What is Shuddhi QA
From requirement to
test suite in 30 seconds
Describe your business flow or upload an FDD — Shuddhi QA generates a complete, structured test artifact that QA engineers can use immediately.
1// Input: Business Flow
2platform = "D365 F&O"
3module = "Accounts Payable"
4flow = "Vendor invoice 3-way match...
5
6// Output: Complete Test Suite
7generate(platform, module, flow)
✅ 8 E2E Test Cases
❌ 5 Negative Scenarios
🔢 6 Field Validations
🔗 4 Integration Tests
Upload any document
PDF, DOCX or TXT — FDDs, BRDs, user stories, Confluence pages. Auto-detects language in the document. Output available in 5 languages.
AI generates structured TCs
Your AI provider (Claude, Gemini, Groq, OpenAI, Together AI or Azure OpenAI) produces 5-column test cases — Title, Step Action, Expected Result, Assigned To, State — ready for ADO or Jira import.
Export in 9 formats
Excel, PDF, Markdown, Jira Zephyr CSV, ADO Test Plan push, TestRail CSV, Gherkin (.feature), Playwright TypeScript, Postman collection. Export All in one click.
Push to Azure DevOps
3-step wizard creates Test Plans, parent + child suites (E2E / Field / Integration / Negative), and links all test cases with step XML.
Core Features
Everything a QA team needs
9 Export Formats
Excel, PDF, Markdown, Jira Zephyr CSV, ADO Test Plan push, TestRail CSV, Gherkin (.feature), Playwright TypeScript, Postman collection. One-click "Export All" downloads everything.
Jira & ADO Auto-fetch
Enter a Jira ticket ID or ADO Work Item ID — Shuddhi QA fetches the summary, description, and acceptance criteria and auto-fills the business flow. Saves 5+ minutes per generation.
Vision & Screenshot Upload
Upload UI screenshots, mockups, or wireframes — the AI analyses visual elements and generates test cases for buttons, fields, validations, and user flows shown in the image.
Multi-turn Refinement
Continue the conversation after generation — ask "add 3 more negative cases", "make TC005 more detailed", or "convert to Salesforce-style". Conversation history maintained per session.
Real-time Streaming
Test cases appear word-by-word as your AI provider generates them. Stop mid-generation and keep partial output. AbortController wired to every request.
Inline Editing
Click any table cell to edit directly in the browser. Done Editing syncs changes back to raw markdown so all exports reflect your edits. Discard restores the original.
5-Language Output
Auto-detects input documents in 22+ languages and generates output in 5 supported languages — English, German, Spanish, French, Portuguese (BR). More languages coming soon.
Google Cloud Billing Sync
Connect your Google account with one tap to see live Gemini billing account details — account name, status, linked project, budget caps, and threshold alerts. USD costs auto-converted to INR with live exchange rate.
Smart Detect — AI Platform & Domain
One click analyses your requirement using a 4-phase enterprise intelligence engine — identifies the platform (D365, Guidewire, SAP...), business domain (Insurance, Banking, HR...), compliance obligations (IRDAI, SOX, HIPAA), and key stakeholders. Automatically selects the right platform and domain chips so you can generate immediately.
Auto-Detection from Text
Paste any requirement and keyword detection instantly identifies the platform and industry domain — no button click, no manual selection needed. 200+ enterprise signals covering Insurance, Banking, Healthcare, ERP, HR, Retail, ITSM, and Web/Mobile.
6 AI Providers — Free & Paid
Choose from Claude (Sonnet + Haiku), Gemini, Groq (free), OpenAI (GPT-4o mini), Together AI (DeepSeek V3), and Azure OpenAI (enterprise BYO). Mix and match — use Groq for free exploration, Claude Haiku for cost-effective enterprise, Claude Sonnet for maximum quality.
Live Cost Indicator
See the estimated cost per generation before clicking Generate — shown only for Claude (paid API). Switch between Claude Sonnet (~$0.58/full run) and Claude Haiku (~$0.15/full run) with one click. Your $5 lasts 8 runs on Sonnet or 33 runs on Haiku.
Shareable Links
Share any generated test case set with a permanent link — anyone with the link can view the full output including platform, module, and all test cases. Great for stakeholder reviews, QA handoffs, and sprint demos. Stored securely in Firestore.
Enterprise Quality Guidance
When an enterprise platform is detected (D365, Guidewire, SAP etc.), a contextual banner explains why Claude produces significantly better results for enterprise workflows — real field names, domain-specific steps, accurate approval chains.
