# Myriade > Reliable AI analytics, even on messy data. Connect your warehouse, get trusted answers in hours — no semantic layer required. ## About Myriade is an AI-native data intelligence platform. It plugs into your warehouse read-only and runs an autonomous reverse-engineering workflow: mapping tables, detecting anomalies, validating metrics, documenting relationships. The result is a living, trusted documentation layer that makes AI analytics reliable from day one — on the messy data you already have, not the clean data you wish you had. The core bet: in the age of AI, the context layer (documentation + business semantics + validated metrics) is the most strategic asset a company owns. It shouldn't be locked into a single warehouse vendor. ## The problem we solve Every enterprise wants AI on their data. Most can't make it reliable. Enterprise data is messy by nature — incomplete catalogs, contradictory metrics, undocumented tables, ten years of legacy decisions nobody remembers. Plug a generic AI (Snowflake Cortex, Databricks AI, BigQuery AI, plain ChatGPT) on top and it hallucinates: numbers that look right but don't match reality, citations to the wrong tables, confident answers to ambiguous questions. Myriade fixes the foundation first. No multi-month cleanup, no upfront semantic layer — the AI agents do the documentation work, your data team reviews and approves, and analytics become trustworthy. ## How it works — from messy to AI-ready in under 48h 1. **Hour 0 — Connect**: Plug Myriade into your warehouse, read-only. Snowflake, BigQuery, Databricks, PostgreSQL, Oracle, Redshift, Fabric, MySQL, and others. Minutes to connect. No prep required. 2. **Hour 4 — Autonomous discovery**: AI agents reverse-engineer your schema, sample your data, detect anomalies, map table relationships, and document ~90% of your tables without supervision. They use statistical analysis, existing metadata (dbt, prior catalogs), data samples, and min/max distributions. 3. **Hour 24 — Validate & complete**: When something is unclear (e.g. "is this price in USD or EUR?"), Myriade asks your team. Your data team reviews, corrects, approves. They stay in control. Built-in quality checks. Hallucination-proof. 4. **Hour 48 — Trusted analytics**: Business teams ask questions in natural language. Data teams reclaim their roadmap. The catalog is complete, correct, and live. Every answer is sourced and verified against cited tables. ## Core capabilities — one integrated intelligence layer Five modules, one platform: - **Documentation**: Auto-generated, continuously maintained table and column descriptions. Synced back to your existing catalog. Searchable by the whole team. - **Context**: Business semantics, ownership, metric definitions, lineage — the knowledge that makes AI answers correct. - **Analytics**: Natural language analytics (NL2SQL) with citations. Deep analysis agents for complex multi-source questions. Every answer is verifiable. - **Quality**: Continuous anomaly detection, freshness/completeness/consistency checks, proactive alerts before issues reach dashboards. - **Transformations**: AI agents suggest and generate dbt fixes, SQL optimizations, and transformation models. Your dbt project remains the source of truth. ## Why Myriade, not your warehouse's AI? Four reasons Snowflake Cortex, Databricks AI, BigQuery AI, or Fabric can't deliver the same outcome: 1. **Portability — you own your context layer.** Export your full semantic layer, definitions, and validation history anytime in open, portable formats. If you leave, you leave with everything. Warehouse-native AI locks your intelligence into that warehouse forever. 2. **Multi-source.** Connect Snowflake, BigQuery, Databricks, Oracle, PostgreSQL together or one at a time. Ask questions that join across sources. Switch warehouses without rebuilding your intelligence. 3. **Transparent, predictable pricing.** Flat platform fee plus per-user seats. Not a surprise invoice tied to compute consumption your vendor controls. 4. **Aligned incentives.