askyourdatabase.com

AskYourDatabase does not publish plan names, prices, or numeric usage limits on the pages available here. What the site does make clear is the decision structure: Desktop and Website Chatbot are separate products, each with its own plan considerations, and private deployment is handled as an enterprise discussion rather than a self-serve tier. To choose, start from who needs the answers and where they will ask questions, then review plans and usage allowances on the pricing page for current figures.

The three paths, and who each one fits

The site frames the choice around "the people who need answers and where they will ask questions."

Two details in that table are easy to misread. First, "local connections" in Desktop do not mean offline AI — the site states that questions and necessary response data can reach a model provider through the AI gateway. Second, hosting the application yourself does not automatically remove external model calls; only a locally hosted model changes that, and it is scoped separately.

What "usage allowance" actually controls

The pricing page is the only place with current plan and allowance figures, and the site directs you there to "review plans and usage allowances." Because no numbers are published on the pages available here, treat the allowance question as something to verify directly rather than assume.

What you can reason about before you look:

• Desktop allowances govern how much analysis one person or a small team can run from a connected machine. Since each user connects from their own computer with database access, the number of seats and the query volume both matter.
• Website Chatbot allowances govern usage by your end users, not your team. A chatbot embedded in a customer portal can generate far more questions than an internal analyst would, so the allowance that fits Desktop may be the wrong shape for an embedded deployment.
• Private deployment is not a usage tier at all — it is a scoping conversation covering hosting, model endpoint, network access, and operational responsibilities.

If your main constraint is cost per question rather than features, the site also offers a direct comparison between AskYourDatabase and BlazeSQL, which is worth reading alongside the pricing page.

How to decide in practice

1. Name the asker. Internal analyst or operations user → Desktop. Your product's end user → Website Chatbot. Neither, because you have hosting or data-handling constraints → enterprise discussion.
2. Check the allowance against realistic volume. For Desktop, estimate questions per person per day. For an embedded chatbot, estimate questions per end user per month, then multiply.
3. Confirm the data boundary before you commit. For Desktop, confirm which machine holds database access and that the AI gateway path is acceptable. For the chatbot, confirm authentication and per-user data access, then test that each user can retrieve only the records they are allowed to see.
4. Compare plans on the pricing page, and compare products if cost is the deciding factor. The AskYourDatabase vs. BlazeSQL comparison covers the product-level trade-off.

Getting accurate answers once you have picked a plan

Plan choice and answer quality are separate problems, and the site treats the second one as a setup discipline rather than a feature:

• Connect approved data. Use a database account limited to the tables or reporting views you need, and start with read-only permissions for analysis. Connection guides exist for PostgreSQL, MySQL, SQL Server, BigQuery, and other supported engines.
• Define your business terms. Revenue, active customers, date ranges, and table relationships need explicit definitions, plus verified SQL examples and schema context, so a familiar business phrase maps to the intended data.
• Check before you explore. Compare the first answers against a trusted report and inspect filters, joins, and totals before relying on them.

This matters for plan sizing too: a deployment where business terms are undefined will burn questions on clarifying and correcting, which shows up as higher usage against the same allowance.

What to verify on the pricing page

Because the available pages do not state prices, plan names, or allowance numbers, confirm these directly before deciding:

• Current plan names and what each includes
• The usage allowance attached to each plan, and what happens when it is exceeded
• Whether Desktop and Website Chatbot are billed separately, since they are separate products
• What the enterprise path requires in terms of hosting, model endpoint, and network access
• Whether a locally hosted model is available and how it is scoped

The site's own framing is the most reliable starting point: choose based on who needs answers and where they will ask, then review plans and usage allowances for the numbers.

AskYourDatabase does not guarantee that your database data stays entirely on your machine. According to the site's own product walkthrough, "Local connections do not mean offline AI: questions and necessary response data can reach a model provider through the AI gateway." So if your question is whether connecting from your own computer keeps everything local, the answer is no — the AI processing step can involve an external model provider. If you need stronger control, the site points to private deployment as an enterprise discussion, and notes that even self-hosting the application can still involve external model API calls. A locally hosted model is described as a separate deployment scope.

What actually leaves your environment

The key distinction is between where your database lives and where the AI runs. AskYourDatabase Desktop connects from your computer and requires your computer to have database access. That part is local. But the natural-language question you type, plus the data needed to produce a response, can travel to a model provider through the AI gateway.

In practice this means:

• Your question text is sent for interpretation.
• Necessary response data — the values needed to generate an answer, SQL, or chart — can also reach the model provider.
• Your database connection itself stays on your side; the app reaches your database from your machine.

The site does not specify exactly which rows or columns are transmitted, so treat "necessary response data" as the operative phrase and test with non-sensitive data first if that boundary matters to you.

The three deployment paths and what each implies

The product walkthrough separates the options by who needs answers and where they will ask:

The site explicitly frames private deployment as an enterprise discussion covering application hosting, the model endpoint, network access, and operational responsibilities. That is the path to take if the default gateway behavior is unacceptable for your data.

