Reporting and analysis on your live database. No warehouse required.
Semavex connects to the PostgreSQL, SQL Server, MySQL or MariaDB database your business already runs on. Your team gets saved reports, scheduled delivery and AI explanations of the numbers, without building a warehouse or writing SQL.
- Read queries only
- Every query validated
- Index-aware
- Encrypted credentials
The reporting stack, without the stack
Most reporting projects start with pipelines, a warehouse and a modelling layer before anyone sees a number. Semavex reads the database you already run, so a report shows your data as it is when the report runs, not as of last night's load.
- Your database
- ETL pipelines
- Data warehouse
- Modelling layer
- BI tool
- Analyst queue
- Your database
- SemavexSemantic model, reports, analysis, delivery
Live by default
Every report reads your database when it runs. There is no copy to refresh and no sync to wait for.
Nothing to maintain
No pipelines to build, schedule or repair. Semavex reads your tables and their indexes where they already are.
One set of definitions
Measures and business terms live in a semantic model. Change a definition once and every report that uses it follows.
Use it on its own, or next to the BI tools you already have.
One platform, from your schema to your inbox
Vexon, the AI agent in Semavex, works with your team at every step, from the first model to the last delivery.
- Step 01
Model
Describe your business in plain language. Vexon drafts the semantic model: tables, relationships, measures and synonyms, each with a confidence level for you to review. Versions and rollback included.
- Step 02
Ask
Ask in plain language, in any language, or start from the questions Vexon suggests. Follow-up questions keep the scope of the conversation.
- Step 03
Analyze
Period comparisons, rolling periods, running totals, rankings and top-N, computed in your database. Vexon explains each result from exact statistics over every row in scope.
- Step 04
Report
Save any answer as a report. Group it with subtotals, format the numbers, add conditional formatting, or change it just by asking.
- Step 05
Deliver
Email reports on a schedule as PDF, CSV or Excel. On Enterprise, data alerts check your thresholds as often as every 15 minutes.
See it in action
Three short videos of the real product: a 90-second overview, connecting live databases to an AI-built semantic model, and Vexon turning questions into scheduled reports.
Semavex in 90 Seconds: Ask Your Live Data Anything, No SQL Required
Semavex is an AI-powered analytics platform that lets business teams query live transactional data in plain language. Ask a question, get an answer, refine it with follow-ups, and turn the analysis into reports, without writing SQL.
From a business question to a finished report
Describe the report you need. Vexon turns the request into a validated query plan, Semavex runs it on your database, and you get a formatted report back, with joins, aggregations and window functions included.
The whole reporting platform
Everything between your database and the people who need the numbers, in one place.
Saved reports
Group by any column with subtotals and a grand total computed on the server. Number formats, conditional formatting, and full CSV or PDF export.
Report chat
Change a saved report by asking: add a column, change the period, rename a metric. Changes to what the report covers wait for your confirmation.
Scheduled delivery
Visual CRON schedules with email delivery as PDF, CSV or Excel. Each run reads your live database.
Time and ranking analysis
Period comparisons, rolling periods such as the last 30 days, running totals, ranks, and top-N with the rest grouped together.
AI explanations
Vexon explains a result from exact statistics computed over the whole result, not a sample. A deep mode reads the rows in detail when you need more.
Guided drill-down
When a result shows a notable pattern, such as one row carrying most of the total or a sharp change, Vexon suggests the next question to ask.
Semantic layer
Measures, relationships, synonyms and business context, with versions, rollback, and draft or active states.
Roles and history
Custom roles built from fine-grained permissions, and a history of every agent request with its status, duration and AI cost.
Four database engines
PostgreSQL, SQL Server, MySQL and MariaDB, with index metadata read from each system catalog. Snowflake is coming soon.
Safe to point at the database your business runs on
Vexon never writes SQL. It produces a structured plan that Semavex checks against your semantic model and a whitelist, then turns into parameterized SQL with deterministic code. What reaches your database is a read query planned around your real indexes.
- Read queries only. We recommend a read-only database account.
- Index-aware plans: composite and partial indexes, cardinality-guided joins
- A time limit on every query
- By default, reports never block your application's writes
- Credentials encrypted with AES-256-GCM, keys in Azure Key Vault
- Every tenant in its own isolated schema
Set it. Forget it. Find it in your inbox.
Build recurring schedules with a visual CRON editor and choose the format: PDF, CSV or Excel. Each run reads your live database, so the numbers are current. Failed runs are retried, and if a report cannot run, its owner gets an email saying why.
- Visual CRON schedule builder
- PDF, CSV or Excel attachment
- Automatic retries, and an email when a run is skipped
Questions teams ask before connecting
Is Semavex a text-to-SQL tool?
No. A text-to-SQL tool turns one question into one query. Semavex is a reporting and analytics platform: a semantic model of your business, saved reports with subtotals and formatting, scheduled delivery, AI explanations of results and, on Enterprise, data alerts. Its AI agent never writes SQL. Vexon produces a structured query plan, which Semavex validates and translates into SQL.
Do I need a data warehouse?
No. Semavex connects directly to your PostgreSQL, SQL Server, MySQL or MariaDB database, so there are no pipelines to build and every report reads the data as it is when it runs. If you prefer to keep reporting off your primary server, connect Semavex to a read replica instead.
Is it safe to run on our production database?
Semavex only sends read queries, built by deterministic code from a plan that has passed a whitelist check, with every filter value passed as a parameter. Queries are planned around your indexes and run with a time limit, and by default they never block your application's writes. We recommend connecting with a read-only database account.
Does Semavex replace Power BI, Tableau or Looker?
That depends on what you use them for. Teams without a BI stack use Semavex as their reporting platform: saved reports, scheduled delivery and analysis on live data. Teams that already run a BI tool use Semavex next to it, for the ad hoc questions and recurring reports that would otherwise wait in an analyst's queue.
Does the AI see our data?
Vexon never connects to your database and never runs queries. It works from your semantic model, which can include sample values from columns not marked as sensitive. To explain a result, Semavex shares statistics and a sample of rows, and shows you what will be shared and what it costs before it runs. Drill-down suggestions use a few rows of the result.
Which databases does Semavex support?
PostgreSQL, SQL Server, MySQL and MariaDB, each read with its own index metadata. Snowflake is coming soon.
Simple, transparent pricing
A platform subscription that covers infrastructure, support, and maintenance, plus an AI credit wallet your team controls.
- Up to 10 users
- Vexon AI agent suite: semantic model builder, data agent & reporting agent
- Up to 10 data sources
- 20 report templates
- 20 report schedules
- Scheduled report delivery
- Saved reports & exports
- Per-user credit allocation
- Auto top-up wallet
- 1 annual schema health review
- 1 annual report audit
- 1 onboarding session (year one)
- Shared cloud infrastructure
- Priority email support · 24h SLA
- 99.5% uptime SLA
- Negotiable user seats
- Negotiable data source limit
- Dedicated Azure instance
- VNet isolation
- Data alerts: threshold checks on your live database, as often as every 15 minutes
- Quarterly business reviews
- 4 onboarding sessions (year one)
- Dedicated Slack support channel · 8h SLA
- Direct escalation path for critical incidents
- 99.9% uptime SLA
- Available add-ons
- White-labelling
- Custom integrations
- Additional onboarding sessions
30-day free trial on Professional · Cancel any time during trial at no charge · Annual subscription begins automatically after trial