12 machine learning models · Groq + JEV AI · PDF report
Turn Any Log Into Clear Answers
Upload a system, server, network or application log. Get the story behind it in seconds: health score, timelines, anomalies, attacks, root causes and exact settings to change, for investors and developers alike.
- No sign-up
- Encrypted per upload
- Deleted after one hour
orders-service.log · Java Application Log · Illustration
Health Score
82
Good
Incidents Found
1
4 of 5 models agree
Root Cause
Pool
Connection exhaustion
Security
100
No threats
Events per 5 minutes
INFO GET /api/products 200 34ms
WARN Payment gateway responded slowly (2140ms)
ERROR HikariPool-1 - Connection is not available
INFO Published OrderCreated event for order 50231
Detects the format automatically. System, server, network and application logs all work.
- Nginx
- Apache
- Syslog
- SSH / auth.log
- UFW
- iptables
- Cisco ASA
- JSON Lines
- logfmt
- Spring Boot
- Log4j
- Hadoop
- Spark
- Zookeeper
- OpenStack
- Android Logcat
- Windows CBS
- Proxy Logs
- Kubernetes
- Custom Formats
Features
Everything a Log Can Tell You
From the first glance for a board meeting to the exact stack trace for a fix, in one place.
Works With Any Log
Format, timestamps, severity and components are detected automatically. The Drain algorithm learns message patterns, so even custom logs make sense.
Timelines and Replay
See volume, errors and severity over time, then replay the log like a live feed to watch an incident unfold.
Anomaly Detection
Five models vote on every time window. Flagged periods explain exactly which signals moved and by how much.
Security Findings
Brute force, logins after guessing, privilege escalation, port scans, floods, SQL injection, scanners and leaked secrets.
AI Root-Cause Analysis
Groq writes plain-English summaries and root-cause hypotheses. JEV by TypeSafe decides severity, urgency and business impact.
Ask Your Log
Ask questions in plain English and get answers that cite the exact log lines they come from.
Log Explorer
Filter millions of lines by severity, pattern, time and text, directly from the encrypted file.
Live Tail
Stream logs from your terminal or browser and watch rates, patterns and alerts update every second.
PDF Report
A polished report with the summary, charts, incidents, findings and recommendations to keep or share.
Two Audiences
One Upload, Two Clear Views
The same analysis, told the way each reader needs it.
For Investors and Leaders
- A single health score with a plain-language verdict
- Availability, error budget and stability trend
- Incidents explained as business impact, not stack traces
- Security risk rated from secure to critical
- An executive summary you can paste into a board update
For Developers and SREs
- Top error patterns, exceptions and stack traces
- Slowest endpoints with p50, p95 and p99 latency
- Event flow and rare transitions from a Markov model
- Alert thresholds, timeouts and rate limits computed from your data
- A searchable explorer across every line
How It Works
Three Steps to Answers
No agents to install, no accounts, no configuration.
Upload Your Logs
Drop .log files, a .gz or a .zip / .tar.gz archive up to 100 MB, or pick a sample. Everything streams straight into encrypted storage.
We Analyse Everything
Parsing, statistics, 12 models and a security scan run in seconds, then AI adds the story.
Explore and Act
Read the summary, drill into incidents, ask questions and download the PDF report.
Machine Learning
Twelve Models, One Clear Verdict
Anomalies are only flagged when at least two independent models agree, which keeps false alarms down.
Isolation Forest
Isolates windows that behave unlike the rest
Local Outlier Factor
Finds windows unusual compared with similar periods
PCA Residuals
Spots broken correlations between message types
Robust Z-Score
Flags sudden spikes and drops in volume and errors
Markov Chain
Learns normal event order and flags odd sequences
K-Means
Groups message patterns into themes and clients into segments
DBSCAN
Surfaces rare one-off messages hiding in the noise
Change-Point Detection
Pinpoints the moment behaviour shifted
Holt-Winters
Forecasts volume and errors with a likely range
Apriori Rules
Finds errors that reliably happen together
Periodicity Detection
Discovers cron jobs and missed runs
Drain Template Mining
Turns millions of lines into a few hundred patterns
Privacy by Design
Your Logs Stay Yours
Logs contain IP addresses, usernames and sometimes secrets. We designed every step to keep them private.
Streamed
Your file never touches disk unencrypted.
Encrypted
AES-256-GCM with a key only your link unlocks.
Analysed
In an isolated worker process.
Redacted for AI
AI sees statistics, never raw lines.
Shredded
Deleted automatically after one hour.
Questions
Frequently Asked Questions
Straight answers about formats, privacy, accuracy and the AI behind the analysis.
Anything that is plain text with one event per line. The format is detected automatically: Nginx and Apache access logs, Apache error logs, syslog (RFC 3164 and 5424), SSH and auth logs, UFW, iptables and Cisco ASA firewall logs, JSON lines, logfmt, Java, Spring, Log4j, Hadoop, Spark, Zookeeper, OpenStack, Android logcat, Windows servicing logs and more. Unknown formats still work: timestamps, severity and message patterns are learned from the text itself.
Your Next Incident Is Already in a Log
Find it before your users do. It takes less than a minute and nothing is kept.