> For the complete documentation index, see [llms.txt](https://docs.finout.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.finout.io/user-guide/optimize/finout-investigation-agent-beta.md).
# Finout Investigation Agent \[Beta]
Finout's Investigation Agent turns a detected cost finding into a decision-ready recommendation. When you open an finding on the Finding page for a supported finding type, Investigation Agent automatically reviews the resource's cost and usage data and builds two things for you on the spot: reasoning that explains what it found and why, and a next step that tells you what to do about it - with a way to act on it directly from the finding.
### **Investigator analysis: the logic behind the recommendation**
At the top of every investigation, Investigation Agent shows a short, AI-generated explanation of what it looked at and what it concluded. This reasoning always covers the five things:
* What was investigated
* What was found
* The key data points that support the finding
* The estimated cost impact
* Anything you should validate before acting on it
Investigation Agent reasons only over the cost, usage, and data Finout already has for that resource - it does not invent metrics, owners, or savings figures. If a data point is missing, the reasoning tells you it's missing rather than filling in a guess.
#### Example: Rightsizing
> Instance `i-1111112222223333` (`r8g.16xlarge`, 64 vCPU / 512 GiB) on `AmazonEC2` shows a CPU peak of 1.77% over the 7-day observation window. Downsizing to `r8g.large` (2 vCPU / 16 GiB) would reduce estimated monthly amortized cost from $2,714.57 to $84.83, a potential saving of $2,629.74/month. Memory utilization data is unavailable and requires validation before resizing
### **Next step: what to do about it**
Below the reasoning, Investigation Agent gives you one short, action-oriented sentence telling you what should happen next - always based only on the data available for that finding, never a generic suggestion.
Next to it, you'll find multiple buttons tied directly to that recommendation:
* Share with Owner - generates a short, copyable message summarizing the finding, the recommended action, and the estimated savings, so you can send it to the resource owner to validate or simply keep them informed. You don't need to re-explain the investigation yourself - the message carries the context with it.
* Draft Ticket - generates ticket-ready content for your ticketing tool, including a title, a description of the proposed change, the justification, implementation steps, and the main risk to be aware of.
#### Example: Share with Owner message (Rightsizing)
> Hi team - Finout's Investigation Agent flagged `i-1111112222223333` instance for rightsizing. Utilization has averaged well below capacity over the past 30 days, and downsizing from 32 vCPU / 128 GiB to 8 vCPU / 32 GiB would save an estimated $840.96/month (Amortized Cost: $1,121.00 → $280.00). Can you confirm there's no upcoming workload change that needs the current capacity before we schedule this?
### **FAQs**
\
**Does Investigation Agent take any action automatically?**
No. Investigation Agent only generates recommendations and copyable content - the Share with Owner message, and Draft Ticket content. It never changes a resource, sends a message, or creates a ticket on your behalf; you review and send that content yourself.
**Which finding types and providers does it support today?**
Rightsizing and Idle resource findings, across AWS, Azure, and GCP. Investigation Agent is in Beta, and support for more finding types is on the roadmap.
**What if there isn't enough data to fully support a recommendation?**
Investigation Agent still gives you a recommendation, but it tells you exactly what's missing - for example, unavailable memory utilization data, or an unconfirmed resource owner - so you know what to check before acting, rather than being asked to just trust the output.
**Can Investigation Agent invent data it doesn't have?**
No. It only reasons over the cost, usage, and metadata Finout already has for that resource. It never fabricates metrics, owners, costs, or implementation steps to fill a gap - if something's missing, the reasoning says so.
**What if I want to dig deeper than what's shown?**
Today, you can navigate to the Billy page to ask follow-up questions about the resource.
**What cost basis are the savings shown in?**
Cost savings are calculated based on your account’s default cost type. Every savings figure comes with a cost type badge (for example, Amortized Cost) so you always know which cost basis the numbers are based on.
**Can I send the ticket straight to Jira or ServiceNow?**
Not from Investigation yet. **Draft ticket** produces ready-to-use text (Summary, Changes, Justification, Implementation steps, Risks) that you copy into your team's tool. Finout doesn't create or send anything on your behalf. The ability to automaticlly open a ticekt on your bhealf is comming soon
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.
Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:
```
GET https://docs.finout.io/user-guide/optimize/finout-investigation-agent-beta.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
