The decision engine
A BI tool shows you a chart. Adaptrix tells you why it moved.
Ask a question in plain language. Adaptrix reasons across your systems, explains which drivers moved the number, forecasts what happens next — and shows the sources, reasoning and audit trail behind every answer.
- Explained answers, with provenance
- 39 connectors across 11 categories
- Self-hosted AI on sovereign EU infrastructure
- 5-language UI
Ask → explained answer
From question to decision in one pass
This is the core loop of the product. One question in, one explained, cited, auditable answer out — with the “why” as the climax, not a footnote.
Ask in plain language
No query language, no dashboard hunting. Type the question the way you would ask a colleague.
Watch the reasoning
Adaptrix streams the steps it takes — which sources it reads, how it joins them — as an auditable trace, not a black box.
Get the why, not just the what
The answer lands with a forecast and confidence band, and a driver breakdown whose signed contributions sum to the change.
Act on a recommendation
Each answer closes with a recommended next step — and an immutable audit ribbon recording what was read and shown.
Q
11.8s·auditedWill we hold our 95% on-time delivery target this quarter?
- Parsing the question — metric, segments, horizon
- Linking 4 connected sources
- Forecasting with a 95% confidence interval
On-Time Delivery · 12 wk actuals + forecast
forecast 85.0%
Why it moves
Late inbound from one supplier cascades to 3 assembly lines.
Forecast85.0%95% CI · 82.9%–87.1% by Q1
- −1.8 ptsSupplier B late inbound · 60% of the move
- −0.8 ptsLine 3 changeover time
- −0.4 ptsTransit / weather
Net change= −3.0 pts
Adaptrix recommends
- Dual-source Supplier B critical parts
- Add a 2-day inbound buffer on Line 3
Projected · recover to 94% on-time by Q1
TraceGDPR · checkpoints cleanSQL + provenance attachedAudit logged
Scripted replay of the real answer flow — illustrative data.
The why-toolkit
Four ways Adaptrix answers “why?”
Charts tell you what happened. The decision engine ships the data science that explains it — honestly framed, down to contributions that add up.
Explainable driver attribution
When a metric moves, Adaptrix decomposes the change into signed driver contributions (SHAP explainability) that sum to the delta. Not a vague correlation heatmap — a walk from the expected value to the actual one, driver by driver.
Automated root-cause analysis
Behind each driver sits a cause. Adaptrix breaks a moved metric into ranked branches — each with its share of the change — so the conversation starts at “what do we do about supplier B” instead of “who can pull the data”.
Forecasting, with confidence
Every projection ships with its uncertainty: a confidence band on the chart, not a single overconfident line. You see the range of outcomes before you commit budget to one of them.
- Forecasts on any metric you can ask about
- Confidence intervals rendered on the answer chart
- Re-forecasts as new data syncs in
Anomaly detection
Adaptrix watches your metrics for out-of-pattern movement and flags it while it is still a blip — before it becomes the surprise in the monthly report.
- Learns each metric’s normal seasonal rhythm
- Flags deviations with the drivers behind them
- Feeds anomalies straight into root-cause analysis
Auto-built knowledge graph
It maps your data automatically
Connect your systems and Adaptrix discovers how they relate: which tables join, on which keys, across which sources. The result is a living semantic layer — the map the decision engine reasons over when a question spans CRM, warehouse and files.
- Join paths discovered from the data itself — no weeks of manual modelling
- One semantic layer across databases, warehouses, SaaS tools and files
- Every cross-system answer cites the path it took through the graph
Interactive — hover or tab through a system to trace its auto-discovered relationships. Illustrative sources.
Scroll-driven
01 / Question
keep scrolling
Will we hit Q1 margin?
Question typed in plain language. No SQL, no joins, no dashboard rebuild. Routed through the Director Agent.
