AI Investigator · Early access

Your analyst spends 4 hours
answering “why did X drop?”
Inferah does it in minutes.

AI investigator built on vertical skill packs and your team's feedback. Hypothesis trees over your data, not LLM brainstorming. Starting with food delivery.

Limited early access · onboarding 1 customer per week
∴ REAL INVESTIGATIONS · DELIVERED IN SLACK
⚠ Fail rate spike · 6 stores
CRITICAL · 9/10
∴ Kitchen capacity exceeded. Prep time +47% at 6 stores. Courier, platform, external all ruled out. Lost revenue today ~$8,400.
📊 4 sources🌳 Pattern A⏱ 4m
CSAT trending down · 7 days
REVIEWING · 6/10
∴ Pricing change Day −3 correlates with CSAT drop. Secondary: “cold food” mentions +18% in tickets. Awaiting full comment analysis.
📊 5 sources🌳 Pattern C⏱ 7m
“Why is GMV down 12% in Dubai?”
ANSWERED
∴ Order frequency drop in returning cohort. MAU stable, AOV stable. 73% of drop in customers whose FIRST20 promo expired 8 days ago.
📊 3 sources🌳 Pattern E⏱ 3m
Morning brief · auto 06:00
PROACTIVE
∴ Ops recovered overnight. Fail rate resolved. 3 things to watch: Cairo CSAT, Dubai vendor onboarding, Riyadh weekend supply.
📊 8 sources🌳 4 patterns⏱ overnight

Not another LLM
on your warehouse.

Generic AI tools brainstorm, hallucinate SQL on schemas they don't know, and forget your context. Inferah is built differently.

“JUST PLUG AN LLM INTO OUR WAREHOUSE”
Generic AI analytics
  • ×Brainstorms ideas — no structured method
  • ×Hallucinates SQL on unfamiliar schemas
  • ×Each question starts from zero — no memory
  • ×Same answer regardless of who's asking
  • ×Vendor-defined methodology, take it or leave it
INFERAH APPROACH
Vertical knowledge × your data × accumulated context
  • ∴Hypothesis trees enforce systematic investigation
  • ∴Vertical skill packs handle schema mapping
  • ∴Feedback log makes next investigation smarter
  • ∴Your team contributes to your own skill pack
  • ∴Calibrated confidence — auditable, not opaque
HOW IT WORKS

Connect once. Then it runs.

Connect
Read-only to your warehouse, OLTP, CRM, dbt. 50+ connectors.
Research
Proactive, anomaly detection, scheduled, or ask anything. Four modes.
Tune
Works out of box. Edit trees, upload docs, set feedback rules when you want.
Answer
Conclusion, confidence, sources, recommended action — in Slack.
THE PRODUCT

A workspace that compounds with use.

app.inferah.com / workspace / investigations
Today
Investigations3
Active alerts
Morning brief
Workspace
Hypothesis trees
Skills & prompts
Knowledge base
Metric library
Data sources
Settings
Feedback log
Investigations
12 this week · 3 active now
⚠ Fail rate spike · 6 storesINVESTIGATING
∴ Vendor capacity · prep +47% · 7/10 rising
CSAT trending downREVIEWING
∴ Pricing change Day −3 · 6/10
Out of the box for business users. Power-user depth for analytics teams.
LAYER 01 · VAULT
Vertical skill packs
100+ metrics, 5 investigation patterns, hypothesis trees by failure mode. Food delivery v1 today.
LAYER 02 · STATE
Your data, your definitions
Your schema, KPIs, baselines — co-built with your analytics team. Speaks your language.
LAYER 03 · FEEDBACK
Compounds with use
Every 👍/👎 recalibrates priors. Gets smarter per-customer, not just per-product.

Built for food delivery first.

Domain knowledge curated with operators, mapped to your schema in onboarding.

100
metrics
5
patterns
23m
avg investigation
Operations
fail rate · prep · promise accuracy
30
Customer
CSAT · repeat · churn
20
Commercial
GMV · AOV · take rate
20
Supply
vendors · couriers · OOS
15
Growth
activation · LTV · CAC
15
Read-only. Tenant-isolated. Your data stays yours.
READ-ONLYTENANT ISOLATEDQUERY AUDIT LOGSSO + SCIMGDPR · UAE PDPL

Ready to investigate
in minutes?

Early access is limited. We onboard one customer per week, mapping your schema and customising your skill pack directly.

Or email hello@inferah.com