Platform

A DBRE that joins your team and earns your trust

SIXTA is onboarded like a person — introduced to your infrastructure, your naming conventions, your policies, and your team. It builds context over time, just as a new hire would. It starts by observing. Over time, it earns the authority to act.

Meet Your New DBRE

How SIXTA thinks
From your infrastructure through to your team — every layer builds on the last. SIXTA connects directly to your databases and runs inside your network.
Your Infrastructure

Databases

Direct read-only connection to MySQL and PostgreSQL

Monitoring

Your existing observability stack — CloudWatch, Performance Insights, Percona PMM

Events

Deploys, incidents, schema changes from your CI/CD and incident tools

read-only · no PII collected · processed locally
Connect

Database Telemetry

Connects directly using native protocols. Collects query patterns, schema metadata, and configuration — continuously.

Monitoring Integration

Ingests metrics from your existing monitoring. Supplements direct database telemetry without replacing your current stack.

Change Awareness

Receives signals from your deploy pipeline, incident management, and source control. Knows when something changed.

Learn

Baselines

What normal looks like for each database. Time-of-day and day-of-week patterns. Workload fingerprints.

Anomaly Detection

Groups related signals into incidents. Distinguishes real problems from noise. Classifies severity automatically.

Proactive Monitoring

Catches new queries before they cause problems. Tracks capacity trends and headroom. Detects query plan regressions.

builds and consults
Knowledge

Pattern Recognition

Recognises problems it has seen before and resolves them instantly — without re-investigating from scratch.

Your Context

Runbooks, architecture docs, escalation paths, and business criticality — referenced in every investigation.

Institutional Memory

Every investigation is documented and searchable. SIXTA gets faster and more accurate with every incident.

SIXTA

Reasoning Engine

Investigates anomalies, diagnoses root causes, and proposes fixes. For known patterns, resolves instantly from past experience. For novel problems, reasons through your environment step by step.

investigates · diagnoses · proposes
Act

Investigate

Analyses query plans, examines locks and sessions, checks replication health, assesses capacity — the same diagnostic steps your best DBA would take.

Recommend

Proposes specific fixes with evidence and expected impact. Every recommendation shows its reasoning and links to your runbooks.

Execute

With your approval, applies safe changes under strict guardrails. Every action is policy-gated, auditable, and reversible.

Slack · Microsoft Teams · or your own systems
Your Team

Ask questions

"What's causing the CPU spike on db47?" — SIXTA investigates and responds in your workflow with findings and recommendations.

Receive findings

Proactive alerts delivered to Slack, Teams, or your own internal systems. Not dashboards — clear explanations of what happened, why, and what to do about it.

Stay in control

You decide what SIXTA is allowed to do. Approval workflows, change windows, and blast-radius limits are all configurable.

Transparency

Shows its reasoning every time

Consent

No changes without your approval

Auditability

Every action logged and traceable

Boundaries

You define where it can act

Reversibility

Never acts without an undo path

Isolation

Deploys inside your network. Air-gapped option available.


Connect. Learn. Act.
SIXTA takes level-1 and level-2 DBRE work off your plate. Three phases, each building on the last — earning authority progressively, just like a new team member would.
Phase 01

Connect

SIXTA deploys inside your infrastructure and establishes read-only connections to your MySQL and PostgreSQL instances. It can also ingest from AWS CloudWatch, Performance Insights, Datadog, and Percona PMM, but doesn't depend on them.

SIXTA does not collect personally identifiable information. With a local model and custom integrations, SIXTA can run fully air-gapped — no outbound connections required.

Phase 02

Learn

SIXTA maps your database topology, catalogs query patterns, and builds a baseline of normal behaviour. It learns the difference between a genuine anomaly and a Tuesday morning batch job, a month-end reporting spike, or a deploy that temporarily increases connection count.

This understanding is specific to your environment. No generic thresholds.

Phase 03

Act

When something deviates, SIXTA traces the root cause to a specific query, table, or missing index and delivers the analysis to Slack, Teams, or your own internal systems. Initially, your team reviews every recommendation. As SIXTA proves its accuracy, you can unlock autonomous actions for safe, risk-assessed operations.

