Compounding intelligence your company owns. Trinity is the operating system for the AI-native company: persistent AI agents that run your company's work, keep their memory, share what they learn, and ship their own improvements, all inside your own AWS account. Every agent's state is versioned in git and every action is recorded in an append-only audit log, with human approval gates where you need them. It works with any model (Claude, OpenAI, Gemini, or open weights) and is open source under Apache 2.0. Your infrastructure, your data, your agents: export everything and walk away whenever you like.
Most AI tools are rented. Whatever they build up about your business lives on someone else's platform, and it starts over when you switch vendors. Trinity is built the other way round. It is the operating system where an AI-native company's agents and workflows run with permissions, state, governance and receipts, deployed in your own AWS account so that the intelligence it builds up belongs to you.
What Trinity does
Persistent agents. Each agent runs in its own isolated container with its own filesystem, credentials and memory. It runs around the clock and keeps its context across restarts.
Works on its own. Agents run on cron schedules, webhooks and event triggers, so work gets done whether or not anyone is at a keyboard.
Reachable from anywhere. Use the web UI or REST API, or reach agents through Slack, Telegram, WhatsApp and browser voice.
Multi-agent orchestration. Orchestrator/worker patterns, parallel fan-out, shared folders and event pub/sub between agents.
Shared, not siloed. Your whole team works with the same agent and its growing context, with four-tier role-based access (admin, creator, operator, user).
Observability. Per-agent execution history, cost tracking, context usage, error rates and a fleet health dashboard.
Guardrails. Per-agent timeouts, capacity limits, read-only mode and a human-in-the-loop approval queue.
Receipts, not promises
Every agent's state is version-controlled in git, so you can roll back to any point. The audit log is append-only, keeps 365 days and records every tool call. A SHA-256 hash chain is available for compliance mode, and you enable it per install. Credentials are encrypted with AES-256-GCM and injected at runtime.
Yours to keep
Single-tenant. It runs in your VPC and your data stays in your network.
Model-agnostic. Use Claude, OpenAI, Gemini or open-weight models with your own keys.
Open source under Apache 2.0, free for commercial use and self-hosting. Export agent definitions, skills and memory at any time.
Trinity Enterprise
For teams that need more governance, Trinity Enterprise adds user management, export of the audit log to your SIEM, and an audit dashboard. It's licensed separately and held by you. Contact hello@ability.ai.
About Ability AI
Ability AI is the lab for AI-native companies, and the lab behind Trinity and Cornelius, its cognitive architecture for making a company's work compound. We run our own company on Trinity.
Highlights
Compounds, and you own it. Agents keep their memory and context across restarts. What they learn builds up as an asset in your AWS account, not on a vendor's platform.
Governed, with receipts. Agent state is versioned in git, the audit log is append-only with 365-day retention and records every tool call, and approval queues put a human gate on the actions that matter.
Runs in your account, on any model. It's single-tenant and inside your VPC, and works with Claude, OpenAI, Gemini, or open-weight models. Apache 2.0 open source with full export, so there's no lock-in.
AWS Marketplace now accepts line of credit payments through the PNC Vendor Finance program. This program is available to select AWS customers in the US, excluding NV, NC, ND, TN, & VT.
Trinity software is free. You pay only for the AWS EC2 instance you run it on, billed per hour. You pick from nine instance types in two sizes. The large sizes (m5.large, m6i.large, m7a.large, t3.large, t3a.large) carry fewer compute resources. The xlarge sizes (m5.xlarge, m6i.xlarge, m7a.xlarge, t3.xlarge) carry more. The m-series and t-series families differ in processor generation and burst behavior. Hourly cost rises with the instance you select, so you match the size to your workload. No separate license fee applies.
Top-of-mind questions for buyers
What specs do the large and xlarge instance sizes give me for running Trinity?
The large sizes carry fewer vCPUs and less memory than the xlarge sizes. Trinity's own guidance suggests about 4 vCPU and 8 GB for the platform plus its supporting services. The xlarge options give more headroom when you run several agents at once.
Am I charged when the EC2 instance running Trinity is stopped?
