Blog · Published · Updated
Questions with visible SQL: how the workflow works
Short answer
Asking a database questions with AI makes sense when the definition of the measure and the SQL query that ran are visible under the result. In our demonstrator the system reads only four approved views with an account that cannot write, and a question costs $0.004400. At Syntalith the application costs from €3,000 net. At 300 data questions a month of 15 minutes each, it pays back in 1.5 months.
How we workDetails ↓
- 01Call30 minutes with an engineer, free
- 02PrototypeA working system on your process and data, free
- 03ImplementationTarget and fixed price in the contract; 50% at signing
- 04Acceptance50% once the agreed target is met
- analysis with an engineer and a free prototype
- €0
- team in ChatGPT or Claude, up to 10 people
- from €1,150
- automation, agent or application for one process
- from €3,000
We belong to the OpenAI and Anthropic partner programmes and learn through their official courses. What this means ↓
Author: Syntalith · 3 min read
Statistics
Data and AI: statistics 2026
1.5 mo.
payback of a €3,000 application at 300 data questions a month: €2,618 of work today, €661 with checking and the model. Scenario.
$0.004400
According to the Syntalith demonstrator (2026), an answer to a data question with the SQL query shown cost $0.004400.
Source: Syntalith, data questions with SQL demonstrator (2026) ↗
4
According to the Syntalith demonstrator (2026), the system reads only 4 approved database views and has no permission to write.
Source: Syntalith, data questions with SQL demonstrator (2026) ↗
42%
According to the European Commission (2026), 42% of EU SMEs and mid-caps already use AI or algorithmic management, mostly in a limited and experimental way.
Source: European Commission, Flash Eurobarometer 578 (2026) ↗
20.0%
According to Eurostat (2025), 20.0% of EU enterprises with at least 10 employees used AI.
Figures checked: 22 October 2026
What „sales this month” actually means
The question sounds simple, but the result depends on the period, the order status and the definition of sales: orders placed, invoiced or paid. So the screen shows the question, the definition of the measure and the query used for the calculation beside the result. An analyst can check whether the system counted what was asked.
In our demonstrator, the questions cover net revenue, margin, active customers and average order value. The system reads only four approved views and cannot write anything to the database. A question outside the approved measures gets no guessed answer. A recorded run of 12 questions cost $0.052794.
- 01Questionin plain language, within the approved measures
- 02Definitionthe measure with its period and status
- 03SQLa query against an approved view
- 04Resultthe figure with the query and the definition beside it
- 05Recordthe question, the definition and the query kept in the history
- 4 business questions in the public scope
- Synthetic data, a local tracing environment
- A validator and a read-only role, with no full security audit of the SQL
- No measurement for ambiguous definitions and wider schemas
What it costs and when it pays back
- Today
- 300 data questions × 15 min = 75 h (€2,618)
- With the system, a person checks the output in 25% of the time
- €655
- Model, 2 calls per case, Claude Sonnet 5.5
- €6
- Cost of one case
- €8.73 today, €2.20 with the system
- Saving
- €1,957 a month
Payback of €3,000
1.5 months
The volume, the time and the checking share are the scenario's assumptions. €34.90 an hour is the EU average hourly labour cost for 2025 according to Eurostat. Model: 4,000 input tokens and 400 output tokens per call at Anthropic's rates, at the ECB rate of 1.1205 USD per euro on 22 October 2026.
The count covers the work around a case: searching, reading, writing and handing over. What remains is the time to check the result, 25% of today's work in the scenario. Before the quote, we measure both times on 20 to 100 cases the team has already handled, in a free test that takes 5 working days.
The scenario assumes 300 questions a month that an analyst answers today in 15 minutes, by writing a query and checking the result. The model cost per question is small, and the payback depends on how many questions fit within the approved measures.
Approved measures before you widen the questions
Start by choosing approved measures and database views, and compare the query and the result with the analyst's answer before you widen the scope. A record of each run is easier with tracing in OpenAI Agents. If the questions are fixed, a ready report in a BI tool will answer faster and more cheaply. Questions in natural language make sense when the board asks varied questions about the same measures and an analyst spends time turning them into queries. Checking the figures in reports is covered in a management report with no made-up numbers.
Fixed questions about the same measures
a report in a BI tool
no build
Varied questions about approved measures
natural-language questions with visible SQL
from €3,000
Questions that need new definitions
an analyst with AI as a help
training
Where to start
The first scope covers a few measures with a clear definition and an analyst who compares the results with their own queries. To prepare: the definitions of the measures, the approved views and a read-only account. We build this kind of rollout as part of AI applications. For the free analysis, a description of one case from recent weeks and of the tools it passed through is enough. You do not need to send confidential documents, and you get a written plan within 2 working days.
Worked example · Scenario
How we count the payback: data questions
We run the same count for your data questions in the free analysis.
€1,815
a month for 12 h a week of report preparation
Payback: 1.9 mo.
The hours for each firm size and checking in 15% of the old time are our assumptions. €34.90 an hour is the EU average hourly labour cost for 2025 according to Eurostat, for companies with 10 or more employees. We measured the model cost in our report system: a commentary with a check of the figures costs about €0.0002, converted at the NBP rate of 21 October 2026 and the ECB rate of 1.1205 USD per euro on 22 October 2026.
See the full calculation →
