Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallBefore investing in a little-known AI company, verify the legal issuer, customer demand, product performance, costs, ownership, and the exact security being offered. A polished demo, a financing announcement, or a Form D filing is not enough. Treat each important claim as a question to verify against documents, customer evidence, and independent records—and keep a written list of what remains unproven.
The steps below are designed for a general investor evaluating a private company. The public-filing guidance is U.S.-specific; legal and regulatory obligations vary with the company’s jurisdiction, industry, product, and offering structure.
1. Identify the company and the investment you are actually considering
Pin down the issuer and its related entities
Start with the full legal name of the entity issuing the investment, its jurisdiction of formation, and any subsidiaries or trading names. Record the founders, directors, and the name of the security you are being offered. Do not assume the brand in a pitch deck is the entity that owns the product, signs customer contracts, employs staff, or issues your shares or note.
Ask for an entity chart showing which company holds the intellectual property, employs the team, signs customer and supplier agreements, and receives investment proceeds. This helps distinguish the issuer from similarly named businesses, a founder’s previous company, or an affiliated fund.
Recommended Free Tools
#1 Best Overall
Translate the pitch into a specific product claim
Write down who the customer is, what problem they pay to solve, where the product fits in the customer’s workflow, and which part—if any—depends on AI. Ask the company to label each feature as generally available, in a paid or unpaid pilot, on the roadmap, or shown only in a demonstration. For each material claim, request a dated supporting document or record; mark anything that has only been stated by management as unverified.
2. Check public records, but understand what they establish
Search SEC records when the U.S. offering makes them relevant
Search SEC EDGAR using the issuer’s exact legal name and, if known, its Central Index Key (CIK). If the company has made an offering that requires a notice, review its Form D and any amendments. Compare the issuer name, related persons, reported offering details, and filing dates with the company’s own documents. The SEC Division of Corporation Finance’s Form D FAQ and the SEC’s Form D filing instructions describe the filing process.
For specified exempt offerings, Form D is generally due within 15 calendar days after the first sale. The SEC defines the first sale for this purpose as when the first investor is irrevocably contractually committed. That timing rule is not a general deadline for every private-company financing. The SEC staff FAQ also says it reflects staff views and has no legal force or effect.
A Form D is a notice—not SEC approval, an audited financial statement, a complete capitalization table, or a guarantee that an offering is legitimate. Its presence does not verify the company’s product or business claims; its absence, by itself, does not prove that a company or offering is fraudulent.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Look for other records that fit the company’s circumstances
Depending on where the issuer operates and what it sells, review relevant corporate registries, court records, patent records, procurement databases, and regulatory records. Search by legal entity as well as by brand and key principals. A missing result is not proof that no business activity, legal obligation, dispute, or regulatory issue exists.
3. Verify that customers pay, use, and renew
Separate different kinds of customer evidence
Request a customer list divided into paid production deployments, paid pilots, unpaid pilots, and prospective customers. Ask the company to define what it means by “customer,” “user,” and “AI user,” including the unit counted and reporting period. A signed production contract, a paid pilot, an unpaid trial, a letter of intent, a waitlist entry, and a benchmark result are different kinds of evidence; do not treat them as interchangeable.
With the company’s permission, speak directly with a representative sample of current and former customers. Ask what they deployed, what the product replaced, who approved the purchase, how often it is used, what measurable result changed, and whether renewal or expansion is planned. Ask former customers why they stopped using it. A customer reference selected by the company can be useful, but it is not a substitute for corroborating records.
Reconcile revenue with the underlying records
Compare reported revenue with signed contracts, invoices, collections, credits, and churn. Separate recurring subscription revenue from one-time services, implementation, and integration work. Review customer concentration, cohort retention and expansion, implementation time, and how much reported backlog has actually converted into paid work. These records help distinguish headline sales from revenue that is collected and repeatable.
4. Evaluate the product beyond a curated demo
Design an evaluation that reflects real use
Arrange a demonstration, but define the evaluation tasks yourself rather than relying only on examples selected by the vendor. Use representative inputs, edge cases, and failure-prone or adversarial examples, then compare results with a conventional baseline or the incumbent workflow. Ask for the evaluation data and methodology, error rates by task or user group where relevant, the amount of human review required, latency, uptime, and evidence that results repeat outside a curated demonstration.
Set expectations about who will run the evaluation, what data can be used, and how results will be recorded. Do not infer production reliability from a successful demo or describe an evaluation as your own unless you actually performed it.
Trace dependencies and calculate the cost of delivering the work
Ask for a production-stack diagram that identifies foundation models, cloud and accelerator providers, retrieval or data vendors, open-source components, and human support. Request the cost per completed customer task at observed usage and under stressed usage, plus gross margin after inference and support costs. Review capacity commitments, rate limits, exposure to supplier price changes, and the fallback plan if a critical provider changes terms or withdraws access.
