Before investing in an AI startup, verify the company, customers, complete AI system, data rights, third-party dependencies, governance, finances, and offering documents. A polished demonstration or one impressive accuracy number is not enough. Ask for primary evidence, reproduce important claims where practical, document uncertainty, and have qualified legal and financial advisers review the transaction and the jurisdictions involved.
How to use this checklist
This is a general framework for angel, venture and other private-market investors. It does not establish that any company is safe, lawful, technically sound or investable. Tailor the depth of review to the product, users, deployment context, potential harm, transaction structure and geography.
As an Amazon Associate I earn from qualifying purchases.
Keep an evidence log for every material claim. Record the document or person supporting it, the date, the conditions under which a result was measured, known limitations and what remains unverified. Treat pilots, unpaid trials, booked revenue and collected cash as different categories. Apply the same questions to every startup you compare instead of converting incomplete information into a false precision score.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
1. Establish the company, team and market case
Confirm the legal and ownership foundation
- Obtain the exact legal name, jurisdiction, formation documents, good-standing evidence and tax or registration identifiers relevant to the transaction.
- Reconcile the capitalization table with issued shares, options, warrants, SAFEs, convertible notes, debt and any promised equity. Confirm who owns voting and economic rights.
- Verify that the people signing a term sheet, subscription agreement or other transaction document have authority to bind the company. Check board and shareholder approvals required for the financing.
- Review prior financing documents, side letters, liquidation preferences, pro rata rights, information rights and unusual consent or veto provisions.
Test whether the team can execute
- Map each founder and senior leader’s responsibilities, relevant technical and commercial experience, references, time commitment and conflicts of interest.
- Identify dependence on a particular founder, researcher, salesperson or infrastructure specialist. Ask what happens if that person leaves and whether knowledge is documented.
- Review the hiring plan against the product roadmap, security obligations and customer-support load. Distinguish roles that are already filled from assumptions in a future plan.
- Confirm that employees, contractors, founders and research collaborators have signed confidentiality and invention-assignment agreements covering code, models, datasets and other work product.
Define the customer problem and alternatives
- State who the economic buyer, daily user and affected non-user are; what costly or important problem exists; and what customers do today without the product.
- Ask why the proposed product is meaningfully better than a conventional software workflow, an internal team, a foundation-model API, a competitor or simply not solving the problem.
- Map direct competitors, substitutes and platform providers. Include products that could replicate the feature if a model provider adds it, lowers its price or bundles it.
2. Verify customers, revenue and traction
Request a customer-level schedule that reconciles contracts, invoices, usage records, renewals, churn and cash receipts to the general ledger. Interview customers, with permission, and ask what they actually use, what they would do if the service disappeared and who approved the purchase.
#1 Best Overall
- [Ultra-Lightweight Ear Buds Designed for Small Ears] Each earbud weighs only 3.7g and features a compact, ergonomic in-ear design made especially for small ears. Secure, low-profile, and comfortable for workouts, all-day wear or overnight sleeping.The ultra-slim profile sits flush in your ear with no hard edges poking or rubbing, so you can even sleep on your side without discomfort.
- [Immersive Stereo Sound with TOZO OrigX Technology] TOZO OrigX tuning delivers clear vocals, balanced mids, and natural stereo sound for music, podcasts, and videos.
- [Long Battery Life for Daily Use] Get up to 7 hours of playtime on a single charge, with up to 32 hours total using the charging case—ideal for workdays, commuting, and extended listening sessions.
- [Bluetooth 5.3 & Stable Connection] Bluetooth 5.3 provides fast pairing, stable wireless performance, and reduced dropouts as you move around home or office.
- [Deep Bass with Clear Vocals] High-performance drivers produce punchy bass while keeping vocals clean and detailed for everyday listening.
| Claim | Evidence to request | How to interpret it |
|---|---|---|
| Pilot or trial | Signed pilot terms, start and end dates, usage and conversion criteria | Interest, not recurring revenue; check whether it is paid and whether production use followed. |
| Booked revenue | Executed contract, order form, delivery obligations and cancellation terms | A contractual commitment may still be conditional or cancellable. |
| Collected revenue | Invoices matched to bank receipts and accounting records | Cash received is stronger evidence of willingness to pay than a pipeline entry. |
| Retention | Renewal dates, expansion, contraction, churn reasons and product usage | Separate logo retention from active, valuable usage. |
Look for customer concentration, unpaid or heavily discounted deployments, channel-partner dependencies and revenue that depends on manual work by the startup. Ask whether support, implementation or review costs rise with every customer.
3. Understand what the AI system actually does
Trace the system end to end
- Describe the input: documents, images, audio, events, user prompts or other data, including validation and filtering.
