Rule-based chatbots vs AI assistants
The old kind never surprises you and never helps much. The new kind helps a great deal and has to be constrained.
Rule-based bots fail the moment someone goes off-script
A decision tree can only handle the paths someone anticipated. Real questions arrive phrased in ways nobody predicted, and the bot answers with a menu — which is why customers learned to type 'agent' immediately.
- Menus that never contain the option the customer wants
- Maintenance burden growing with every new branch
- Customers bypassing the bot on sight
- No coverage for questions outside the tree
What rule-based bots do well
Complete predictability. For a fixed process with a known set of paths — a returns workflow, a triage routing form — a decision tree is transparent, testable, and cannot say anything unexpected. That is a real advantage where it applies.
What AI assistants do well
Handle phrasing nobody anticipated, cover a large body of content without anyone enumerating paths, and improve when you write a page rather than when you edit a flowchart.
Phrasing-independent
The same answer whatever words were used.
Scales with content
New pages extend coverage automatically.
No tree to maintain
Nobody enumerates the paths.
The cost of the newer approach
It must be constrained. A decision tree cannot invent; a language model can, which is why grounding, refusal, and citations are not optional extras but the price of the flexibility.
Common questions
- Is a rule-based bot ever the better choice?
- Yes — for a fixed process with a small number of known paths where predictability matters more than coverage. A structured returns workflow is a decision tree, not a question.
- Can I combine them?
- Commonly, yes: an assistant for open questions, with structured flows for defined processes like collecting details for a return.
From the blog
All posts- EvaluatingAI chatbots for documentation sites: six compared on citations, gaps and price (2026)Matter Chat, DocsBot, kapa.ai, Biel.ai, Docsie and Inkeep for docs sites: section-level citations, refusal behaviour, content-gap reports, docs-generator installs and whether you can buy without a sales call. Checked on the vendors' own pages.Read
- EvaluatingThe best AI chatbot for a Webflow site, compared honestly (2026)Matter Chat, Social Intents, Ultimo Bots, Chatling, WeblyChat and Chatbase on a Webflow site: Apps marketplace or custom code, the paid-site-plan gate, CMS coverage, citations, handoff and pricing. Checked on the vendors' own pages.Read
- EvaluatingChatbase alternatives for teams that need cited answers (2026)Matter Chat, Chatbase, Intercom Fin, Tidio Lyro, DocsBot and Chatling compared on citations, refusal behaviour, handoff and what the entry price meters. Checked on the vendors' own pages, with the test to run on all six.Read
Answer honestly. Capture the rest.
Point Matter Chat at your site and see what it can — and can't — answer. It's honest about both.
No credit card. 2 minute setup.
Every answer cites the source it came from. When there isn't one, it says so — and hands the visitor to you.
Installs on the tools you already run.