AI-Review Gate
Before pushing to Azure DevOps or downloading the Jira CSV, you must explicitly tick "I've reviewed these test cases." The Push button stays disabled until confirmed — preventing 50 hallucinated TCs from accidentally landing in your sprint. Accountability stays with the human reviewer, not the AI.
Domain Intelligence — 6 Verticals
Layer business-domain knowledge on top of any platform. Pick Insurance, Banking, HR, ERP, Retail, or Healthcare — the AI adds vertical-specific terminology (FNOL, KYC, HL7, PHI) and tags compliance-sensitive scenarios so you don't miss audit-critical paths. Mix any platform with any domain: Salesforce + Insurance produces underwriting and policy-binding test cases automatically.
Quality Scoring per Test Case
Every generated test case gets an AI confidence score — Good (90-100%), Fair (70-89%), or Needs Review (<70%). Colour-coded badges let you prioritise which TCs to review first. Pair with the AI-Review Gate to focus eyes on the Fair/Needs Review ones before pushing to ADO.
PWA Install & Power Shortcuts
Install Shuddhi QA as a desktop or mobile app — launches in its own window, works offline for previously generated TCs. Power users get a command palette (Ctrl+K) to jump anywhere instantly, plus shortcuts for Generate (Ctrl+Enter), Theme Toggle (Ctrl+Shift+T), and Settings (Ctrl+,).
Supported Platforms
14 platforms · ERP, CRM, ITSM, Insurance — all in one
Smart Detect auto-identifies the platform from your requirement. Platform and domain chips auto-select so you jump straight to generation. Switch platforms and the AI adapts completely — no ERP jargon in web app tests, no REST API language in D365 tests.
D365 Finance & Ops
AP · AR · GL · Fixed Assets · Procurement · Inventory · 7 more
D365 CRM / CE
Leads · Opportunities · Cases · Activities · Marketing
D365 Business Central
Purchase · Sales · Finance · Manufacturing · Service
Salesforce Sales Cloud
Lead-to-Opportunity · CPQ · Forecasting · Reports
Salesforce Service Cloud
Cases · Knowledge · SLAs · Omni-Channel · Live Agent
Salesforce CPQ
Product Catalog · Price Rules · Quotes · Contracts
General Web App
Auth · RBAC · Payments · API Integration · Admin Panel · 10 more
Mobile Application
Biometric Auth · Push · Offline · In-App Purchases · GPS
See it in Action
From description to test suite in 30 seconds
Type a business process, upload an FDD, or click a sample chip — Shuddhi QA returns a full structured test artifact instantly.
How it works
Four steps to your test suite
1
Select Platform & Module
Choose from D365, Salesforce, Web App, or Mobile. Module list updates automatically.
2
Describe or Upload
Type your business flow, paste a user story, or upload a PDF/DOCX FDD. 5 output languages supported with auto-detection of input language.
3
Generate & Edit
Your AI provider streams test cases in real time. Click any cell to edit. Discard restores the original.
4
Export or Push to ADO
Download Excel, PDF, or Jira CSV — or push directly to an Azure DevOps Test Plan with one click.
Security First
Your keys stay in your browser
Your API keys (Groq, Gemini, Claude) are stored only in your browser's localStorage — never on our servers. ADO PAT and Jira tokens are also browser-only. The Cloudflare proxy never logs or stores your credentials.
✓ API keys stored in browser localStorage only
✓ No eval() or new Function()
✓ ADO URL allowlist enforced
✓ CORS restricted to Cloudflare Pages origins
Your Groq / Gemini / Claude key
localStorage only
↓ sent only with your request
Browser → /api/claude
Cloudflare proxy
↓ forwarded to AI provider
/api/claude → Groq / Gemini / Claude
SSE stream
Your ADO PAT / Jira token
localStorage only
↓ sent only with your request
Browser → /api/ado
Cloudflare proxy
↓ forwarded to Azure DevOps
/api/ado → Azure DevOps
REST API
👩💻 BUILT BY A QA ARCHITECT, FOR QA ENGINEERS
"I spent 4 hours writing test cases for a D365 invoice flow.
Now it takes 18 seconds."
I'm Ramya — Senior QA Architect with 10+ years of enterprise testing experience across D365 F&O, Salesforce, Guidewire and Playwright. I built Shuddhi QA because I was tired of writing the same test cases manually for every project.
