Research
Generative & Agentic AI
Generative AI creates content and answers. Agentic AI goes further and takes action. This area covers the models, assistants, agents and retrieval systems that enterprises are putting into production.
Topics in Generative & Agentic AI
Foundation Models & LLMs
The large models behind generative AI and how to choose among them.
Overview
Large language models and other foundation models are now available through cloud APIs, embedded in business software and as open-weight models that organizations can run themselves. Copilots and assistants bring them into daily work. Agents connect them to tools so they can complete multi-step tasks.
Value depends on grounding these systems in accurate, permitted information and on controlling what they are allowed to do. Retrieval-augmented generation (RAG) and well-governed knowledge sources are often more important than the choice of model.
New report: Governing Agentic AI 2026
A 67-page playbook for enterprise AI agents covering autonomy levels, risk tiering, agent identity, MCP and A2A connectivity, human oversight and audit trails, with 10 ready-to-use templates.
What leaders need to get right
- Model strategy. Match models to tasks on capability, cost, latency and data terms.
- Grounding. Connect AI to curated, access-controlled knowledge.
- Action controls. Limit what agents can do and require approval for high-impact steps.
- Measurement. Track quality, adoption and cost from the first deployment.
Research report
Put guardrails around your AI agents before they scale
Get the Agent Autonomy Scale, the Six-Plane Agent Control Model, an agent registry template, a pre-deployment readiness checklist and a model agent governance policy.
Featured research report
Governing Agentic AI 2026
67-page playbook for governing enterprise AI agents: autonomy levels, risk tiers, agent identity, MCP controls and human oversight. Includes 10 ready-to-use templates.
Questions leaders should ask
- Which models are approved for which kinds of data?
- How do we ground AI answers in approved sources?
- What actions can our AI agents take without a human?
- How do we measure answer quality over time?
Other research areas
Related research reports
In-depth reports with ready-to-use templates.
- Governing Agentic AI 2026$59.99 PDF report
- AI-Ready Data$59.99 PDF report
Related research reports
Reports connected to this topic.
Latest insights on Generative & Agentic AI
SageMaker Training Sessions: Six Governance Tests Before Deployment
AWS now offers serverless training sessions with custom post-training loops. Here is what enterprise teams must test for quality, security, cost and reproducibility.
RAG Access Revocation: The Hidden Gap Between Source Permissions and AI Answers
Permission changes do not instantly propagate through RAG connectors, indexes and caches. Learn how to measure access revocation and test AI search security.
How to Evaluate Enterprise AI Agents Before Giving Them Autonomy
A practical enterprise AI agent evaluation framework covering business outcomes, prohibited actions, multi-step tests, repeated trials, approvals, pilot gates and continuous assurance.