** Warehouse vendors profit from your compute spend — every optimization they don't make is their revenue. Myriade includes a FinOps audit and typical clients cut 27–40% of warehouse compute in the first quarter. Guaranteed savings on the FinOps sprint, or you don't pay. ## Who we serve - **Data teams** get an AI-powered copilot: autonomous documentation, quality reviews, SQL optimization, dbt transformations, complex analysis. Analyses that used to take two days take minutes. - **Business teams** get autonomy: 5 minutes from question to insight, in natural language, under the governance of the data team. - **Leaders (CDO, COO, CEO)** get a data-driven roadmap with high-level strategic insights. Like a pocket data analyst. Typical buyers: Chief Data Officers, Heads of Data, Data Engineers, Data Platform teams, BI teams. ## Use cases Retail analytics, logistics optimization, revenue forecasting, ESG/CSRD reporting, e-commerce performance, customer 360, financial reporting, inventory management, payment & fraud, marketing attribution, HR/people analytics, multi-market expansion, supply chain visibility, regulatory compliance, risk management, SLA monitoring, product analytics, cost allocation, churn prediction, data quality scoring, data warehouse audit, sales performance, pricing optimization, geospatial analytics, campaign ROI, demand planning, process mining, AI model monitoring, vendor performance, expense analytics, executive dashboards. ## Pricing Transparent, predictable pricing — a flat platform fee plus per-user seats. Annual commitment, 12-month minimum. All prices exclude VAT. Fair-use policy applies to AI credits. ### Plans **Startup — for startups and small teams** - €149/mo (or €1,490/yr annual, -17%) platform fee - Standard user: €19/mo · Power user: €69/mo · Viewer: €9/mo - Min. 2 seats · 1 SQL source · Hosted AI - 240 org credits + 150 per Standard / 600 per Power - 3,000 API calls/mo · Standard SSO · Centralized hosting - Top-ups: 300 credits €119 · 3,000 credits €999 - Billed monthly or annually **Business — for established data teams (Most Popular)** - €649/mo (€7,788/yr) platform fee - Standard user: €49/mo · Power user: €169/mo · Viewer: €9/mo - Min. 5 seats · Up to 4 data warehouses · Hosted AI - 1,200 org credits + 240 per Standard / 960 per Power - 10,000 API calls/mo · Standard SSO · VPC or self-hosted - 1 CSM onboarding + 2 data team workshops/year - Top-ups: 300 credits €99 · 3,000 credits €799 **Enterprise — for mid-market and large organizations** - Custom pricing · Min. 20 seats · Annual commitment - Unlimited sources · Your choice of AI (hosted or BYOK) - Fair-use org credits · Unlimited API calls - SAML + SCIM · VPC or self-hosted - 1 CSM onboarding + 4 data team workshops/year ### User types - **Standard** — queries and catalog access - **Power** — full access + advanced AI agents (deep analysis, transformations) - **Viewer** — read-only, €9/mo across all plans ### 30-day pilot Prefer to try before committing? Run a 30-day pilot on your own data: 1. Custom demo on your stack, challenges, and team goals 2. Team meeting with your operational and data leads 3. Setup in 72h: documentation, warehouse cost optimization, conversational analytics 4 workshops included (Discovery, Use cases, AI agents, Autonomy). If the pilot is successful, lock in 20% off a 2-year commitment. ## Value creation ### Hard ROI - **Warehouse cost reduction: 27–40%.** FinOps agent analyzes query patterns, table usage, and compute waste. Typical findings: unused tables still being refreshed, inefficient queries on schedule, oversized clusters. For a €100k/yr Snowflake bill, expect €27–40k in identified savings — usually covering the platform cost. - **Grunt work eliminated:** - Empty/outdated catalog → auto-generated, continuously updated (40+ hrs/quarter saved) - Data quality firefighting → proactive anomaly detection (5–10 hrs/week saved) - "What does this column mean?" → descriptions and lineage always current (2–3 hrs/week saved) - Writing dbt fixes for known issues → auto-generated fix suggestions (1–2 hrs/incident saved) ### Strategic ROI - **Data team leverage: 2–3× capacity.** Data engineers spend 30–50% of their time on ad-hoc requests. Myriade handles the repetitive work; your team focuses on higher-value projects. - **Time-to-insight: days → minutes.