Ellipsis is an AI-based software development tool that serves as a virtual software engineer. It is designed to augment an engineering team's abilities by reviewing and writing code, answering queries about source code, and even implementing features or fixing bugs.
The tool utilises Large Language Models to understand source code and generate code as per user requests. Ellipsis offers capabilities such as automated code reviews, where it leaves insightful summaries and easy-to-accept suggestions, and also converts requirements into working, tested code.
It also provides pull requests as a service by automating bug fixes and making simple changes. Another feature is addressing pull request comments by interpreting them and writing corresponding, tested and working code.
The tool puts a strong emphasis on security, stating that it does not store, train on, or commit to one's source code without explicit permission from the user.
It supports a variety of languages, frameworks, and libraries, and is compatible with GitHub.
Isolated Cloud Sandboxes: Each agent runs in its own sandbox, ensuring that multiple agents do not interfere with each other while executing tasks.
Agents as Code: Agents are defined as YAML files in the repository, allowing for version control and easy updates through standard Git workflows.
Scoped Permissions and Budget Caps: The platform allows for granular control over permissions and spending limits for each agent session, enhancing security and cost management.
Integration with Existing Tools: The application integrates seamlessly with tools like GitHub, Slack, and Linear, enabling agents to respond to events and perform tasks directly within these platforms.
Full Audit Trail and Observability: Every session is recorded and searchable, providing a comprehensive audit trail for accountability and analysis of agent activities.
Cloud Agents: Run coding agents in isolated cloud sandboxes to automate tasks based on events from GitHub, Slack, or Linear.
Code Review: Automatically review pull requests and provide inline comments on code changes as they are made.
Automations: Schedule tasks or trigger actions based on specific events, such as cron jobs or mentions in communication tools.
Background Agents: Execute tasks in the background without blocking other processes, ensuring parallel execution of multiple agents.
Analytics: Track and analyze agent performance, spending, and session history for better resource management and decision-making.
Isolated Sandboxes: Each agent runs in its own sandbox, preventing interference between sessions and ensuring that multiple agents can operate simultaneously without conflicts.
Scoped Permissions and Budget Caps: The platform allows for granular control over permissions and spending, enabling teams to set limits on agent usage and maintain budgetary oversight.
Audit Trail and Observability: Every session is recorded and searchable, providing a comprehensive audit trail that helps teams track actions and decisions made by agents over time.
Integration with Existing Tools: Agents can be triggered by events in tools like GitHub, Slack, and Linear, allowing for seamless integration into existing workflows and enhancing productivity.
Agents as Code: Agents are defined in YAML files within repositories, enabling version control, easy updates, and deployment through standard Git workflows.
New organizations receive a one-time $100 credit ($10 for individual accounts) that applies to usage before billing.
Pricing is usage-based, charging for CPU and memory metered by the second, tokens at cost with no markup, and a 40% platform fee on token costs.
There are no per-seat fees; the entire team can use Ellipsis, paying only for what agents consume.
Support packages are available: Standard at $5,000/month, Advanced at $7,500/month, and Premier at $15,000/month, each with varying levels of support.
For AWS deployments, costs include AWS charges for CPU ($0.142/vCPU-hour) and memory ($0.024/GB-hour), in addition to the platform fee.
