QueryMind

All of your company data, for AI.
The only way is to connect
the database directly.

AX that analyzes company data with AI
and brings it into everyday work.

“How did the sales-order amount change from January to February?”

February's sales-order amount was KRW 34,638,400,
an 8.3% decrease from January.

Synthetic ERP data demo · query conditions included

SQL for your question

Connect the AI you already use
with your existing databases
ChatGPT
Claude
PostgreSQL
SQL Server
Oracle

Bring company data into
one workflow through AI.

ChatGPT and Claude do not know your company data. Upload a file and they know only that file, as it was on the day you uploaded it.

QueryMind is an AX product that connects company data to ChatGPT and Claude and helps teams bring AI into everyday business work. Check the state of the business and compare changes across periods right in the chat window. Review the query conditions and SQL behind each answer, then ask your next question.

Turns the question typed into the AI chat into SQL and queries company data
ChatGPTClaudeQueryMind Workspace
QueryMindQuery criteria · authorization check · SQL execution · results with conditions
ERP
Academic · Administrative
Business DB
Cloud DB
Existing databases · read-only queries

Ask where you already work

Ask questions of company data in ChatGPT, Claude, or your internal business screens.

Compare on the same basis

Business definitions are applied so you can see the differences between periods and the breakdown by item.

Decide with the evidence in view

Review the query conditions and SQL used in the answer. Business databases are queried read-only, and people make the final decisions.

For the same question,
the same business criteria apply.

Before querying ‘sales-order amount’, QueryMind first settles which amounts to total and which date to go by. It checks query authorization, runs the query, and returns the conditions applied to the answer alongside the result.

“Please compare the sales-order amounts for January and February.”
Business definitions applied in the demoSales-order amount · order date · login company
SUM(DomAmt) · OrderDate
SEMANTIC LAYER

Turns business terms into query criteria

Decides what to aggregate, over which period, and in which unit. Everyday business terms are turned into SQL query conditions, and when the criteria are unclear, the user is asked to clarify.

See a query example
SELECT
Policy checkAuthorization · read-only · execution conditions
RESULT + AUDIT
CONTROLLED EXECUTION

Checks authorization before the query, keeps a record after it

Authorization and execution conditions are checked before each query, and the question, SQL, and result are kept as an audit record afterward. In Enterprise environments, the configuration is reviewed against your existing security policies.

See security and deployment options

In the AI you already use, talk with your company's data.

Choose the AI you want to use and connect QueryMind. For now, this uses each service's custom connection feature.

ChatGPT

Ask questions of company data in the ChatGPT chat window and see comparisons across periods. You can change the conditions and ask again in the same conversation.

Claude

Check company data on the same business criteria in the Claude chat window. Use the query conditions behind each answer to shape your next question.

Whether the custom connection feature is available depends on your AI account and organization settings. Press the Get started button to see the connection steps and a first question to ask.

  1. 01 · Choose your AIStart in ChatGPT or Claude.
  2. 02 · Connect QueryMindFollow the steps to add QueryMind, then log in to confirm your data access permissions.
  3. 03 · Ask your first business questionAsk a question of the connected data and review the result with its query conditions.
Internal-only screens · on-premises · restricted networks

QueryMind Enterprise is configured to fit your organization's data and security environment.

Enterprise consultation ↗

One question returns
both the numbers and the query conditions.

“Compare the sales-order amounts for January and February, and show the aggregation conditions too.”

An actual MCP demo run on synthetic ERP data. The sales-order amounts for January and February 2026 were compared using the same metric and conditions.

Sales-order amount, January and February 2026
Difference between the two months−KRW 3,128,600 (−8.3%)

Actual MCP demo on 2026.11.08 · synthetic data · not an accuracy benchmark or a response-time measurement.

The same criteria are applied when comparing, too.

Scroll the table sideways to see all the query conditions and check items.

