitive or cheap to do early and expensive to fix later: incorporation, founder stock with vesting, the 83(b) elections, and IP assignments. The equity plan and 409A valuation come when you start hiring and granting options. Fundraising documents come when you raise. Commercial contracts and user-facing terms come when you start selling and put a product online. The mistake is not doing everything at once; it is skipping the day-one items because the company "isn't big enough yet to need lawyers."

Related Articles

This article provides general information and is not legal advice. The law varies by jurisdiction and changes over time, and several areas discussed here — including tax rules (notably Sections 83, 409A, and 1202), non-compete enforceability, and privacy requirements — are unsettled or fast-moving. Consult qualified counsel and a tax advisor before acting on any matter described above.