A phrase screen for research candidates
This page looks for AI-related wording in company descriptions. It does not call an AI model, examine filings, measure AI-derived revenue or estimate a stock's future return. The first stage requests up to 50 actively listed candidates from NASDAQ, NYSE and AMEX using the selected sector, market-cap, price and provider-volume filters. Returned exchange identifiers are checked again, and the company profile must confirm USD.
The result depends on which candidates the provider returns first. A company outside that response cannot appear, regardless of its involvement in AI. Funds and ETFs are excluded from the request. The volume filter uses the provider's volume field; it is not an average-volume or executable-liquidity guarantee. Selecting Any removes that numerical filter. Changes apply only when Find AI Stocks is pressed.
What the matching count measures
The fixed phrases are: artificial intelligence, machine learning, deep learning, neural network, generative AI, large language model, natural language processing, computer vision, robotics, data science, AI solutions, AI platform, AI-powered, AI-driven, autonomous, automation, predictive analytics, cognitive computing, edge AI, and AI infrastructure. Matching ignores case and counts each phrase once if it occurs anywhere in the description. It does not understand negation, context or how material a product is to revenue.
For example, a hypothetical description saying "industrial automation and robotics" receives two matches. A description saying "artificial intelligence" three times receives one. The first count is larger, but that does not establish a stronger AI business. A phrase such as "does not develop artificial intelligence" still matches. Open the matching terms in each row to inspect the full provider description and test that interpretation yourself.
Read the coverage summary before the ranking
After a run, the summary separates returned candidates, excluded listings, unavailable profiles, unconfirmed or non-USD currencies, and matching descriptions. Missing quotes or analyst fields do not remove a matched company: its row remains with unavailable values. Failed requests produce a visible incomplete-result message. An empty table therefore does not prove that no companies satisfy a broader AI investment theme.
Rows are ordered by phrase count, then provider analyst label, then available target/price spread. These are display rules, not empirically validated weights. Among otherwise tied rows, a known negative spread appears before an unavailable spread; missing data is not treated as zero. The table does not assign a purchase recommendation.
Separate a target comparison from a return forecast
If a hypothetical available quote is $80 and the consensus target is $100, the displayed spread is (100 / 80 − 1) × 100 = 25%. If the quote rises to $95 with the same target, the spread becomes about 5.26%. The change can come entirely from the denominator, without any analyst revising a view. A missing or nonpositive price or target leaves the spread unavailable.
Analyst consensus summarizes opinions. The target endpoint reports an aggregate forecast, not a guaranteed future transaction price. This page cannot establish a shared target date, horizon or probability of success. Check the original research, relevant security and currency before using the comparison. The spread excludes dividends, fees and taxes and does not quantify downside risk.
Data timing and next research steps
The current app caches screener data for two minutes, profiles for five minutes, batch quotes for 30 seconds, and analyst fields for one hour. Those durations do not eliminate provider delay. The page runs on opening or an explicit screening action, without background polling. Quote times are displayed when supplied; descriptions and targets can have different ages. Reset clears the result and cancels acceptance of an older run.
Use a matched phrase to form a question for further research. Verify the business claim in company filings, distinguish product availability from meaningful revenue, and inspect valuation and risks separately. Pineify's coding agent can help draft a chart indicator, but this page does not connect a broker, execute trades or validate a trading strategy.
Sources and method
- FMP company screener
- FMP profile endpoint reference
- FMP analyst grades summary
- FMP price target consensus
Method reviewed October 21, 2026. The descriptions and prices in the examples are hypothetical.