Query conditions used in the demo
Check itemBasis applied in the demoWhat it tells you
MetricSales-order amount · DomAmtWhat was totaled
PeriodOrder date · OrderDateWhich date was used as the basis
ScopeLogin company · stopped orders excludedWhich conditions were applied
RecordQuery tool · audit IDWhich query the answer came from

The conditions above are business definitions used only for this demo. In customer environments, each company's own business definitions apply.

Design background · the DART-SQL study and its source

What the research shows: even with the same model, results change with how the query is designed.

The DART-SQL study compared the effects of question rewriting, database context, and execution feedback. It shows that results vary with how the query is composed, even on the same model.

These are experimental results from an external paper. They do not represent QueryMind's own accuracy or a ranking of current models.

Improvement that persisted across evaluation sets

Execution accuracy (EX) of DAIL-SQL and DART-SQL

DAIL-SQLDART-SQL
Spider-dev
76.2
83.1
Spider-test
76.9
81.8
Spider-Realistic
68.9
79.3
Spider-DK
61.5
71.4
050100% EX
Mao et al. · ACL 2024 · Table 1 ↗Original figure ↗

Which components change the result

Spider-dev results as components are removed

Spider-dev · execution accuracy (EX, %)
DART-SQL · full configuration
83.1
Question rewriting removed
84.0
Context input removed
80.4
Execution correction removed
80.6
Rewriting + context removed
80.2
Rewriting + correction removed
80.4
Context + correction removed
79.9
DAIL-SQL · baseline
76.2
DAIL-SQL · Zero-shot
78.6
Mao et al. · ACL 2024 · Table 3 ↗Original figure ↗

GPT-3.5-turbo-16k-0613 · no self-consistency applied · EX compares the execution results of the generated SQL and the gold SQL. The configuration with question rewriting removed scored higher at 84.0%, and adding components does not always guarantee an improvement.

DART-SQL is external research. It does not indicate QueryMind's own accuracy or a ranking of current models.

A conversation that asks for the sales-order amount
and checks the query conditions as well.

This walks through asking for the January and February sales-order amounts and checking the difference and the aggregation conditions. See the actual demo alongside an easy-to-read summary of the results.

Answer confirmed in the demo Synthetic ERP data

January 2026
KRW 37,767,000
February 2026
KRW 34,638,400
Difference between the two months
−KRW 3,128,600 (−8.3%)

Sales-order amount DomAmt · order date OrderDate
Login company · stopped orders excluded

ClaudeQueryMind MCP
Actual demo of Claude calling the QueryMind MCP. Greetings and on-screen guidance around the edges were edited out.

QueryMind returns the data and the conditions applied, and Claude chooses the tools and writes the response.

Recording and editing notes

Recorded on 2026.11.08 with synthetic ERP data. Greetings, on-screen guidance around the edges, and waiting periods were edited out, while the question, tool calls, and responses were kept. The video length cannot be used to judge actual response time or speed differences.

The ERP demo uses synthetic data. The academic business screen is available under the ‘Academic screen’ tab.

Query and compare data with AI,
and start your team's AX.

Examples of the results you can get for each kind of business question. The actual scope depends on the connected data and your business definitions.

Sales · Purchasing

Sales-order and customer analysis

“Compare last month's and this month's sales-order amounts, and show the breakdown by customer too.”

What you can seeAmount by period · change and rate of change · breakdown by customer

Finance · Planning

Plan vs. actual review

“Compare each department's budget and spending on the same basis.”

What you can seeBudget and spending by department · comparison basis

IT · Security

Query evidence and access records

“I'd like to see the query conditions and the SQL that was run for this answer.”

What you can seeQuery conditions · executed SQL · audit records

Universities · Administration

Academic and operations status

“Show the number of enrolled students by department and the enrollment-status criteria used for the count.”

What you can seeHeadcount by department · enrollment-status criteria used for the count

These are usage examples, not quotes from customer testimonials or deployment records.

