Agentic Analytics
On all your data

One source of truth for your entire business, from raw data to company-wide agentic workflows. So every team works with the same numbers, every time.

9 Active deployments
460+ Employees connected
€65M+ Tracked annual revenue
25k+ Analytics API requests / day

Trusted by

Deployed with open source components

Our open-source Semantic Nexus architecture gives your organisation complete visibility into its data — from raw sources to AI-powered querying.

Domain Reports

Three purpose-built reporting domains aligned to how companies are structured: Commerce Atlas (go-to-market, unit economics), Finance Atlas (margins, financial operations), and Operations Atlas (utilisation, capacity). Each delivers periodic dashboards and trend analysis tailored to the domain's owners and targets.

Details →

Data Model Explorer

A unified, navigable interface that maps your entire data architecture. Source tables are shown post-transformation with aggregations for fast analysis. Each critical metric gets its own dedicated page with a scorecard, numerator/denominator breakdown, and business context provided by the client.

Details →

Data Entry Tests

Automated, client-specific tests that continuously validate data entries and business workflows. Each test is assigned an owner, a deeplink for direct resolution at the source, a context description, and a unique test ID — keeping data quality accountable and easy to communicate across teams.

Details →

Pipeline Freshness

A live flowchart of your full data pipeline from extraction through transformation to consumption. Every stakeholder can see how data moves across your architecture — which systems are involved, what the current status is, and where bottlenecks occur. Invaluable when planning refactors or expansions.

Details →

MCP Servers

We configure multiple Model Context Protocol servers tailored to specific use cases and agentic workflows — each one exposing the right slice of your semantic layer to the right AI system. Whether it's a customer-facing agent, an internal analyst assistant, or an automated reporting workflow, every MCP server is purpose-built for how that agent actually needs to query your data.

Details →

Built on best in class tooling

We work with the leading data stack tools, keeping clients in full control of their own infrastructure. We have a tooling-agnostic approach in designing the Semantic Nexus, with a strong preference for open-source and self-hosted applications.

Sources

Product database (PostgreSQL, MySQL) · Bookkeeping (Exact Online, Yuki) · CRM (HubSpot) · ERP (Float, Infor) · Customer care (Freshdesk, Zendesk) · Spreadsheets (Google Sheets, Excel)

Extraction · streaming (product database)

Debezium

Extraction · batch (all other sources)

Airbyte Hevo Data

Warehouse · real-time (from the stream)

ClickHouse

Warehouse · batch

BigQuery Snowflake Microsoft Fabric

Transformation

dbt

Semantic layer

Cube Microsoft Fabric

Consumers · AI models

Claude OpenAI Gemini

Consumers · MCPs

nao analytics

Consumers · AI agents

Quality Guardian LangChain agents

Consumers · Reports

Data Studio Power BI

Consumers · Your own apps

Next.js Django

Maintained by reliable frameworks

The Semantic Nexus is underpinned by three frameworks that cover architecture, performance tracking, and security.

SDA

Standardised Data Architecture

Best practices for building a data pipeline and data model that is robust, flexible, and low cost — refined over years of real client work.

Details →
IPF

Integrated Performance Framework

A systematised way of presenting the data model that strikes the right balance between simplicity and depth for all stakeholders.

Details →
MSP

Modular Security Protocol

A structured set of security checks designed for client-side and open-core architectures, where the security environment is inherently more complex.

Details →

Depended upon daily by long term clients

How one of our clients, Freeday, runs its reporting and operations on the Semantic Nexus we built and maintain for them.

Freeday

Freeday builds AI digital employees for customer service and finance teams. Maxq designs, runs and maintains their complete analytics stack: Airbyte, dbt and Cube on Snowflake, with the Quality Guardian on top.

Visit Freeday →

The starting point

Ahead of a growth-investor round, Freeday needed one set of metric definitions that finance, sales, operations and the board could all trust: ARR, NRR, CAC payback and the performance of every digital employee. The data sat in six systems: the Freeday platform itself, its ClickHouse event logs, HubSpot, Airtable, Exact Online and Google Sheets.

What runs today

  • ·Airbyte loads all six sources into Snowflake every night. The Exact Online connector is an open-source Maxq contribution that Freeday runs on its own virtual machine.
  • ·dbt builds 79 models and runs 131 tests in one 14-minute job each morning. Every metric is defined once, in version control, with its tests and documentation next to it.
  • ·Cube serves 13 cubes and 14 views to every consumer: the Data Studio reports for the commercial, operational and financial teams, the KPI portal, the investor sheet, Claude through an MCP server, and the metrics embedded in Freeday's own platform.
  • ·The Quality Guardian runs 11 data-entry tests every morning and writes the fixes back to HubSpot and Airtable, so the three systems stay in step without anyone reconciling by hand.
6
source systems, one model, one set of definitions
79 · 131
dbt models and tests, built daily in 14 minutes
13 · 14
cubes and views, serving six consumers from one semantic layer
90M+
conversation events of the digital employees, modelled per case

Is there business value we can create together?

Schedule half an hour with Philip where you show how your data is set up now, together you will explore what one source of truth would change for your team, and whether it's worth the investment.

Schedule a Call