AI Ops

AI Ops is a set of nullplatform features that take recurring operations work off your plate: things like catching an outdated runtime before it becomes an incident, holding a risky deploy until the right checks pass, or following a migration across every team. Everything runs on your own data and under the rules your organization defines.

🚀 Early release

AI Ops features are in early release and may not be available in your organization yet. If you'd like to try them or want to know more, reach out to us and we'll get you set up.

What's included

Each feature works on its own: you can gate deploys with checklists without ever writing a workflow, or use Insights without action items. They also combine well. For example, to roll out an AMI update:

1. A workflow scans your scopes on a schedule, finds the ones running the outdated image, and opens an action item for each one, assigned to the team that owns it.
2. A checklist gates the deploy that applies the fix: it checks rules, asks an external system, or waits for a human sign-off before the deploy runs.
3. An Insights dashboard shows how many scopes are still pending across the organization.

You build each piece the same way: install the AI plugins in your coding assistant, describe what you need in plain language, review the result in the UI, and activate it.

Set up the AI pluginsInstall the nullplatform plugins in your coding assistant (Claude Code, Kiro, Cursor, and others), configure your API key, and run a first prompt. Every other feature is built from here.WorkflowsAutomations that run on a schedule, on a platform event, or on demand. They connect action items, the data lake, notifications, agents, and your own tools, with AI steps where a decision needs judgment.ChecklistsApproval gates made of rules nullplatform evaluates, external checks answered by a workflow, and human sign-offs. Attach them to the actions you want to control, such as production deploys.Action itemsRecords of a finding: what's wrong, on which entity, who owns it, and an AI-suggested fix the team can review and apply in one click.InsightsDashboards built from a question in plain language, backed by the data lake. Save and share them so the whole organization works from the same data.AI assistantAsk questions, investigate failures, and generate CLI commands or configurations from the console, using the context of the page you're on.

Ready-to-use suites

Nullplatform runs these suites on its own production environment. Each one is open source in nullplatform/workflows and comes with a tutorial of about 30 minutes.

Suite	What it does	Start here
Cost intelligence	Prices every Kubernetes scope daily, opens right-sizing suggestions where it finds waste, and verifies the savings after the change	Right-size Kubernetes workloads
Runtime lifecycle	Keeps a catalog of AWS Lambda runtimes from the AWS docs and opens a migration item per affected scope	Migrate deprecated Lambda runtimes
AMI rollout	Detects scopes running an outdated image and tracks the rollout with action items that close themselves when the scope is updated	Detect and roll out AMI updates
Smart checklists	Gates deploys on a Jira ticket, an AI architecture audit, or a risk score that decides who approves	Gate deploys on a Jira ticket, Audit every deploy against your architecture rules, Match approval friction to deploy risk
Deploy analysis and library inventory	Scores the risk of every deployment and records what each deployed asset depends on	The deploy-analysis and lib-inventory suites in nullplatform/workflows

Learning paths

From hello world to a working AI Ops suite

1. 1Set up the AI pluginsInstall the plugins in your coding assistant and run a first prompt.10 min
2. 2Your first workflowDescribe a hello world workflow, run it, and activate it.10 min
3. 3Roll out AMI updatesA nightly scan that opens an action item per outdated scope and closes it once the scope is updated.25 min
4. 4Right-size Kubernetes costsPrice every Kubernetes scope daily and open right-sizing suggestions the team applies in one click.30 min

Gate deploys with checklists

1. 1Your first checklistDescribe a two-item gate, dry-run it, and attach it to a real deploy.10 min
2. 2Gate deploys on a Jira ticketBlock a deploy until its linked Jira ticket is approved.20 min
3. 3Match friction to deploy riskScore the risk of each deploy and require more approvals for riskier ones.25 min
4. 4AI architecture complianceAn agent audits every deploy against your architecture rules before it runs.25 min

Building blocks

Workflows and checklists are built on these features. Each one works on its own, and most workflows use at least one of them.

Data lakeQuery the state of your entire organization with SQL: applications, deployments, scopes, approvals, and more, all in one queryable place.ApprovalsDefine policies that require a human sign-off before sensitive actions run, like production deployments or secret access.NotificationsRoute platform events to Slack, webhooks, or your own services, and filter them so each team only gets what it needs.Entity hooksReact to entity changes with your own logic: enforce naming rules, update repositories, or trigger external systems when something changes.

Also see AI usage and quotas to monitor AI consumption across your organization.