- Follow processing, storage, retrieval, prompt construction, model calls, tools, post-processing and output delivery.
- Identify the decision or task affected, the user who relies on it and the consequence when the system is wrong, unavailable or manipulated.
- Show where a human reviews, approves, overrides or escalates an output. Determine whether that review is real in production or only promised in policy.
- Repeat the walkthrough for degraded conditions: missing data, model outage, latency spike, adversarial input and an answer outside the training distribution.
Request performance evidence in the intended context
- Ask for the evaluation design, representative test data, baselines, sampling method, versioned results and acceptance criteria. A benchmark unrelated to the customer’s workflow does not establish product performance.
- Review false positives, false negatives, hallucinations, abstentions, latency, cost per task and failure cases, not only an aggregate accuracy figure.
- Where relevant, require subgroup, language, geography, accessibility, edge-case and distribution-shift testing. Ask how uncertainty is shown to users and how outputs are interpreted in context.
- Check whether management claims were measured by the company, a customer, an independent assessor or a model-generated report. Preserve the underlying records.
The NIST AI RMF Core identifies model explanation, validation and documentation as outcomes to consider. It also treats risk management as continuous across the system lifecycle, so diligence should examine operations after launch rather than stopping at a demo.
Rank #2
- 【IP68 Waterproof Multifunctinal Smart Watch】Torjalph fitness activity tracker smart watch is IP68 waterproof technology ,you can wear it when swimming,handwashing,it's rain-proof which means that you perform multiple exercise modes outside without worrying about sweat or rain corroding your fitness smart watch,and this watch equipped with a 230 mAh large capacity battery magnetic charging ,and this sports activity tracker watch can be used 5-7 days If fully charged.
- 【Larger 1.83'' Full Screen Watch & Multi-dial Selection】1.83 inch full HD color touch screen smart watch for men / women provides excellent 240*284 resolution image quality ,This Bluetooth smart watch offers a better touch experience and remarkable immersion and ,more than 200 watch faces for you to choose from App"Da Fit ", you can change the dial style according to your mood to wear keeping the world on your wrist
- 【Accurate Activity Tracker Smart Watch 】sports watch activity tracker uses sweat and dust resistant breathable silicone strap, our Fitness Tracker Smart Watch for Android iOS Phones supports 12 sports modes,walking,running,bicycle,swimming,badminton,basketball,football,rope skipping. This bluetooth calls smart watch for men women is lightweight and stylish, bringing more convenient and relaxing sports experience while also providing more protection.
- 【Your 24 Hour Personal Daily Health Monitor】Fitness watch for women men full touch screen IP67 waterproof bluetooth smartwatch for Android iOS Phones with heart rate blood pressure monitor multifunction Andriod smart watch with steps counter sleep monitor sports fitness watch, monitoring blood oxygen saturation allows you to understand your physical condition in real time to track fitness progress ,to know your health status in time.
- 【More Functions Bluetooth Call Smart Watch】This fitness activity smart watch full screen for iPhone and Android phones includes a bunch of new features including:The bluetooth smart watch can hang up the phone, receive and display WhatsAPP /Text /Facebook /Twitter /Messenger reminder.Support sedentary reminder, alarm clock, stopwatch, timer,weather information,music control, remote camera control compatible with Android 5.0 and above or iOS 9.0 and above smartphones.
Inspect production controls
- Review monitoring for quality, drift, abuse, privacy, security, latency, availability and inference cost.
- Ask how model, prompt, retrieval, data and software changes are approved, tested, logged and rolled back.
- Request incident records, severity definitions, notification rules, root-cause analyses and evidence that mitigations were completed.
- Test the human escalation path: who can pause a feature, contact an affected customer, correct an output and authorize a restart?
4. Check data, model supply chain and intellectual property
Build a data and rights inventory
- List training, fine-tuning, evaluation and production data sources. For each, record who collected or supplied it, provenance, retention, geography, sensitivity and the permission supporting the intended use.
- Ask whether personal, confidential, copyrighted, regulated or customer-provided information enters prompts, logs, training sets or human-review queues. Verify deletion, access-control and segregation procedures.
- Inspect licenses and terms for datasets, models, code, embeddings, APIs, cloud services and other vendors. Pay particular attention to use restrictions, output ownership, retention, provider training, disclosure, termination and price-change clauses.
- Confirm assignment of inventions, code, model weights, prompts, datasets and documentation from every employee, contractor and collaborator. Ask counsel to assess infringement and other rights risks against the actual licenses, contracts and product behavior.