What to verify before you roll out

The site's own guidance for the Website Chatbot path is the most concrete privacy instruction available: plan authentication and user-level data access before rollout, and test that each user can retrieve only the records they are allowed to see. That is an access-control check, not a data-residency guarantee, but it is the step the vendor tells you to do.

For any deployment, a practical sequence:

1. Connect with a limited account. Use a database account restricted to the tables or reporting views you need, and start with read-only permissions for analysis.
2. Define your business terms. Add verified SQL examples and schema context so a familiar phrase maps to the intended data — this reduces the chance of the model pulling broader data than intended.
3. Check the first answers against a trusted report. Inspect filters, joins, and totals before relying on output.
4. Confirm the model endpoint question with the vendor if you need to know precisely what is transmitted and to whom. The public material describes the gateway behavior but does not enumerate the data fields.

Bottom line

If "private" means "nothing leaves my network," AskYourDatabase's standard Desktop and Website Chatbot paths do not meet that bar as described — the AI gateway can send questions and necessary response data to a model provider. If you need hosting or data-handling terms set explicitly, private deployment is the route the vendor offers, with the caveat that self-hosting the app can still involve external model API calls unless you separately scope a locally hosted model.

AskYourDatabase is an AI-powered tool that turns natural-language questions about a SQL database into SQL, answers, and charts, so you can query data without writing queries by hand. It supports PostgreSQL, MySQL, SQL Server, BigQuery, Snowflake, and other engines, and it comes in two separate products: a Desktop app for your own team's analysis, and an embeddable Website Chatbot for your users. It fits teams and product builders who need answers from a database but don't want every question to go through a SQL developer.

The core idea

You ask a business question in plain language. AskYourDatabase translates it into SQL, runs it, and returns an answer — plus a chart when you want one. The point is to skip the manual query-writing step while still letting you inspect what the AI actually did.

Two products, two audiences

The choice depends on who needs answers and where they'll ask.

Private deployment is a separate, enterprise-level discussion rather than a self-serve option.

How it works, step by step

The documented flow runs from a business question to a checked answer:

1. Connect approved data. Use a database account limited to the tables or reporting views you need. For analysis, start with read-only permissions. Follow the connection guide for your engine.
2. Explain your business terms. Define what "revenue," "active customers," date ranges, and table relationships mean. Add verified SQL examples and schema context so a familiar phrase maps to the intended data.
3. Check, then explore. Compare the first answers against a trusted report. Inspect filters, joins, and totals before relying on the results, then ask follow-ups or request a chart.

The "check the answers" step is built into the product's own framing — it expects you to verify, not just accept.

What to know before you commit

• Local connections don't mean offline AI. Even when you connect from your own computer, questions and necessary response data can reach a model provider through the AI gateway. If that matters to you, treat hosting and model endpoint as explicit decisions.
• Self-hosting still involves external calls. Hosting the application yourself can still involve external model API calls; a locally hosted model is a separate deployment scope.
• Access control is your job for the chatbot. Before rollout, test that each user can retrieve only the records they're allowed to see.
• Pricing and plan limits are listed on the site's pricing page; check current plans and usage allowances there rather than assuming a tier.

When it's a good fit

AskYourDatabase makes sense if you have a supported SQL database, a set of recurring business questions, and either a team that wants self-serve analysis (Desktop) or users who should query data inside your product (Website Chatbot). It's less suited to situations where no external model calls are permitted, or where you need a fully local model without a separate deployment effort.

Use Desktop if you are an analyst, operations person, or business user who wants to ask your own database questions from a Windows or macOS app. Use the Website Chatbot if you are a product team that wants to put database questions inside a website or customer portal for your users. They are separate products, so the choice depends on who needs the answers and where they will ask. If you have hosting or data-handling requirements, private deployment is a separate enterprise discussion.

The core difference: who asks, and where

AskYourDatabase turns questions about your database into SQL, answers, and charts. Both products do that, but they sit in different places.

Choose Desktop if you want to analyze data yourself

Desktop is the fit when the person asking the question is also the person who needs the answer. You connect from your computer, ask a question, inspect the SQL and results, then ask a follow-up or request a chart.

Two conditions matter before you start:

• Your computer needs database access. Local connections do not mean offline AI. Questions and necessary response data can reach a model provider through the AI gateway, so "local" describes the connection, not the whole processing path.
• Start with read-only permissions. For analysis, use a database account limited to the tables or reporting views you need.

A practical workflow, based on how the product describes its own process:

1. Connect approved data. Use a limited database account and follow the connection guide for PostgreSQL, MySQL, SQL Server, BigQuery, or another supported engine.
2. Explain your business terms. Define revenue, active customers, date ranges, and table relationships. Add verified SQL examples and schema context so a familiar business phrase maps to the intended data.
3. Check, then explore. Compare the first answers with a trusted report. Inspect filters, joins, and totals before relying on the result.

