Create
Three alerts on checkout, orders-svc and payments-gw become CASE-241004 (P1).
OllyThe autonomous SRE and security agent
Agents that watch every service, dashboard and threat, investigate on their own, open the case, page the right people and line up the fix.
Paused while you read. Move away to continueThe real Olly screen, playing illustrative scenarios with sample data. Point at it to pause and scroll back, open the player with ⤢, or type your own question.
Cases on autopilot
From the first alert to the fix
Olly opens the case, works out how bad it is, hands it to the team that solved it last time, wakes them up and has the remediation waiting when they arrive.
Triage
18% of checkouts failing, about $42k an hour, starting with orders-svc v3.1.0.
Assign
To team-payments: they resolved 4 of the last 5 escalations like this one.
Page
@maya paged in PagerDuty, #inc-checkout opened in Slack with the evidence.
Remediate
Rollback PR #2291 to v3.0.4 is ready. One approval and checkout recovers.
A team of agents
An agent on every service, talking to each other
Give each service, dashboard or threat surface its own Olly. They watch around the clock, and when one sees something, it asks the others before it wakes anyone up.
- Service watch learns each service's normal and flags what isn't.
- Dashboard watch reads your dashboards like your best engineer would.
- One story instead of five alerts: agents share findings before they escalate.
Olly
orders-svc
payments-gw
Checkout dashboard
Security sentinel
Security
The security analyst that never sleeps
Olly hunts across identity, endpoint, cloud and application telemetry in one investigation, so a login spike and a process on a laptop stop being two tickets in two tools.
- Correlates every source. Okta, CrowdStrike Falcon, AWS CloudTrail, Wiz and your own logs, joined by user, host and IP.
- Contains, with your approval. Isolate a host, revoke sessions, block an IP, each one a button you approve.
- Maps it to MITRE ATT&CK and writes the case your auditors will ask for.
Investigation: Okta brute force (P1)Live
41 accounts targeted from 203.0.113.47 in 6 minutes
T1110One login succeeds: j.park, MFA push accepted
T1078Falcon flags credential dumping on LAPTOP-JPARK
T1003j.park's role calls ListSecrets in prod for the first time
T1552Olly: host isolated, sessions revoked, IP blocked, SecOps paged
Contained in 3m 12s
Dashboard and service watch
It reads your dashboards so you don't have to
Point Olly at a dashboard or a service and it keeps watching: what moved, why, and whether it matters, pinned right on the widget and sent where your team will see it.
Dashboards / Checkout & payments healthWatched by Olly
1/2 Checkout success fell to 84% at 20:31, the same minute payments-gw started rejecting charges with 400. Both follow the orders-svc v3.1.0 rollout. Posted to #inc-checkout
Scheduled tasks
Ask once. Olly reports forever.
Turn any question into a schedule. Olly runs the investigation, not just the query, and delivers it to Slack, email or a ticket.
/schedule every Monday 09:00: which errors cost us the most revenue last week, and which release caused each? Send to #eng-leads and email vp-eng@acme.ioScheduled task created
- When
- Mondays at 09:00, Europe/London
- Investigates
- Logs, traces, APM and product analytics
- Delivers to
- #eng-leads vp-eng@acme.io
Monday09:00
Olly APP09:00
3 error patterns cost $118k in lost checkouts last week. The largest began with orders-svc v3.1.0 (missing Idempotency-Key header), fixed by rollback on Thursday.
Weekly revenue-impacting errors
To vp-eng@acme.io, from Olly
Global memory
It remembers how your company works
Who owns what, which deploy windows are off limits, how your team likes its answers. Olly keeps what it learns, for you, your team or the whole organization, and you can see and edit every memory.
Orgteam-payments owns payments-gw and the charge flow
OrgNo production deploys on Fridays after 14:00
TeamPage SecOps for any P1 identity alert
YouDana wants p95, not averages, in every answer
Used 3 memoriesp95 charge latency is 1.84s, up from 310ms. I've assigned the case to team-payments and paged their on-call. It's Friday 15:10, so I've prepared the rollback but held it for approval.
Plugins
Works with the tools you already run
Olly reads from them, writes to them and acts in them, with the permissions you give it. Anything else with an API connects through MCP.
- Slack
- Gmail
- CrowdStrike
- PagerDuty
- Okta
- GitHub
- Jira
- AWS
- Microsoft Teams
- ServiceNow
- Wiz
- Any MCP server
In the product today
See Olly work in Coralogix
Plain language in, evidence out
Every answer comes with the query and the chart
Ask a question and Olly writes and runs the DataPrime queries itself, across logs, metrics and traces, and draws the chart that answers it. Open any result in Explore to keep going.
Root cause
From alert to root cause, on its own
Olly reads the Case, gathers evidence across every signal, forms hypotheses and tests them, then writes up the root cause with its confidence, the evidence behind it and the steps to fix it.
Skills
Teach it how your team works
Write your runbooks and conventions as Skills, for yourself or your whole team. Olly loads the right one when a question calls for it, or when you call it with /.
Your models, your data
Private by design, on by your choice
Pick the model
Each user chooses GPT, Claude or Gemini, and how much effort Olly puts into reasoning.
Private inference
Models run in private deployments inside Coralogix’s own cloud networks, in your region. Prompts and answers are never used to train foundation models.
Bound by your permissions
Olly only sees what the user asking can see, and every action waits for the approval you require. Admins set usage limits per user, group and API key.
SOC 2 Type II; ISO/IEC 27001, 27017, 27018, 27701 and 42001; PCI DSS v4; HIPAA; GDPR
From the teams using it
Hours of digging, down to minutes
“I used to spend up to an hour digging through data. Now, Olly scans my telemetry and gives me what I need in about 5 minutes.”
“Olly makes every interaction smarter, faster, and more actionable, saving me hours of digging through data.”
“I want to give props to Olly. I’ve been using it a lot for debugging, and it’s an incredible time saver.”
How Simpplr uses Olly to turn hours of troubleshooting into minutes
AI ready at scale
Observability that was built for AI,
Observability that was built for AI,
not repackaged for it
Run your agents and teams on the right foundation, so they don't approximate, they know
14 days, all features, no credit card required.
Alert detection time<1ms
Platform uptime99.9%
Data ingestion capacity10x