How we test and score commercial software
Most "best CRM" content ranks tools by feature count. We think that's the wrong lens now. The question worth answering in 2026 isn't "does this tool have an AI feature" — nearly all of them do — it's "can an autonomous agent actually act inside this system without a human re-keying half of it."
What we actually check
- Data model, not marketing copy. Can the underlying object model represent an agent-initiated action (a price change, a renewal offer, a support resolution) as a first-class, auditable event — or does "AI" mean a chat widget bolted on top of a static record?
- API and permission depth. Can an external agent read and write the fields that matter (deal stage, entitlements, usage data) through a documented API, with scoped permissions, or is agent access an afterthought?
- Pricing and packaging fit. Does the vendor's own pricing model assume seat-based usage (a red flag for a tool meant to support consumption or outcome-based commercial models), or does it flex with usage and agent-driven activity?
- Real usage, not the sales demo. Where possible we use trial accounts and public documentation rather than relying on vendor briefings alone.
Who writes this
The Agentic Stack is written by the team behind FR Advisory's Agentic Commercial Model, a diagnostic framework used to benchmark how ready a commercial organisation — not just its tools — is for the agentic era. That's the lens every review here is written through.
How we make money
Some links on this site are affiliate links, meaning we may earn a commission if you sign up through them, at no extra cost to you. This never determines which tool we recommend — see our full disclosure for detail on how that separation works in practice.