Product Positioning
Edge Delta is a telemetry-native AI SRE (Site Reliability Engineering agent), officially described as a "production immune system." It is built on the telemetry pipeline architecture the company has run for years, enabling the agent to reason directly over live data rather than querying tools from the outside. Observability is the underlying architecture, and AI SRE is the capability layer running on top of it.
Core Philosophy
Edge Delta believes that software release velocity accelerates year after year, models can already write code, open pull requests, and perform actions in production systems, yet most tools for monitoring production still assume that humans will read every alert and make every decision. Edge Delta is built on a different assumption: monitors patrol continuously, investigations begin before anyone is paged, and by the time an engineer opens their laptop, the root cause and its evidence are already prepared. This requires that telemetry be processed where it is generated, that baselines be learned from one's own traffic, that the agent carry evidence for every conclusion, and that a rollback path be written before any change is released.
Levels of Autonomy
Autonomy is earned step by step. The agent starts at the Propose level: it freely reads telemetry data, drafts changes, and seeks approval before any write. It works in parallel with the existing tech stack and gains more permissions in increments that the user approves and can revoke at any time. Every write goes through a human-defined gate; there is no direct path from a log line to a tool call.
The Anomaly-to-Fix Process
- Detect anomalies before anyone is paged: monitors patrol continuously, and the agent triggers based on baselines learned from the data, with no need to manually set thresholds.
- Contain before the blast radius grows: signals are captured within the detection layer and correlated before reaching anything that can change the production environment, with multiple alerts merged into a single analyzed issue.
- Autonomy is granted through approval: every write goes through an approval gate, chat-initiated actions are approved item by item, and scheduled tasks run pre-approved playbooks.
- Fixes come with evidence: remediation plans come with the root cause, timeline, and proposed changes, and are executed by a human-approved or pre-approved playbook, with rollback automatic and pre-written.
Agent Types
Edge Delta provides four agents out of the box and supports building custom agents:
- AI SRE: triages alerts and performs root cause analysis.
- AI Security Engineer: reviews pull requests for vulnerabilities and monitors for suspicious activity.
- AI Software Engineer: performs in-depth code reviews and proposes targeted fixes.
- AI Work Tracker: keeps tickets in sync with what actually happened, creating or updating tickets when work is missing or unclaimed., and links back to the source. Custom agents can be tailored to team workflows, tools, and processes, and have the same connectors, guardrails, and auditability as built-in agents.
How to use and setup
Agents connect to the services, data sources, and tools users already run through out-of-the-box connectors, with new connectors added regularly. Setup takes only a few minutes: connect the stack, activate the needed agents, and see initial results within an hour. Start without a credit card, with SSO and SAML included. Deploy alongside the existing stack in a single namespace for a two-week side-by-side measurement, without affecting existing collectors and data. Use helm uninstall one command to remove it, with no migration, no rerouting, and no reconfiguration of anything.
Data privacy and security
User data belongs to users, and Edge Delta never uses user data to train models. Agents start at the Propose trust level, only read and draft changes, apply nothing without approval, and every action is logged for auditing. Connector credentials are encrypted, and the platform is SOC 2 Type II certified.
Applicable scenarios
Suitable for SRE teams, security teams, and engineering teams. SRE teams typically start with AI SRE; security teams start with AI Security Engineer; engineering teams adopt AI Software Engineer for in-depth PR review and remediation suggestions; AI Work Tracker suits all teams. Agents do not replace engineers, but take on the heavy lifting of alert triage, data correlation, initial investigation, first-round code and security review, and ticket maintenance, transforming engineers from first responders into decision-makers.
Billing and limits
Pricing is based on token usage. You can start for free, choose the agents you need (out-of-the-box or custom), and run them in production within minutes.
No information available
The official text does not provide details such as specific price amounts, free quota limits, detailed technical limits of each agent, the complete list of supported connectors, or data storage geographic locations; this information is not yet available.