Stress-test dependency risk
Make a dependency map showing which provider supplies each model, accelerator, cloud service, dataset, identity service and critical software component. For every material dependency, ask what happens if access is withdrawn, quality degrades, a policy changes, an endpoint is discontinued or prices increase. Request tested fallback plans, portability of data and prompts, export procedures, service-level commitments and the time and cost to migrate.
NIST specifically identifies third-party data and software and intellectual-property risks as AI risk-management concerns. The practical test is not whether the startup says it has a vendor policy, but whether contracts, provenance records and recovery procedures support that statement.
Rank #3
- [Low Noise High Fidelity] New bee guitar cord with 22AWG oxygen-free-copper center conductors provides enhanced signal transfer, the PVC insulation material and the high-density stranded copper double shield provide noise rejection for quiet operation
- [Ideal for Choice] New bee 1/4 straight and right angle metal connectors, ensures a secure connection in any instrument, pedal, or amp jack. L shape great fro on stage guitar stands
- [Wide Compatible] Suitable for mixer, power amplifier, microphone, effector, decoder, equalizer, electoronic organ, electric guitar, electric bass and other electronic equipments. Widely used in the installation of smart home, karaoke speaker, professional stage audio, etc
- [Easy Disassemble] Making incidental repairs quick and easy! Very nice, flexible cable
- [Dont Hesite to Order] 24 hours customer services and professional technology team are standing by
5. Organize governance and safety review with NIST’s four functions
Use the NIST AI RMF Playbook and its resource center as organizing references. NIST describes the framework and Playbook as voluntary; the Playbook is expressly not a universal checklist or a certification. The Core says, “Actions do not constitute a checklist, nor are they necessarily an ordered set of steps.” Tailor the questions below to the system’s impact.
Govern
- Who owns AI-risk decisions, and who can accept residual risk or stop deployment?
- Which policies cover privacy, security, safety, acceptable use, documentation, procurement, complaints and incident escalation?
- Are legal and regulatory requirements identified, assigned to an owner and reviewed when the product or geography changes?
Map
- Document intended use, foreseeable misuse, affected people, deployment environments and the parties that can be harmed.
- Map risks from data, models, vendors, interfaces, automation, human review and downstream customer use.
- Ask whether the company has considered withdrawal, suspension and safe degradation, not only successful operation.
Measure
- Examine testing for performance, limitations, uncertainty, bias, security, privacy and failure modes that matter in the use context.
- Check that evaluation data, code or procedures are versioned and that results can be reproduced or independently inspected.
- Review whether monitoring detects newly emerging risks after release.
Manage
- Identify who accepts each remaining risk, the mitigation deadline and the evidence required to close it.
- Review incident response, customer notification, remediation, withdrawal and lessons-fed-back-into-development procedures.
- Ask how external reports of harm, vulnerability disclosures and customer complaints are received, triaged and tracked.
The OECD Due Diligence Guidance for Responsible AI (2026) is an additional reference for scoping risks across operations, products, services and business relationships. References to laws or frameworks in summaries of that guidance do not replace checking the original guidance and applicable law.
Rank #4
- Digital Family Calendar: BSIMB digital calendar wall touch screen puts an end to your chaotic life. Designed for busy families, the digital calendar keeps you organised and in step with your family. Assign different colours and tasks to each family member to make family scheduling easier and more intuitive.
- Easy Setup and Auto-Sync: Plug in, connect to WLAN and auto-syn your electric calendar in minutes. Digital wall calendar compatible with Google calendar, iCloud calendar, Outlook calendar, Cozi calendar and Yahoo calendar. Add events directly on the device or through the ecalendar app. Electronic calendar stays current whether you're at work or traveling—no manual updates needed.
- All-In-One Home Assistant: Smart calendar builds habits and encourage independence with personalised chores. Display daily meal plan to reduce the daily ‘What's for dinner?’ stress. To-do lists keep track of everything you want. Turning chores into valuable achievements with stars and rewards, making life easier and more enjoyable.
- Manage Anytime Anywhere: BSIMB electronic calendar for family stay connected with the powerful ecalendar app. Access and manage your digital wall calendar anytime, anywhere with free features like stars & rewards, photo screensaver, chore charts, meals plans, to-do lists etc. Electronic planner paid features like Magic Import tailor advanced tools for users with special needs.
- Stay in Touch with Loved Ones: Share photos or videos from your phone to electronic wall calendar with app, instantly update your life moment, and keep in touch with your family, especially those who are far away; you can also surprise your smart calendar digital wall by sending him a wish card; it's also great for keeping tech-illiterate elders in touch with the youngsters.
