Company examples

Agentic AI companies: products, roles and what to check

Agentic AI companies build models, tools or infrastructure for software that can take multiple steps toward a task. Autonomy still depends on permissions, tools and human approval. The Microsoft fiscal 2025 annual report describes Copilot Studio and GitHub Copilot; Salesforce fiscal 2026 results describe Agentforce. Those are product examples, not a ranking, and agent features are only part of each reported business. A working demo does not set a valuation.

Capability boundary
Check what the agent may do, which actions require approval, and how failures are detected.
Market Taxonomy
Frontier model labs, agent orchestration platforms, enterprise vertical agents, and compute providers.
Research Tooling
Use company disclosures and dated financial data; an AI summary is not an audit opinion.

Start with the output, not the AI label

AI trading products do different jobs. Decide whether you need research, screening, chart review, code generation, parameter testing, signals, or journaling. Then judge the output you can inspect and the limits the product states.

Company examples and their different roles

These are examples with public product disclosures, not a ranking or a complete industry list. Their overall revenue includes businesses beyond agents.

  • Microsoft (MSFT): its fiscal 2025 annual report describes Copilot Studio for creating agents and GitHub Copilot for software work. Assess cloud and software exposure separately from agent-specific sales.
  • Salesforce (CRM): its fiscal 2026 results report Agentforce as an enterprise software offering. Compare agent adoption with renewal, implementation and service costs.
  • Model labs, infrastructure suppliers and application vendors solve different problems. Compare businesses within a role before comparing margins or customer adoption.

Evidence to request before calling a company a leader

Product demos, customer counts, bookings, annual recurring revenue and recognized revenue answer different questions. Record the metric definition, measurement date and denominator.

  • Deployment: distinguish a signed deal, a pilot and a paying production workload.
  • Reliability: inspect approval controls, evaluation scope, incidents and recovery costs.
  • Economics: compare reported gross margin with inference, support and implementation costs; scale does not guarantee margin expansion.
  • Durability: test whether customers can switch suppliers or reproduce a feature; proprietary data alone does not establish pricing power.

Research a public company with Pineify

Start with a public ticker and one reporting period. Ask the Finance Agent for available statements and the earnings transcript, then match any agent-product claim to the filing. For Microsoft and Salesforce, separate agent features from cloud, software and service revenue. Build a valuation only with assumptions you can point to; the AI output is not an audit or a price target.

Frequently asked questions

Use AI as a reviewable tool

This material is for information and product education. It is not investment advice. Trading can result in losses, and historical tests do not predict future results. Review generated code, assumptions, and data before using them in a trading workflow.

Research AI companies