Find out whether the company develops its own model or builds its product on third-party models. Either approach can support a viable business; the difference affects supplier dependence, control, costs, and what might make the product hard to replace.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Use AI risk frameworks for questions, not as a seal of approval
NIST’s AI Risk Management Framework is voluntary and intended to support trustworthiness considerations in AI design, development, use, and evaluation. NIST lists the framework as released in 2023, its generative AI profile as released in 2024, and says AI RMF 1.0 is being revised. It can provide a vocabulary for discussing risks; it does not certify a company or establish product quality, legal compliance, or investment merit.
5. Check data rights, intellectual property, and security
Follow data from collection to use
Request an inventory covering data used for training, fine-tuning, evaluation, retrieval, and inference. For each source, ask who collected it, what contractual or legal permission supports its use, what restrictions apply, whether it contains personal or confidential information, whether customers can opt out, and whether submitted data is retained or used to train shared models. Confirm how the company handles deletion and access requests where those obligations apply.
Confirm rights to the technology and work product
Review the licenses for models, datasets, and third-party code. Check employee and contractor invention assignments, patent and trademark claims, trade-secret controls, and any disputes or notices. Confirm that the issuer—not just a founder, contractor, or affiliate—owns or has adequate rights to the technology it sells. A pitch-deck description such as “proprietary data” or “proprietary model” is not a substitute for the underlying agreements and records; legal review may be needed to assess them.
Assess security controls and contractual exposure
Ask for the security architecture and evidence concerning access controls, encryption, logging, incident response, and vulnerability management. Review customer security commitments, incident history and remediation, and any independent audit or certification. For any report or certification, check its date, scope, exceptions, covered systems, and the legal entity it covers. Read contracts to understand who bears responsibility if the system produces a harmful or materially incorrect result.
Best Value
6. Map governance, regulation, and legal exposure
Identify the markets and decisions the product affects
Map where the system is offered and what decisions it informs. Depending on the product and market, relevant issues may include privacy and data protection, consumer protection, employment, health, financial services, safety, export controls, and sector-specific rules. Have counsel familiar with the company’s markets assess which requirements apply; a general checklist cannot determine compliance for every deployment.
Ask who is accountable for model changes, evaluation, incident escalation, customer-facing claims, and board oversight. Review litigation, customer complaints, regulatory inquiries, insurance, indemnities, and contractual restrictions.
Distinguish a draft recommendation from a binding rule
A SEC Investor Advisory Committee Disclosure Subcommittee document dated November 18, 2025, was a draft for discussion at a December 4, 2025, committee meeting. It recommends that the SEC consider issuer definitions of AI, disclosure of board oversight, and separate discussion of material AI effects on internal operations and consumer-facing matters. It is not an adopted SEC rule or a legal requirement for a private startup. Those topics can still serve as useful prompts for asking how the company defines AI and oversees material risks.
7. Reconstruct ownership and read the actual investment terms
Reconcile the capitalization
Request the current fully diluted capitalization table and reconcile it with the stock ledger, charter, board approvals, options, warrants, SAFEs, convertible notes, debt, liens, and earlier financing documents. Look for promised equity, side letters, liquidation preferences, anti-dilution provisions, conversion caps or discounts, information rights, voting rights, transfer restrictions, and obligations to participate in later rounds. Confirm which entity owns material IP and signs customer and supplier contracts.
Model the security you would receive
Read the actual subscription, stock-purchase, SAFE, note, or other security documents. Model ownership and proceeds under multiple financing and exit outcomes, including dilution and downside scenarios. Compare valuations only after comparing the securities’ rights and the assumptions behind the valuation. Ask qualified legal and tax advisers to review the documents and your eligibility and jurisdiction. A public filing does not mean the SEC has checked the investment terms for you.
8. Make a decision memo with evidence labels and conditions
Before committing capital, write a short memo that separates what is verified from what is asserted or unresolved. For each material conclusion, label the evidence as independently verified, corroborated, management-provided, inferred, or unresolved. Cover the investment thesis, customer proof, product evidence, unit economics, defensibility, key dependencies, governance and legal exposure, capitalization, investment terms, and downside case.
List the evidence that would change your view and set conditions that must be satisfied before funding—for example, direct customer verification, documentation of data or IP rights, security remediation, or clarification of financing terms. If comparing multiple opportunities, assess each on the same dimensions rather than compressing trade-offs into a single “AI moat” score:
Quick Recap
- Customer urgency and willingness to pay
- Product performance and implementation burden
- Gross margin and compute exposure
- Data and intellectual-property position
- Distribution and retention
- Supplier concentration
- Governance and regulatory risk
- Cash runway
- Valuation, dilution, and security rights
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