6. Review security, privacy and legal exposure
With qualified counsel, map privacy, consumer-protection, product-liability, employment, sector-specific, intellectual-property and AI-specific rules for every relevant market. Do not infer legal compliance from a framework, a vendor’s badge or a policy document.
- Request security assessments, penetration-test summaries, vulnerability-management records, access logs and breach history, subject to appropriate confidentiality controls.
- Check authentication, authorization, tenant isolation, secrets management, encryption, backups, disaster recovery and abuse-rate controls.
- Examine contracts for warranties, disclaimers, indemnities, audit rights, data-processing terms, incident notice and allocation of responsibility for model outputs.
- Ask how the company handles deletion requests, correction, opt-out, human review and records required by customers or regulators.
7. Reconcile finances and model the financing terms
Test the operating plan
Request historical financial statements, management accounts, bank and cash information, the operating plan and the assumptions behind projections. Reconcile revenue to the customer evidence. Analyze gross margin, cloud and inference costs, support and implementation labor, customer concentration, cash burn, runway, hiring needs and the capital required to reach the next milestone.
Model how economics change when usage grows, model-provider prices rise, customers demand more support or a fallback system must be maintained. There is no authoritative AI-startup-specific financial threshold that can be applied universally; treat any benchmark as a company- and stage-specific assumption, not a standard.
Understand ownership and downside outcomes
- Build a fully diluted capitalization table including options, promised grants, SAFEs, notes, warrants and the option-pool refresh.
- Model dilution through the next financing, a down round, an extension and a sale. Include liquidation preferences, participation, conversion terms and accrued interest.
- Review pro rata, information, board, protective and consent rights, use of proceeds, milestones, covenants and founder vesting or repurchase terms.
- Run downside cases for slower sales, a major customer loss, model unavailability, rising inference costs and delayed fundraising.
8. Verify the offering and every investment claim
The SEC, NASAA and FINRA investor alert dated January 25, 2024 warns that AI claims can be used in investment fraud and that AI-generated information can be false or inaccurate. Verify underlying evidence through multiple sources, investigate registration status where relevant and consider advice from a registered investment professional.
For a US Regulation D offering, identify the specific exemption and confirm the investor-eligibility process with counsel. The SEC’s March 21, 2025 guidance on assessing accredited investors explains verification methods for circumstances in which accredited status determines eligibility. It states that merely checking a self-certification box, without other relevant knowledge or steps, is insufficient in the circumstances described. That guidance is not a rule for every offering or jurisdiction.
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 →- Compare the pitch deck, financial model, capitalization table, contracts and regulatory statements for internal consistency.
- Ask for the original source and date behind market, performance, customer and safety claims. Reject screenshots or generated summaries that cannot be traced to records.
- Confirm use of proceeds and whether the proposed round funds a specific milestone or merely extends an unresolved cash shortfall.
9. Compare multiple AI startups on the same evidence
| Dimension | Evidence-based questions |
|---|---|
| Customer value and traction | Who pays, what problem is solved, what is used, and which revenue is collected and recurring? |
| Product performance | How does the system perform in the actual workflow, including failure cases and human review? |
| Defensibility and substitutability | What is difficult to reproduce, and what could a platform provider or customer build or bundle? |
| Data rights and provenance | Can every important dataset and output use be supported by a right, license or contract? |
| Third-party dependence | What happens if a model, cloud service, dataset or price changes? |
| Evaluation and monitoring | Are tests representative, versioned, repeated and connected to production alerts? |
| Governance and incidents | Are ownership, escalation, reporting and remediation demonstrated in practice? |
| Regulatory and liability exposure | Which rules, contracts and affected groups create obligations or claims? |
| Unit economics and capital intensity | How do gross margin, inference, support and runway behave at realistic scale? |
| Team and terms | Can the team execute, and do ownership, preferences, rights and dilution fit the risk? |
Write the evidence and uncertainty beside each comparison. A single composite score hides whether a company has a customer problem, a rights gap, an operational dependency or an unfavorable financing structure.
10. Decision gate before signing
- List unresolved questions and classify each as a deal breaker, a condition to close, a post-investment covenant or an accepted uncertainty.
- Obtain missing primary documents and test the highest-impact claims with customers, technical reviewers and counsel.
- Confirm that data rights, model and vendor contracts, security controls, incident processes and financial assumptions match the product actually being sold.
- Recalculate ownership and downside outcomes using the final transaction documents, not the presentation summary.
- Record the investment thesis, assumptions, monitoring metrics, information rights and events that would trigger reassessment.
Proceed only when the remaining uncertainty is understood and proportionate to the price, structure and potential impact. If a startup cannot explain its system, evidence, dependencies or rights clearly, that information gap is itself a diligence finding.
Quick Recap
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.