The common failure point is step 3. If "revenue" is not defined the same way in your schema as in your head, the SQL will be valid but the answer will be wrong. Verified SQL examples and schema context are what close that gap.

Choose the Website Chatbot if your users need answers

The Website Chatbot is for product teams adding database questions to a website or customer portal. You configure a chatbot and connect it to approved data, then integrate the chat interface into your application.

The setup work is different from Desktop in one important way: you are exposing data to people who are not you. The product guidance is explicit here — plan authentication and user-level data access before rollout, and test that each user can retrieve only the records they are allowed to see.

That test is the real acceptance criterion. A chatbot that answers correctly but shows one customer another customer's rows has failed, regardless of how good the SQL is.

What about private deployment?

Private deployment is not a third option you pick from the same menu. It is an enterprise discussion for organizations with hosting or data-handling requirements. You agree on application hosting, the model endpoint, network access, and operational responsibilities with the enterprise team.

Two limits to keep in mind:

• Hosting the application yourself can still involve external model API calls.
• A locally hosted model is a separate deployment scope.

So "private" does not automatically mean "no data leaves your environment." Confirm which parts stay inside before you treat it as a compliance answer.

How to decide

Ask these in order:

1. Who needs the answer? Your own team → Desktop. Your customers or portal users → Website Chatbot.
2. Where will they ask? In a desktop app → Desktop. Inside your product → Website Chatbot.
3. Can the asker reach the database directly? Desktop requires database access from the user's computer. If that is not possible or not appropriate, the Website Chatbot model fits better.
4. Do you need per-user access control? If yes, budget for authentication and user-level data access planning before rollout — this is not optional for the chatbot.
5. Do you have hosting or data-handling constraints? Then start an enterprise conversation about private deployment, and clarify the model endpoint and whether a locally hosted model is in scope.

If you are still weighing cost and usage, review plans and usage allowances, or compare AskYourDatabase and BlazeSQL.

AskYourDatabase turns natural-language questions into SQL, answers, and charts across PostgreSQL, MySQL, SQL Server, BigQuery, Snowflake, and other supported engines. To get accurate results, set it up in three stages: connect a restricted, read-only database account; define your business terms and schema context; then validate the first answers against a trusted report before relying on them. The setup below applies to the Desktop app and, with extra authentication work, to an embedded Website Chatbot.

1. Connect approved data with a limited account

The connection is the foundation. If the account can reach every table, the model can too.

• Use a database account scoped to what you need. Limit it to the specific tables or reporting views relevant to your questions rather than granting broad access.
• Start read-only for analysis. Read-only permissions prevent a generated query from modifying data while you're still testing.
• Follow the connection guide for your engine. AskYourDatabase documents setup for PostgreSQL, MySQL, SQL Server, BigQuery, and other supported engines.
• Confirm network access. The Desktop app connects from your computer, so that machine needs database access.

One point that surprises people: a local database connection does not mean offline AI. According to the site, questions and necessary response data can reach a model provider through the AI gateway. If your data cannot leave your environment, that is a separate deployment conversation — a locally hosted model is its own scope, and even self-hosting the application can still involve external model API calls.

Desktop vs. Website Chatbot connection differences

These are separate products, so pick based on who needs answers and where they will ask. For the chatbot, plan authentication and per-user data access before you roll out, and test that each user can retrieve only the records they are allowed to see.

2. Define business terms and schema context

This is the step most people skip, and it is where accuracy is won or lost. A phrase like "revenue" means nothing to a model until you say which column, which filter, and which date logic it maps to.

• Define your core terms. Spell out what revenue, active customers, and date ranges mean in your data.
• Describe table relationships. Explain how tables join so the model picks the right path instead of guessing.
• Add verified SQL examples. Provide SQL you trust for common questions so a familiar business phrase maps to the intended data.
• Include schema context. Give enough structure that the model understands what each field represents.

The goal is that a business question maps to the same data a knowledgeable analyst would use.

3. Check the first answers before you explore

Treat early output as a draft, not a result.

1. Compare the first answers with a trusted report. Pick a question you already know the answer to.
2. Inspect the filters, joins, and totals. A correct-looking number can come from a wrong join or a missing filter.
3. Only then ask follow-ups or request a chart. Once the logic checks out, the conversational flow — ask, inspect the SQL and results, follow up, visualize — becomes reliable.

If an answer is wrong, the fix is usually upstream: tighten the account scope, clarify a business term, or add a verified SQL example.

Common pitfalls

• Granting more access than needed. A broad account lets the model query tables that shouldn't be in scope.
• Assuming local means private. Local connections can still send questions and response data to a model provider.
• Skipping the trusted-report check. Without it, you have no baseline for whether the SQL is right.
• Rolling out the chatbot without access testing. Verify per-user data access before launch, not after.

What to do next

Review the plans and usage allowances on the pricing page, or compare AskYourDatabase with BlazeSQL if you're weighing options. For hosting or data-handling requirements, private deployment is handled as an enterprise discussion covering application hosting, the model endpoint, network access, and operational responsibilities.