Software intelligence
Discover.
Compare.
Choose well.
Independent research on the software layer: AI agents, automation, developer tools, customer support and the wider B2B SaaS market. Every tool scored on five axes, every comparison written to answer one buying question. No vendor gets to write their own entry.
- Tools indexed
- 39
- Categories
- 10
- Comparisons
- 11
- Curated stacks
- 6
Top of the index
Live
Composite of capability, usability, ecosystem, value and support, weighted and scored against category peers. Read the methodology
[ 01 ] Editors' picks
The tools we would actually put in a stack this quarter.
[ 02 ] The market map
Ten categories, each with a thesis about where it is heading.
AI Agent Platforms
Frameworks and no-code builders for software that plans, calls tools and completes multi-step work without a human in every loop.
Open category→Workflow Automation
Connective tissue for the stack — moving records, firing webhooks and stitching together the apps that were never designed to talk.
Open category→Customer Support Tech
Helpdesks, inboxes and AI resolution layers — the category where autonomous agents reached production first and hurt fastest when wrong.
Open category→AI Infrastructure & LLMOps
Evaluation, tracing, vector storage and serving — the unglamorous layer that decides whether an AI feature survives contact with real traffic.
Open category→AI Developer Tools
Coding agents, editors and review systems that write, refactor and reason about production codebases.
Open category→Data & Analytics
Warehouses, notebooks, product analytics and the semantic layers that keep a company arguing about the same numbers.
Open category→CRM & Revenue
Systems of record for pipeline, plus the enrichment and signal tooling that now sits in front of them.
Open category→Marketing & Growth
Content engines, lifecycle messaging, CDPs and the analytics that survived the collapse of third-party tracking.
Open category→Productivity & Collaboration
Docs, wikis, project tracking and the AI layers grafted onto all three.
Open category→AI Content & Writing
Long-form generation, brand voice control and the editorial workflow around machine-written drafts.
Open category→[ 03 ] Head-to-head
One buying question per comparison, answered plainly.
Updated 2026-07-19
n8n vs Zapier
Execution-based billing and self-hosting, or the widest connector catalogue in the market?
Updated 2026-06-28
LangGraph vs CrewAI
Explicit state machines, or role-based crews that orchestrate themselves?
Updated 2026-07-28
Cursor vs Claude Code
An AI-native editor, or a coding agent that lives in your terminal?
Updated 2026-07-21
Intercom vs Zendesk
Outcome-priced AI resolution, or the enterprise system of record?
Updated 2026-09-07
YourGPT vs Intercom
Omnichannel no-code you can take all the way, or Intercom Fin on a full suite?
Updated 2026-06-19
Make vs n8n
The most capable no-code canvas, or source-available automation you can host yourself?
[ 04 ] Stacks
The set we would buy, with a bill attached.
AI Agent Hub
First agent in production. You can replay a bad run, not argue about it.
Automation Stack
n8n for the volume. Zapier only for the connector nobody else has.
B2B SaaS starter
Under 15 people. Free tiers that actually work.
Ops, no engineers
Nothing here needs a repo, a deploy, or someone's afternoon in GitHub.
Revenue signal
Clay in front of the CRM. Cap credits and domains in week one.
Lean AI support
The agent takes first contact. Plain, Notion, PostHog and Linear do the rest.
[ 05 ] Insights
Where the market is going, and what it means for what you buy.
The agent pricing reckoning
Per-seat licensing assumed software helps a person work. Agents do the work. Four billing models are now fighting to replace the seat, and three of them will lose.
2026-08-12
Why most agent projects fail in month three
The demo works. The pilot works. Then something changes upstream, nobody notices for two weeks, and trust never comes back. A field guide to the failure that kills more agent projects than capability ever has.
2026-07-30
The consolidation nobody predicted
Unbundling was supposed to be the permanent condition of SaaS. Instead, buyers spent two years cutting vendor counts, and the tools that bundled well are winning categories they were not supposed to.
2026-07-18
Customer support after deflection stopped being the metric
Containment rates became contractual, and then immediately stopped being interesting. What replaced them says more about where support tooling is headed than any product roadmap.
2026-07-04
[ 06 ] Moving fastest
What is actually changing hands right now
Tools where buying behaviour shifted meaningfully this quarter — either because the product moved, the pricing moved, or the category around it did.
Missing something?
Tell us what to review next.
We take submissions from vendors and buyers alike. Nothing gets a paid placement, nothing gets a score it did not earn, and every listing states plainly what we could not verify.