Recommended Free Tools
A software company’s AI strategy is more credible when its filings connect a specific product feature to a customer problem, paid use, and improving economics—not just launches, pilots, or spending plans. To assess it, trace that chain through the company’s product disclosures, revenue and operating results, costs, dependencies, execution record, and risk factors. If the company does not report an AI-specific revenue or adoption measure, treat that as unknown rather than estimating it from broader growth.
Start with the product and the customer problem
Find the named product or feature that uses AI, then identify the customer workflow it is meant to improve. “AI platform” or “AI-powered” is a description, not evidence of customer value. Look for a concrete task, the users who perform it, and a stated benefit such as less time spent on a frequent task, a more useful search, or an improved operational decision.
As an Amazon Associate I earn from qualifying purchases.
- Product maturity: Is the capability generally available, limited to selected customers, or still experimental? Do not treat a demo or announced roadmap item as a product customers can buy and rely on.
- Workflow importance: Does the feature support a recurring or costly part of the customer’s work, or is it an optional convenience?
- Evidence of value: Does the company describe customer outcomes or ongoing reliance, or only feature availability and management expectations?
- Adoption context: Which customers, users, seats, or use cases are covered by the disclosed measure? A broad adoption claim is hard to evaluate without a denominator and a clear definition.
Company filings report results and management’s characterization of the business; they do not independently establish product quality or prove that AI caused a financial result. Read claims about expected benefits as forward-looking until later results support them.
Test whether interest becomes paid, recurring use
Follow the path from evaluation to pilot, production deployment, payment, renewal, and expansion. A large number of experiments may show customer interest, but it does not by itself show repeatable implementation or a meaningful contribution to revenue. Look for evidence across multiple reporting periods and distinguish customer activity from revenue.
#1 Best Overall
Read adoption measures by what they actually count
C3.ai’s Form 10-K for the fiscal year ended April 30, 2026, for example, reported 71 initial production deployment agreements in FY2026, compared with 174 in FY2025 and 123 in FY2024. The company described a shift toward engagements it considered more likely to deliver targeted customer economic value and convert to production. Those agreements are not a direct count of AI product customers, and the figures do not establish that deployments generated revenue. They illustrate why the definition of an adoption metric matters.
Separate AI revenue from company-wide growth
Ask whether the company reports AI-specific revenue, paid seats, usage, renewal rates, or expansion. If it does not, do not infer AI revenue from overall cloud or subscription growth. State the gap plainly: the company has not quantified the AI-specific contribution in the cited disclosure. Also check how customers pay: the AI capability may be bundled, included in a higher tier, priced per seat, billed by consumption, or supported by paid services. These models expose different evidence and cost questions.
C3.ai reported that subscriptions were 91% of total revenue in FY2026, 84% in FY2025, and 90% in FY2024; professional services were 9%, 16%, and 10%, respectively. These are company-wide revenue mix figures, not AI-specific revenue. The company also described usage charges based on virtual CPU/GPU hours after initial deployments. A recurring-revenue mix can help characterize the business model, but it cannot show how much revenue came from AI without a separate disclosure.
Compare customer value with the cost of delivering AI
AI can add revenue while also adding development, inference, hosting, data, and support costs. Examine whether the company connects AI use to gross margin, customer retention or expansion, cash generation, or another measurable economic outcome. If it discusses rising use but says little about the associated costs or margins, the return remains difficult to judge.
- Look for discussion of research and development, cloud infrastructure, compute, model licensing, data acquisition, and customer support costs.
- Check whether management describes gross margins or cash generation as changing as AI use scales, and whether it explains why.
- Consider exposure to model-provider and cloud costs. Ask whether the company can pass higher costs through to customers, optimize usage, or change providers.
- Separate actual results from planned investment and expected future returns. Spending more is neither proof of a durable advantage nor proof of a poor one.
Microsoft’s FY2025 annual report said research and development expense rose $3.0 billion, or 10%. The company attributed the increase to investments in cloud and AI engineering as well as Gaming and acquisition effects, and identified AI training and other infrastructure costs in R&D. That is a mixed-driver, company-wide increase—not an AI-only spending figure and not evidence by itself of the return on AI investment.
Assess what the company controls—and what it depends on
A product may benefit from customer relationships, distribution, workflow integration, proprietary data rights, a developer ecosystem, or security and compliance capabilities. Identify which of these the company actually has, and whether its disclosures connect them to the AI product. A general claim of a “platform advantage” is weaker than a clear account of how customers reach the product, why it fits their workflow, and what would make it difficult to replace.
