A grounded agent vs a general model
Why wiring a general model to your site is not enough, and why grounding is the line between an asset and a liability.
Side by side
| Criterion | General model | Grounded agent |
|---|---|---|
| Source of the answer | Training data | Your documents, at question time |
| Facts specific to your company | Invented | Retrieved |
| Auditability | No traceable source | Every answer attributable to a document |
| General knowledge | Broad | Deliberately bounded to your domain |
| Updating a fact | Wait for a new model release | Edit the document |
Choose the first when: General model
- The use is creative or general: drafting copy, brainstorming, translation.
Choose the second when: Grounded agent
- The agent speaks for your company in front of a customer.
- Wrong answers carry a cost: financial, regulatory, or reputational.
- You need to be able to explain later why the agent said what it said.
Frequently asked
- Can a grounded agent answer general questions?
- You can allow it, but redirecting politely to its domain is usually better. A customer asking a shipping company's agent about the weather is not waiting on a forecast.
Other comparisons
AI agent vs traditional chatbot
The difference is structural, not cognitive: a chatbot follows a fixed decision tree, an agent retrieves, reasons, and acts.
Building an AI agent in-house vs using a platform
The prototype takes a week. What takes a year is everything around it: channels, permissions, audit, and operations.
Designing the handoff from agent to human
When an agent should hand off, what must travel with it, and why a zero handoff rate is a bad sign rather than a good one.
Running an AI agent on WhatsApp Business
What running an agent on WhatsApp actually requires: the account, the 24-hour window, templates, and human handoff.