You control exactly what SIXTA is allowed to do.


What SIXTA actually does

Root Cause Analysis

When an incident hits, SIXTA traces it from the symptom (CPU spike, replication lag, connection pool exhaustion) to the specific cause. Not "your database is slow." More like "unindexed JOIN on orders.user_id in OrderService.processPayment() is causing a full table scan."

Source: Full table scan in OrderService.processPayment()
Table: orders, missing index on user_id
Fix: ADD INDEX idx_orders_user_created (user_id, created_at)
Expected: 94% query time reduction, $47K/yr saved

Proactive Detection

SIXTA doesn't wait for things to break. It identifies degradation patterns days before they become incidents. A query that's getting 2% slower each week. A connection pool that's trending toward capacity. A replication lag pattern that correlates with a specific workload.

Query pattern in PaymentService will exceed
connection pool limits within 14 days at
current growth rate. Recommended action: increase
pool size or optimise query batch processing.

Knowledge Capture

Every investigation, every root cause, every fix becomes part of SIXTA's understanding of your environment. When your senior DBA leaves, their diagnostic intuition doesn't leave with them. SIXTA retains the institutional knowledge that's usually locked in someone's head.

Shows its working

Every recommendation is evidence-backed. Every action is auditable. SIXTA documents what it did and why, building a knowledge base your team can learn from.


You decide how far it goes
SIXTA's autonomy is progressive. Each level unlocks independently, and every action is logged.
01
Observe

Read-only monitoring

SIXTA connects with read-only credentials, ingests telemetry, and builds its understanding of your environment. It watches query patterns, maps topology, and establishes baselines. Nothing is modified. Nothing is written. This is where every deployment starts.

02
Recommend

Analysis with human review

Root cause analysis and specific fixes delivered to Slack or Teams. Your team reviews every recommendation before executing. SIXTA shows its reasoning: what to do, why, and what happens if you don't. This is where most customers operate day-to-day.

03
Act

Autonomous resolution with guardrails

Unlock autonomous actions for risk-assessed, safe operations. Kill a runaway query. Adjust a connection pool. Add a missing index during a maintenance window. Each action type is individually gated. SIXTA assesses risk before acting, runs within strict guardrails, and produces a full audit trail. You control exactly what it's allowed to do.


Runs where your data lives
SIXTA deploys inside your infrastructure. Your database telemetry never crosses your network boundary.
Your VPC / Data Centre
MySQL Instances

Read-only user, native protocol

PostgreSQL Instances

Read-only user, native protocol

CloudWatch / PMM

Optional metric supplement

query telemetry
schema metadata
perf metrics
SIXTA

Container or VM
in your infrastructure

findings
recommendations
audit logs
Slack / Teams

Channels, threads, adaptive cards

Your Internal Systems

Custom integrations via bespoke connectors

Audit Log Store

Your S3, CloudWatch, or SIEM

Air-Gapped Option Available

With a local model and custom integrations, zero outbound connections required. Slack/Teams integration requires a connection to api.sixta.ai for message coordination.

No Cross-Customer Training

Your data trains your DBRE only — never shared

Requirements
Database Access

Read-only user on each monitored instance. MySQL: SELECT on performance_schema and sys, PROCESS, REPLICATION SLAVE. PostgreSQL: pg_monitor role. No schema write access required to start. See full permissions reference.

Compute

Container (Docker/ECS/EKS) or VM. Typical footprint: 2 vCPU, 4GB RAM. Scales with number of monitored instances. No GPU required.

Network

Inbound: none required. Outbound: HTTPS to your configured LLM endpoint, Slack/Teams via api.sixta.ai (or your own messaging systems), and optionally AWS/Datadog/PMM APIs. With a local model and custom integrations, zero outbound connections required. See full network reference.

Setup Timeline

Connect: hours. Baseline learning: days. Full environment understanding: 2–4 weeks. SIXTA gets more accurate over time as it learns your patterns.

See it in your environment

We'll connect to your databases, show you what SIXTA finds, and let the analysis speak for itself.

Meet Your New DBRE