I used Shuddhi QA on a real enterprise engagement — generating and managing 160+ test cases for a major US museum's ERP integration project. The tool saved my team an estimated 21 hours of QA effort on that single engagement alone.
R
Ramya
Senior QA Architect · Bengaluru
D365 F&O
Salesforce
10+ Years QA
"
Used on a real ERP integration project. Generated 160+ test cases covering Accounts Payable, Purchase Orders and GL workflows — with actual D365 field names and navigation paths. Saved the team an estimated 21 QA hours.
Ramya · Senior QA Architect
Enterprise ERP Integration Project · 2024
"
The platform auto-detection is genuinely impressive. I pasted a Salesforce Lead-to-Opportunity flow and it generated 38 test cases with SOQL-specific validation steps — no manual setup needed.
Beta Tester · Senior SDET
Salesforce Sales Cloud · CRM Project
100+
QA Templates
D365 · Salesforce · Web · Insurance
ITSM · HR · Healthcare · Omni-Channel
API · Security & Data
Try Free →
ITSM · HR · Healthcare · Omni-Channel
API · Security & Data
Get Started
Get started in 30 seconds
No deployment needed. Use Groq for free with no API key — or bring your own Claude, Gemini, OpenAI, Together AI or Azure OpenAI key.
⚡ Groq — No Key Needed
or
✦ Gemini API Key (free)
or
🤖 Claude API Key (paid)
or
⚫ OpenAI API Key (paid)
or
🟣 Together AI API Key (paid)
or
☁️ Azure OpenAI (enterprise BYO)
=
✅ Ready to generate
FAQ
Frequently asked questions
Everything you need to know before getting started.
Yes. Shuddhi QA is free forever. New users get 3 free generations immediately (no signup) using a shared Groq server key. After that, sign in with Google or Microsoft for 10 more free generations, or add your own API key for unlimited use. Groq is 100% free with your own key (no credit card). The app itself never charges.
Yes — available now via Model Context Protocol (MCP). Your tenant admin opens Copilot Studio, creates a Declarative Agent, and adds our MCP server URL:
https://shuddhiqa-mcp.ramya9-b.workers.dev/mcp. Once published, anyone with a Microsoft 365 Copilot license can invoke Shuddhi QA from inside Word, Excel, Teams, or Outlook by typing @Shuddhi-QA. Full setup walkthrough at Help → Copilot / MCP Integration.
Any MCP-capable AI client: Claude Desktop, Cursor, Continue (VS Code / JetBrains), and a growing list of community clients. Setup is the same paste-the-URL pattern as Microsoft Copilot. The MCP server exposes five tools:
generateTestCases, fetchJiraIssue, pushToADO, listTemplates, detectPlatform. See setup guide.
Shuddhi QA itself is free. You need a Microsoft 365 Copilot license to invoke any third-party agent from Copilot — that's Microsoft's standard requirement ($30/user/month). The AI generation happens via our free Groq tier by default; if you want higher quality, supply your own Claude / OpenAI / Gemini key per request. We never charge for use of the MCP server.
Typically 10–30 seconds for a single-section format (End-to-End, Field Validation, Negative, etc.). Full Suite generation across all 8 sections takes 1–3 minutes for complex enterprise requirements. Test cases stream in live as they're generated, so you can start reading TC001 before TC002 has finished. Groq is fastest (~10s); Claude produces the highest quality but takes 20–40s.
Start with Groq — it takes 30 seconds, requires no credit card, and is completely free. Go to console.groq.com/keys → sign in with Google → Create API Key → copy it → paste in Settings → AI Provider. Done. See the full setup guide.
Your API key is stored only in your browser's localStorage — never on our servers. It is sent directly to the AI provider (Groq, Gemini, or Claude) with each request and never logged, stored, or visible to anyone else. You can clear it anytime from Settings → AI Provider → Clear all saved keys.
Google shows this warning for apps requesting billing access that are awaiting their verification review. Shuddhi QA has submitted for verification — the review takes 4–6 weeks. The warning is cosmetic only — the app safely reads only your billing account name, status, and budget using
cloud-billing.readonly scope. No data is modified. You can safely click "Continue" or skip billing sync entirely — it is optional.
Depends on the Output Format you pick. All Sections generates 40+ test cases across E2E, Field Validation, Integration, Negative, Smoke, and Test Data (target 50+). Single-section formats (End-to-End / Field Validation / Integration / Negative / Smoke) generate 8–15 focused test cases of that one type. Test Data Format produces one comprehensive data table. Every test case has minimum 10 numbered steps.
