Meet Toby AI Agents
Delegate engineering work to Toby AI Agents.
Build AI teammates you can delegate engineering work to. They investigate issues, run checks and return findings and propose fixes, working in your environment with your tools and approval rules.
SRE AgentWorking✓Logs → 1,204 pool_wait_timeout in 18m
✓Metrics → db.pool.active at max since 14:22
✓Traces → payment.charge p95 2.8s · 8,412 sampled
✓Events → deploy checkout-api v2.14.1 at 14:20
Correlated with AppDev Agent
Awaiting your approval
Roll back checkout-api to v2.14.0
Click Approve rollback to see what the agent does next.
Example run. Illustrative data.
Each Toby AI Agent comes ready for its role.
Set the scope once and the agent works inside it, at any time. Your team gets a finding to act on, and approves any change.
INCIDENT INVESTIGATION
SRE Agent
“Picks up your alert, traces the cause and leaves a rollback for you to approve.”
Skills
Alert triage- Log investigation
- Trace analysis
Reads
Metrics, logs and traces for the services you scope
Delivers
Likely cause, supporting evidence, and a rollback or scaling change for your approval
Where you start an agent
An agent is not a separate console. It opens where the work already is, and inherits the context you were looking at.
01 · From a dashboard
You spot a spike
Open the agent panel beside the chart. It takes the time range, the services in view and your filters with it.
TrueWatch
Context attached from your dashboard
Recurring tasks your teams do, now on autopilot with Toby AI Agents.
Set the rules. Toby AI Agents act on alerts, schedules or questions, with your approval before making changes.
Approval and audit
Runs on infrastructure you control
- Install the agent runtime on a Linux host or in a container, then connect it to your TrueWatch workspace. Your team manages its network and permissions, and the runtime can send its own metrics, logs and traces back so you can watch the agent like any other workload.
SRE Agent
Tracing latency
Security Agent
Scanning IAM
AppDev Agent
Reviewing deploy
FinOps Agent
Finding idle nodes
SRE Agent
Tracing latency
Recorded in audit log
SRE Agent · approved by human
A-2207 · Roll back checkout-api to v2.14.0 · 14:41
Roll back checkout-api to v2.14.0
3 regions · 14 pods · reversible
Recorded in audit log
Roll back checkout-api to v2.14.0
A-2207 · Approved by human · 14:41
HOW IT RUNS
Your agents run on your own host, and every run stays on the record.
The runtime installs in your environment and connects to your TrueWatch workspace. Each alert, schedule or question leaves a run your team can open and review.
Read every stepEach run keeps the tools the agent called, what they returned and the evidence behind the answer.- Put it on a scheduleRun it hourly, daily, on weekdays, weekly or on a cron schedule, and post the result to your team’s chat. Anything that changes a system still waits for your approval.
- Host it yourselfInstall the runtime on a host or in a container you manage, then connect it to your TrueWatch workspace. Your team controls the network, permissions and dependencies.
Picked up bySRE Agent· asked from the checkout-api dashboardlog.search
1,284 timeouts to payments-db
metric.query
p95 2.8s on payment.charge
deploy.events
v2.14.1 shipped at 14:20
Answer
v2.14.1 cut the payments-db pool from 50 to 10.
SRE Agent · every weekday 09:00
checkout-api p95 check
DoneDoneNeeds approvalDoneDoneSame scope every run. Changes still wait for approval.
Dashboard
TrueWatch
data engine
One DQL
Every agent
SRESecurityAppDevFinOps
CustomNo copy of the data.
Your dashboards and every agent read from the same place.
HAND OVER THE RECURRING WORK
Give every team an agent. Keep control. Save time on repetitive work.
Put Toby AI Agents to work on alert triage, recurring checks and investigations. Find a plan that fits the work you want to delegate, so your team can spend less time repeating it.
You have questions. Toby has the answers.
TruePilot works beside your team. You ask a question, and it analyzes the evidence and recommends a fix for your team to carry out. Toby AI Agents take on a role of their own. Each one investigates inside the scope you assign, proposes the action, and runs it once you approve.
Explore Toby AI TruePilot
Get more done with the AI that knows your systems.
- Explore product
Toby AI TruePilot
Ask in plain language. Analyse your current view and build dashboards with AI assistance.
- Explore product
OWL CLI & MCP Server
Connect AI agents and developer workflows to TrueWatch observability data.
- Explore product
AI Agent Observability
Trace agent steps, model calls and tool execution to investigate latency, cost and risk events.
- Explore product
LLM Observability
Track model requests, latency, token usage and cost to pinpoint slow or failed calls.
Keep up with what’s next in observability and AI.
See Toby AI Agents in action.
Start building your autonomous AI engineering team today.