Start small, and use only as much as you need.

01 QueryMind Connect

Top-up credits · in preparation

Use 200 questions free each month, and top up credits when you need more.
No monthly subscription and no automatic payments.

200questions / month freeEach month, the 200 free questions are used first, then your purchased credits.

Starter

1,000 credits

₩49,000

KRW 49 per credit

A light start for business queries

Team

5,000 credits

₩199,000

KRW 39.8 per credit

Everyday use across a team

Business

15,000 credits

₩499,000

About KRW 33.3 per credit

Recurring data use
Set up QueryMind Connect

Choose your AI to see the current connection steps.

1 answered question = 1 creditNo charge for clarifications, refusals, failures, or queries that return no rowsPurchased credits do not expire

Credit top-ups are being prepared. Final payment details, including tax and refund terms, will be published before billing begins. Fees for AI services such as ChatGPT and Claude are separate.

02 QueryMind Desktop

Windows program

Install it on one PC and ask your company database questions. 14-day free trial.
Screens and answers are in Korean.

KRW 199,000 per monthAdd-on packages KRW 49,000 per month each (VAT included)
View details (Korean)
A real QueryMind Desktop screen in Korean: a question asked on sample data, answered with the result and its basis
A real screen recorded on sample data, shown in Korean.

03 Enterprise

Custom deployment

Deployed to fit your internal databases, security policies, network, and operating environment.

Custom quotePriced by integration scope and deployment requirements
  • Dedicated business screens for your company
  • Authorization, approval, and audit configuration
  • On-premises and restricted-network review
Ask about deployment
Q1Can I use it in ChatGPT and Claude?

Yes. Choose the AI you want to use from the Get started buttons above, and we will show you the connection steps. For now, this uses each service's custom connection feature, and the available features and permissions depend on your account and organization settings.

Q2Does it change the source database?

No. Business databases are queried read-only, and no data is changed. The accounts, settings, and audit records needed to run the service are kept separately from the business database.

Q3Is data sent to an external AI?

Yes. With Connect, the content needed for the answer, including your question and the query results, is sent to the AI you selected (ChatGPT or Claude). Organizations that restrict data from leaving their network should consider a separately scoped Enterprise or restricted-network deployment.

Q4When are credits deducted?

Each month, the 200 free questions are used first, and after that one credit is deducted for each question that receives a result. Clarifications, refusals, failures, and queries that return no rows are not charged. Top-up payments are currently being prepared.

Q5Do purchased credits expire?

Under the top-up plan being prepared, purchased credits do not expire. Refund conditions and how to request a refund will be announced once the payment terms are finalized.

Q6Can it be deployed in a restricted network?

This is reviewed as a separate deployment. After checking your network policy, server and model requirements, and data structure, we propose a deployment approach. The deployment method and billing structure differ from Connect.

Q7What is QueryMind?

QueryMind is an AX product that connects company data to ChatGPT and Claude and helps teams bring AI into everyday business work. It queries the current state of the business according to your access permissions and your company's business criteria, and compares changes across periods or the breakdown by item. You can also review the query conditions and SQL used in each answer.

Ask your first business question
of your company's data.

Tell us the business questions your team wants AI to answer and the data you want to connect. blueurban will walk you through the connection steps, deployment scope, and what to prepare.

The submission status will appear here. We will reply to the email you entered.

QueryMind Connect

Get started in ChatGPT

Start asking questions of your company data. Connect QueryMind and log in to see which data you can access.

    Copy the server address for a manual connection
    https://querymind.co.kr/mcp

    Enter QueryMind as the connection name. On the login screen, confirm your account and data access permissions.

    First question after connecting

    Tell me what data I can query in QueryMind.

    For now, we provide custom connection instructions. Listing status in the official plugin and connector directories is still being confirmed. If you do not see the connection menu, check your account or organization permissions.

    Need help connecting? Contact us ↗