Hourly charges meter running instance-hours only. A fully stopped instance does not accrue hourly compute charges. Stopped instances may still incur AWS storage fees for attached volumes. The Trinity software itself carries no license fee at any state.
Why offer both m-series and t-series instance families, and how do they differ for my bill?
The m-series instances deliver steady, fixed compute each hour. The t-series instances use burstable performance, giving baseline compute with short bursts. Both bill per hour. Choose t-series for intermittent agent runs and m-series for continuous, steady workloads.
www.ability.ai
Helpful?
Vendor refund policy
Trinity is free. Ability AI charges nothing; AWS bills the EC2, EBS and IPv4 usage directly.
How can we make this page better?
Tell us how we can improve this page, or report an issue with this product.
Give us feedbackReport a problem with this product or seller
Legal
Vendor terms and conditions
Upon subscribing to this product, you must acknowledge and agree to the terms and conditions outlined in the vendor's End User License Agreement (EULA).
Content disclaimer
Vendors are responsible for their product descriptions and other product content. AWS does not warrant that vendors' product descriptions or other product content are accurate, complete, reliable, current, or error-free.
An AMI is a virtual image that provides the information required to launch an instance. Amazon EC2 (Elastic Compute Cloud) instances are virtual servers on which you can run your applications and workloads, offering varying combinations of CPU, memory, storage, and networking resources. You can launch as many instances from as many different AMIs as you need.
Version release notes
First AWS Marketplace release of Trinity. HTTPS on the instance's public IP with no domain required, admin setup gated on the instance ID, bundled PostgreSQL, and automatic certificate renewal after a stop/start or an Elastic IP change.
Additional details
Usage instructions
Launch the instance with a public IPv4 address and the recommended security group (ports 443 and 80). First boot takes a few minutes.
In the EC2 console, open Instances and copy this instance's Instance ID (it starts with i-) and its Public IPv4 address.
Open https://<public IPv4 address>/. Enter the instance ID when asked, then choose an admin password. An email address is optional; without one, sign in as
admin with that password.
Add a model credential in the browser and create your first agent. The first-run guide walks through attaching a domain and a Cloudflare Tunnel when you are ready to move off the bare IP.
Stopping and starting: without an Elastic IP, AWS gives the instance a new public address after a stop/start. Trinity picks it up and obtains a new certificate within about five minutes; open the new address after that. An Elastic IP keeps the address fixed.
Managing over SSH (optional, open port 22 from your own address only): cd /opt/trinity, then docker compose -f docker-compose.hosted.yml ps or logs -f backend, and sudo ./scripts/deploy/start.sh --hosted to restart. The database is a bundled PostgreSQL container. Backups run nightly under /opt/trinity/trinity-data/backups/; take EBS snapshots as well.
AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.
Trinity is an AI safety, evaluation, and observability platform that helps enterprises build reliable, secure, and production-ready AI and GenAI applications by monitoring prompts, responses, guardrails, and model performance.
Tacnode is a unified cloud native and cloud agnostic data platform built for massive scale. The first database that unifies real-time analytics, online retrieval, and PostgreSQL compatibility in one platform, which handling large-scale data and complex workloads with unprecedented performance, at real-time.
This is a repackaged software product wherein additional charges apply for a pre-hardened, SI Core STIG Hardened image and seller support. Debian 13 is designed for robust and secure cloud environments, providing a stable and reliable operating system tailored for EC2 instances. With its enhanced security features, Debian 13 incorporates strict security protocols that align with industry standards, making it ideal for sensitive workloads. The pre-hardened image ensures compliance with security regulations, streamlining deployment within enterprise environments. Suitable for developers, system administrators, and organizations looking to implement secure cloud solutions, Debian 13 simplifies application hosting, web services, and database management with unparalleled performance and efficiency.
The Clinical Trials Registry and Results Database compiles information on publicly and privately supported clinical trial studies on a wide range of diseases and conditions. Its main goal is to provide an easy access to both privately and publicly funded clinical trials information for patients, their family members, healthcare professionals, researchers, and the public.
Be the first to review this product. We've partnered with PeerSpot to gather customer feedback. You can share your experience by writing or recording a review, or scheduling a call with a PeerSpot analyst.