The client's question and the company's structured data: assortment, prices, addresses, opening hours and terms.
SUB–09 / DATA · EVERYDAY ASSISTANT · HUMAN CONTROL
Company Data / Website Consultant
Advises clients based on the company's approved data: the assortment, prices, addresses, opening hours and pre-order terms. If a fact is missing, it honestly says “I don't know”. Every question and refusal is recorded in the log.
A factual answer with a link to the source or an honest refusal recorded in the request log.
- The answer contains only an approved fact
- No data — it honestly says “I don't know”
- Updating the approved data updates subsequent answers
- The log shows frequent questions and data gaps
01Capabilities
What the subagent can do.
Every function works within set boundaries and returns a result that can be checked before the next action.
Free-form question
Understands phrasings like “how much does it cost”, “where are you located” and “when do you open” without a rigid script.
Search over company data
Works only with approved prices, assortment, addresses, opening hours and pre-order terms.
Factual answer
Returns the fact it found and keeps the link to the specific data source.
Honest “I don't know”
If there is no information, it does not improvise, suggests the available topics and records the refusal in the log.
A team, cloud, or local deployment environment connected to the company's approved data.
Data content, payments, order placement and disputed situations remain under human control.
02Process
Four steps within a familiar workflow.
Context remains traceable: each step shows its input, operation, result, and handoff to a human reviewer.
Receive the question
Receives the client's question in free form.
Find the data
Determines the topic and looks up the fact in the company's approved sources.
Prepare the answer
Returns a confirmed fact with a link to the data source.
Refuse honestly
If the data is missing, it says “I don't know” and saves the request in the log.
Structured resultA factual answer with a link to the source or an honest refusal recorded in the request log.
The result can be reviewed, adjusted, and passed to the next participant without copying a long conversation.
03Advantages
Both the value and its rationale are visible.
Each proposed outcome includes its verification method, limitations, and completion criteria.
Answers without inventions
Plausible improvisation is replaced by a rule: a fact from the data or an honest refusal.
Up-to-date information without retraining
Prices, addresses and opening hours change together with the update of the structured data file.
Transition to action
It can explain the pre-order terms and hand the client over to an agreed ordering scenario.
Demand log
The company sees which topics are asked most often and what information needs to be added.
- The answer contains only an approved fact
- No data — it honestly says “I don't know”
- Updating the approved data updates subsequent answers
- The log shows frequent questions and data gaps
04Roles and perimeter
Who benefits and under what conditions.
Each role receives a defined output; sources, integrations, and control points are agreed before implementation.
| Role | What they get |
|---|---|
| Retail location | Takes repetitive questions about price, address and opening hours off the staff. |
| Marketer | Gets the topics clients actually search for, from the request log. |
| Content administrator | Updates the data without retraining the model and without changing code. |
| Business owner | Protects client trust from invented prices and non-existent terms. |
Boundaries and responsibility
A useful workflow assistant does not become an autonomous decision-maker.
The completeness of the answers is determined by the quality of the company's data. The subagent does not accept payment, does not place an order on its own and does not answer topics outside the approved sources.
- Access rights and allowed sources are set before connection.
- Critical actions require human confirmation.
- Integrations and the deployment environment are selected after reviewing the process.
First step
Choose a ready-made solution or describe your process.
Pick one of the twelve agents and subagents, or fill in the questionnaire for custom development.