Cheaper than one hour of downtime
One plan, one price per workspace. Every feature for your whole team,
with $25 of AI included every month.
- Every feature, for everyone on the team
- Halon, the AI SRE, with $25 of AI every month
- Top up, or bring your own AI key
- 14-day free trial
- The same code we run, AGPL-3.0
- Every feature, nothing held back
- Halon on your own AI keys
- No plan, no trial, no bill
The 14-day trial includes every feature and the $25 of AI. Your card is taken when it starts and first charged when it ends, unless you cancel before then. Prices are in US dollars, plus tax.
$25 of AI, every month
Halon runs on the AI your plan includes. Top up with credits
when you need more, or bring your own AI key. Here is what uses it.
- InvestigationsHalon works an incident across your logs, metrics, deploys, and code, and reports what it found.
- QuestionsMention @Firefight in an incident channel or chat with Halon in the dashboard for an answer grounded in your records.
- Catch-upRun /ff catchup in an incident channel for a summary of what has been tried, decided, and blocked.
- Postmortem draftsA full postmortem written from the timeline and the channel, ready to edit, and any passage rewritten on request.
- Timeline notesWhen an incident closes, Halon reads the channel once and writes what the team tried, found, and decided onto the timeline.
- Renews monthly, top-ups stay
The $25 of AI renews with each billing period and does not carry over. Halon spends it before anything you bought, and credits you buy never expire.
- AI pauses at zero, nothing else does
Your admins hear by email and in Slack when 80% is used and again at zero. Declaring, updating, and resolving incidents never use AI.
- Bring your own AI key
Add your own Anthropic, OpenAI, Google Gemini, or other provider account. Halon uses it first, and your provider bills you directly.
Everything you need to run incidents well
No feature gates. No add-ons.
One price covers everything, for everyone on the team.
Incident management
- Declare and resolve in SlackRun /ff new to open a channel, bring in responders, and start the clock without leaving Slack.
- Incident rolesAssign an Incident Lead and bring in responders from Slack. Everyone knows their job from the first minute.
- Actions and follow-upsCreate action items mid-incident. They survive the postmortem and stay tracked until they're closed.
- Configurable severities and statusesDefine your own severity levels (P1, P2, SEV-1, however your team labels them) and lifecycle statuses.
- Automatic incident timelineEvery state change, role assignment, and status update is captured automatically. No manual logging during a P1.
- RunbooksAttach runbooks to severities, incident types, or services. The right one shows up the moment an incident matches.
- Postmortem editorRich-text editor with revision history and restore. Export as PDF or Markdown.
- Service catalogDefine your services, teams, and ownership in one place. Reference catalog entries on incidents.
- Incident metricsAverage resolution time, active and critical incident counts across your workspace.
- Slack channel managementChannels are created, named, and archived automatically. No manual housekeeping after an incident closes.
Halon, the AI SRE
- InvestigationsHalon works an incident across your connected tools and tells you what it found and where it found it.
- Answers in Slack and the dashboardMention @Firefight in an incident channel, or chat with Halon in the dashboard.
- Catch-up summariesRun /ff catchup for a summary of what has been tried, decided, and blocked, so anyone joining late is current in seconds.
- AI postmortem draftsHalon drafts the postmortem from the timeline and the channel, and rewrites any passage when you tell it how.
- Similar incident searchPast incidents that match what you're seeing right now, with their timeline and resolution.
Platform and integrations
- IntegrationsMore than 50 tools, including GitHub, GitLab, Datadog, Grafana, Sentry, PagerDuty, AWS, Google Cloud, Azure, Kubernetes, Linear, Jira, and Notion.
- MCP server for AI agentsConnect Claude or any agent to your workspace over MCP to investigate incidents and act on them. Anything that speaks MCP can be connected as an integration too.
- Agent permissions and approvalsEvery agent action goes through per-agent grants, human approval for anything high risk, and a record of every attempt.
- Webhooks and REST APISend incident events out to any tool, and declare incidents from any source through the API.
Common questions
What happens after the 14-day trial?
The trial includes every feature and the $25 of AI. You add a card to start it, and the card is first charged $99 plus tax when the trial ends, then every month. Cancel before the trial ends and nothing is charged. If the first payment does not go through, the workspace pauses until an admin updates the card, and nothing is lost.
Does everyone on the team need a seat?
No. The price is per workspace, not per person. Everyone on your team can declare incidents, take roles, post updates, and ask Halon, and none of it changes what you pay.
What does the $25 of AI cover?
Everything Halon does on FireFight’s AI: investigations, answers in Slack and the dashboard, catch-ups, postmortem drafts, and timeline notes. It renews with each billing period and does not carry over. Beyond it, Halon uses credits you top up, which never expire, or you can bring your own AI key.
What happens when the included AI runs out?
Your admins hear about it first, by email and in Slack, when 80% is used and again at zero. If you have bought credits, Halon carries on with them. If not, Halon pauses until you top up or the included AI renews. Declaring, updating, and resolving incidents never use AI, so they keep working.
How do we top up?
An admin buys a prepaid credit pack under Settings, Billing. Credits you buy never expire, and Halon spends the included AI first, so they stay put until you need them.
Can we use our own AI key instead?
Yes. An admin adds an account with Anthropic, OpenAI, Google Gemini, Amazon Bedrock, OpenRouter, or another supported provider under Settings, Workspace, in AI accounts. Halon uses your accounts first, your provider bills you directly, and FireFight charges nothing for that usage. The included AI and any credits are used only when none of your accounts can answer.
Can I cancel anytime?
Yes. No annual contract and no cancellation fee. An admin cancels under Settings, Billing, and the workspace keeps working until the end of the period already paid for. After that it pauses, and your data is kept for 90 days in case you come back, then deleted for good.
Do you offer annual billing?
Not yet. Billing is monthly.
What integrations are included?
All of them, at no extra cost. FireFight connects to more than 50 tools, including GitHub, GitLab, Datadog, Grafana, Sentry, PagerDuty, AWS, Google Cloud, Azure, Kubernetes, Linear, Jira, and Notion. Anything else that speaks MCP connects the same way, and webhooks and a REST API cover whatever is left.
Can I self-host FireFight instead?
Yes, for free. FireFight is AGPL-3.0, and the code you run is the code we run, with every feature. Halon runs on your own AI keys, and there is no plan, trial, or bill.
Install before another incident
14-day free trial. $25 of AI included. 5-minute setup.
Start free trial