AI models can occasionally generate inaccurate steps — this is a known limitation of every LLM, not specific to Shuddhi QA. That's why we built the AI-Review Gate: before pushing to Azure DevOps or downloading a Jira CSV, you must explicitly tick "I've reviewed these test cases and they're ready to push." The Push button stays disabled until you do. This keeps accountability with the human reviewer — not the AI — and prevents 50 fabricated TCs from accidentally landing in your sprint. Every TC also has a quality score (Good 95%, Fair 75%) so you know which ones to read more carefully.
No — sign-in is optional. Anonymous visitors get 3 free generations immediately. Signing in with Google or Microsoft unlocks 10 more free generations and syncs your settings across devices. We only access your name and email — no Gmail/Drive/Calendar access, no posting on your behalf. You can delete your account and data anytime from Settings → Account.
Yes. Configure your ADO PAT (Personal Access Token) in Settings → Azure DevOps. After generating, click the ADO button → tick the AI-Review acknowledgement checkbox → select your project, test plan, and suite → Push. Test cases are created in ADO with full step XML — Title, Step Action, Expected Result, Assigned To, and State. Note: ADO Test Plans requires the "Basic + Test Plans" license — your "Edit Cloudflare Workers" PAT scope must include Test Plans (R&W). See how to create a PAT.
Yes. Click ✏️ Edit in the export toolbar to enter inline edit mode. Every cell in the test case table becomes clickable — change the title, step action, expected result, assigned to, or state. Changes are reflected instantly in all export formats (Excel, PDF, ADO push, Jira CSV, etc.). Edits are stored in your browser's localStorage and persist across sessions.
Yes — that's our Iterative Refinement feature. Hover any test case row → click the 🔄 regenerate icon that appears on the right. A modal opens showing the original TC title. Add optional context (e.g., "Focus on boundary conditions" or "Add steps for currency conversion") and click Regenerate. Only that one test case is updated — the rest of the suite stays intact. Perfect for tweaking one weak TC without losing your edits to the others.
14 platforms — Microsoft D365 F&O, D365 CRM/CE, D365 Business Central, Salesforce (Sales/Service/CPQ), SAP S/4HANA + ECC, ServiceNow ITSM, Guidewire (PolicyCenter, ClaimCenter, BillingCenter), Workday HCM, Oracle ERP, Power BI, generic Web App, and Mobile App. Plus 6 business-domain layers (Insurance, Banking, HR, ERP, Retail, Healthcare) that overlay domain-specific terminology and risk patterns. Each platform has a tailored AI prompt with real field names, navigation paths, and module-specific test patterns.
Yes. Enter your Jira ticket ID (e.g.
PROJ-1234) in the Jira field and click Fetch → — the summary, description, and acceptance criteria auto-fill the Business Flow field instantly. Same for Azure DevOps — enter a Work Item ID (e.g. 1234) and click Fetch →. Configure credentials in Settings → Jira or Azure DevOps tabs. See the setup guide.
No. Your uploaded documents and flow descriptions are sent directly to your AI provider (Groq, Gemini, or Claude) and are subject to their privacy policies. Shuddhi QA does not store, log, or retain your requirement documents on any server. Generated test cases are saved only in your browser's localStorage (last 20 sessions).
Groq is an AI inference platform that runs Llama 3.3 70B — one of the fastest AI models in the world. For Shuddhi QA, Groq is recommended because it is completely free (30 requests/minute, 14,400/day), requires no credit card, and delivers test cases in 10–20 seconds. Sign up at console.groq.com in 30 seconds using your Google account.
Test cases can be generated in 5 languages: English (default), German, Spanish, French, and Portuguese (BR) — all with full QA translation. Upload your FRD in any language — Shuddhi QA auto-detects the input language and generates output in your selected language. 4 additional languages (Japanese, Dutch, Italian, Polish) are listed as "Coming Soon".
Shuddhi QA supports 9 export formats: Excel (.xlsx with separate sheets per TC type), PDF (formatted A4 landscape), Markdown (copy to clipboard), ADO Test Plan push, Jira Zephyr CSV, TestRail CSV, Gherkin (.feature file), Playwright TypeScript, and Postman collection. Access all formats via the export buttons and the More ▾ dropdown after generation.