** Before: ticket → investigation → SQL → validation → response (2–5 days). After: ask Myriade → validated answer with lineage (minutes). - **Self-service that actually works.** Most "self-service BI" projects fail because the data isn't trustworthy or documented. Myriade fixes the foundation. Expected impact: 50–70% reduction in ad-hoc data requests to engineering. ### What we don't claim - We don't migrate old pipelines. The more data already in the warehouse, the better. - We don't replace your data team — we make them more effective. - ROI depends on your current state. If your catalog is already perfect and your warehouse already optimized, value is lower. ## Integrations - **Data warehouses**: Snowflake, BigQuery, Databricks, Redshift, Fabric, Oracle - **Databases**: Postgres, MySQL, Oracle, SQLite, and others - **Transformation / orchestration**: dbt, GitHub - **MCP**: Myriade exposes an MCP server so Claude, Cursor, and other LLM clients can query your governed data directly ## Security - Read-only database access by default - Self-hosted or VPC deployment on Business and Enterprise - Zero data retention with LLM providers (Anthropic Claude by default) - Zero training on customer data - Data samples only, never full tables - Zero-knowledge protection: automatic masking of sensitive fields - Standard SSO on all plans · SAML + SCIM on Enterprise - Bring Your Own Key (BYOK) on Enterprise - RBAC, audit logs, DPA available on request ## Clients **Large retail**: Promod, Jules, Carter Cash, Dickson **Scale-ups**: Vroomly, Recommerce, Ekip, Le Plein From 50-person startups to enterprise retail with complex multi-channel data. ## FAQ **How is Myriade different from a data catalog like Atlan, Alation or Collibra?** Myriade completes your existing catalog rather than replacing it. The AI writes and maintains descriptions collaboratively: when there's ambiguity (e.g. "is this price in dollars or euros?"), it asks the user. Descriptions sync back to your catalog so it stays up to date. If you don't have a catalog yet, Myriade works as a lightweight one. Our vision: in the future, only AI agents will need data catalogs — humans will just ask questions. **Does Myriade replace dbt?** No. Myriade complements dbt — it can suggest fixes and generate models, but your dbt project remains the source of truth. **How long to see value?** 48–72 hours for both catalog automation and cost optimization insights. **What data does Myriade access?** Data samples only, not full tables. Zero data retention with GDPR-compliant LLM providers. Zero-knowledge protection to automatically mask sensitive fields. **Can I self-host?** Yes — available on Business and Enterprise plans. Startup uses our centralized infrastructure. **How do AI credits work?** Each plan includes an org-level AI credit quota plus per-user credits. Credits are consumed by NL2SQL queries, proactive agents, and catalog completion. Roughly 1 credit per question. Need more? Top-up packs of 300 or 3,000 credits. ## What Myriade is not - Not a BI tool (works with Metabase, Looker, Tableau, Power BI, etc.) - Not a replacement for your data warehouse (runs on top of Snowflake, BigQuery, Databricks, …) - Not a replacement for your data catalog (completes it and syncs back) - Not a one-time audit (designed for continuous, autonomous operation) - Not a migration tool for old pipelines ## Links - Website: https://www.myriade.ai - Product — Deep Analysis: https://www.myriade.ai/deep-analysis - Product — Catalog: https://www.myriade.ai/complete-catalog - Product — Quality: https://www.myriade.ai/quality-control - Product — Transform: https://www.myriade.ai/instant-transformation - FinOps / cut warehouse costs: https://www.myriade.ai/cut-warehouse-costs - Pricing: https://www.myriade.ai/pricing - ROI calculator: https://www.myriade.ai/roi - Security: https://www.myriade.ai/security - Case studies: https://www.myriade.ai/case-studies - Blog: https://www.myriade.ai/blogs - Try on open data: https://opendata.myriade.ai/paris ## Contact - Book a demo: https://calendly.com/myriade/30min - Email: contact@myriade.ai