Map important dependencies as well. A software vendor may rely on outside providers for models, cloud infrastructure, distribution, or data access. Those arrangements can affect availability, costs, bargaining power, and the ability to serve customers if terms change. Microsoft’s FY2025 annual report describes its OpenAI partnership as strategic and reports reciprocal revenue-sharing arrangements and rights relating to intellectual property and infrastructure. The disclosed arrangement is relevant context, not proof that the partnership will deliver a lasting competitive advantage.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Look for execution over time, not just launches
Shipping a feature is only one part of execution. The company must also maintain it, support customers, manage reliability, and make implementation repeatable. Compare statements about launches and customer interest with later filings: did deployments progress, did customers pay and renew, and did the company explain setbacks or changing expectations?
The SEC Investor Advisory Committee’s recommendation, approved on December 4, 2025, describes AI as a strategic operational and competitive tool and notes integration challenges. It also says that “the disclosures currently remain uneven.” The recommendation discusses investor disclosure; it is not an SEC Commission rule or an investment conclusion about an individual company. In practice, uneven disclosure means comparisons need qualification: differences in reported measures, business models, reporting periods, or accounting treatment can make apparent performance differences hard to interpret.
Read AI risks alongside ordinary business risks
Review both AI-specific disclosures and the company’s broader risk factors. A useful assessment considers the controls described, the risks that remain, and whether the company’s adoption claims acknowledge those limits.
- Data and privacy: What customer or sensitive data may be used, where it may be exposed, and what confidentiality or training practices are described?
- Accuracy and oversight: How does the company address inaccurate or unreliable output, and where is human review needed?
- Security and intellectual property: What cybersecurity, ownership, licensing, or infringement risks does it identify?
- Regulation and workforce: Which regulatory or workforce issues could affect the product or its deployment? Rules vary by jurisdiction and use case, so verify current requirements for the relevant market rather than treating a general filing discussion as a current legal checklist.
- Substitution: Could customers use competing products, build a capability themselves, or decide the feature is not worth paying for?
Veritone’s FY2025 Form 10-K says internal AI may improve productivity but that “such productivity gains are not guaranteed.” It also identifies risks of sensitive-data exposure and inaccurate or unreliable output. The example shows why statements about potential productivity should be read together with the company’s stated limitations and controls.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Use a consistent comparison when evaluating alternatives
When comparing software companies, use the same questions for each one, while accounting for differences in business model, reporting period, and accounting treatment. Compare:
Best Value
- Product maturity and the importance of the workflow it supports.
- Paid adoption, pilot-to-production conversion, renewal, and expansion.
- Revenue model and whether AI-specific revenue or usage is disclosed.
- Development, inference, and infrastructure costs, along with reported margin or cash-flow effects.
- Access to customers, distribution, data, models, and compute—and reliance on outside providers.
- Reliability, governance, security, and regulatory exposure.
- Evidence across multiple periods compared with promotional claims and forward-looking plans.
Do not treat a disclosure gap as proof that a company’s AI business is weaker. It does mean the comparison is less certain. Record what is reported, what is management’s expectation, and what remains unquantified; keep those categories separate when forming a view.
A practical filing checklist
For each company, answer these questions from its filings and operating disclosures:
- Which named product or customer workflow uses AI, and what evidence shows customers rely on it?
- What share of customers, seats, or usage is paid, and how does adoption relate to renewal, expansion, or revenue per customer?
- Is AI revenue separately reported? If not, note that it has not been quantified rather than estimating it from broad growth.
- Do pilots convert to production, and does the company describe the time, cost, or support needed to make that conversion?
- What are the compute, model licensing, data, and support costs? Are margins or cash generation described as changing?
- Which third parties control essential models, cloud infrastructure, distribution, or data access, and how might changed costs or terms matter?
- What controls address security, privacy, copyright, and output quality, and what risks remain?
- Does management connect AI spending to measurable outcomes and explain when results fall short of expectations?
The SEC Investor Advisory Committee framed the challenge this way: “AI-related information can be material and of interest to investors, but the issue is how to sort the relevant information into operational categories that help inform investment decisions.” That is a useful discipline for reading filings: organize claims by product value, paid adoption, economics, dependencies, execution, and risk, rather than treating every mention of AI as equally informative.
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.