Specialised agents reason across your stack
1
SQL Agent — pulled 12 weeks of margin rollups from your data warehouse and joined the returns ledger
2
Forecast Agent — projected the next 13 weeks with a 95% confidence interval
3
RCA Agent — SHAP attribution across 150+ variables; defects on the premium product line explain 62% of the margin leak
4
Graph Agent — walked the auto-built Knowledge Graph to the Week 45 ops report; GDPR pre-flight clean
Insight, root cause, and a decision
High confidence
answer ready · 14.2s
Margin slips 1.4 pts by Q1 — 62% traces to premium-line returns.
Adaptrix recommends
- Re-certify 6 QC inspectors
- Rebalance 2 supplier contracts
Projected · +10-pt margin recovery
source: data warehousesource: returns ledgersource: ops documentsaudit log
Illustrative scenario — figures shown for demonstration, not a customer result.
The workbench
Every source lands in a governed catalog
Data Assets is where your connections live: profiled tables, a data-quality score per source, and KPI definitions Adaptrix discovers on its own — on-time delivery, return rate, order value — before anyone writes a metric spec.
Data AssetsCatalog
Operations DB · PostgreSQL
orders
96/100· Excellent
Rows
1.24M
Columns
18
Owner
Data Platform
Refreshed
Hourly · 12 min ago
Every customer order with status, value and promised ship date. Profiled and quality-scored on each refresh; used across on-time delivery and returns analysis.
The actual Data Assets view — catalog, quality scoring and auto-discovered KPIs. Illustrative tables and values.
Ask a question, keep the answer as a dashboard
Every answer you like becomes a live tile. My Dashboard is the board you compose by asking — each card still carries the reasoning, sources and math behind its number.
My Dashboard4 widgets
My Dashboard
composed from your conversationsConfigure in chatOn-time delivery
from chat94.2%+3.1 pts
vs 91.1% prior quarter · target 95%
“Show me on-time delivery for Q2”
Weekly revenue
from chat€1.82M+8.4%
“Add weekly revenue to my dashboard”
Anomaly alert
from chatReturn rate up in Region West
17.4% vs 11.2% baseline — outside expected range. Investigate the driver?
Auto-flagged on refresh
Revenue forecast
from chat€1.97M±6% · 90% confidence
“What’s our revenue outlook next month?”
Average order value
from chat€1,214−1.2%
30-day rolling · 8,420 orders
“Track average order value”
Open orders
from chat312+18
42 due to ship today
“How many open orders right now?”
The actual My Dashboard view — tiles composed from saved answers. Illustrative data.
Trust, built in
An engine you can hand the keys to
Autonomy is only useful when you can supervise it. Adaptrix pairs agentic execution with explicit approvals, full auditability and European data sovereignty.
Agentic execution
Adaptrix does not stop at answering. It can carry out multi-step work — pull the data, run the analysis, assemble the report, schedule the follow-up — as one autonomous flow.
Human-in-the-loop
Before an agent acts on your systems, it asks. Approvals are explicit, scoped, and recorded — automation with a hand on the wheel, not a leap of faith.
Governed by design
Role-based access controls who sees what. Every question, source read, and answer is written to an immutable, hash-linked audit trail. GDPR-native; SOC 2 and ISO 27001 are on the roadmap.
Sovereign by default: Adaptrix runs self-hosted open-weight models on its own EU infrastructure. Your data never leaves for a third-party AI provider.
It fits your stack — and your brand
Whitelabel
Your brand, your colors. Ship Adaptrix answers to your own users under your own identity — one theme switch, no re-implementation.
39 connectors, 11 categories
Databases, warehouses, files, SaaS tools and event streams — read-only connections that sync on your schedule.
5-language UI
Ask questions and read answers in any of five interface languages. The decision engine speaks your team’s language.
PostgreSQL
MySQL
Snowflake
MongoDB
Salesforce
HubSpot
Google Analytics
Shopify
Stripe
AWS
Google Cloud
Azure
Excel
Google Sheets
QuickBooks
KafkaAsk it why.
Bring a real question from your business and watch a live answer — reasoning, sources and math included — or model what it saves you first.