EstimAI — Support Documentation

Quick start (2 minutes)

Step 1: Install

Install EstimAI from the Atlassian Marketplace. It will appear in your Jira sidebar as “EstimAI”.

Step 2: Get an API key

EstimAI uses your own AI provider key — you pay directly to OpenAI or Anthropic at their rates.

OpenAI (recommended):

  1. Go to platform.openai.com/api-keys
  2. Create a new key
  3. Copy it (starts with sk-...)

Anthropic (alternative):

  1. Go to console.anthropic.com
  2. Create a new API key
  3. Copy it (starts with sk-ant-...)

Step 3: Configure

  1. Open EstimAI → click ⚙ Settings
  2. Select your AI provider
  3. Paste your API key
  4. Click “Save settings”

Step 4: Estimate

  1. Open any Epic or issue in Jira
  2. Or navigate to EstimAI in the sidebar
  3. Enter the issue key (e.g., KAN-4) or paste the full URL
  4. Optionally add design screenshots, Figma link, or grooming notes
  5. Click “Estimate story points”

Features guide

Loading issues

EstimAI supports all Jira project types:

Enter an issue key (KAN-4), paste a full Jira URL, or let EstimAI auto-load from the current issue context (when used as an Issue Panel).

Design screenshots

Upload up to 5 screenshots of your design. AI analyzes UI complexity, number of components, and interaction patterns.

Supported: PNG, JPG, WebP. Images are resized to 1200px max before sending.

Figma integration

  1. Go to figma.com → Account → Personal access tokens → Create new token
  2. In EstimAI Settings → add your Figma Personal Access Token
  3. Paste any Figma frame URL into the “Prototype link” field
  4. Click ↓ to auto-fetch the screenshot

Note: The URL must include a node-id parameter (right-click a frame in Figma → Copy link).

Grooming transcript

Paste your meeting notes, Grain/Notion AI transcript, or manual notes. EstimAI extracts technical decisions, risks, and constraints to improve estimate quality.

Sub-tasks and platform dropdown

Each sub-task has a platform label (Frontend, Backend, iOS, Android, Desktop, QA, DevOps). This is used to:

When an Epic with Jira sub-tasks is loaded, they auto-populate with AI-detected platform labels.

Calibration

EstimAI fetches completed tasks from your project (last 90 days by default). These appear in each result card as historical anchors — real tasks your team completed, with actual hours logged.

Calibration window can be adjusted in Settings.

Planning Poker feedback

After estimating:

  1. Run your team’s Planning Poker session
  2. Select the team’s agreed SP for each task
  3. Click “Save feedback”

EstimAI tracks AI accuracy over time. The more you use it, the better the calibration becomes.

FAQ

Q: Why do I need to provide my own API key? A: EstimAI is “bring your own key” — you connect directly to OpenAI or Anthropic. This means your data never passes through our servers, and you pay AI costs at provider rates (typically $0.01–0.05 per estimation).

Q: Is my data safe? A: Yes. EstimAI runs on Atlassian Forge — all data stays within Atlassian’s infrastructure. Your API keys are encrypted per-installation. We never store issue content or AI responses after the session. [Full Privacy Policy]

Q: Which AI provider should I use? A: Both work well. GPT-4o mini (OpenAI) tends to be slightly faster and cheaper. Claude Haiku (Anthropic) produces more detailed analysis text. You can switch between them in Settings anytime.

Q: Why is my sub-tasks count showing 0? A: EstimAI tries multiple strategies to find child issues. If it still shows 0:

Q: The Figma screenshot isn’t loading — what’s wrong? A: Common causes:

  1. Token not saved — go to Settings and re-save your Figma token
  2. Wrong URL format — right-click a specific Frame in Figma → Copy link (URL must include node-id=)
  3. No access — ensure your Figma token has access to the file

Q: Can I use EstimAI for Linear / Asana / Notion? A: Currently EstimAI is Jira-only. You can still use it manually: enter a task description in the “Epic description” field without loading from Jira.

Q: How accurate are AI estimates? A: Accuracy improves over time as calibration data builds up. In our testing, after 20+ feedback sessions, AI matches team estimates within ±1 SP about 80% of the time for well-described tasks.

Q: Does EstimAI work on Jira Data Center / Server? A: No. EstimAI is built on Atlassian Forge and supports Jira Cloud only.

Pricing

EstimAI is free with unlimited AI estimations — all features included (vision, Confluence context, historical calibration, Figma). It is bring-your-own-key: you connect your own OpenAI or Anthropic key and pay only your provider’s usage; EstimAI itself is free.

Contact & support

Email: support@estimai.org
Response time: Within 24 hours on business days

For bug reports, please include: