For Product Managers, Support Engineers, QA & Technical Writers

Stop Waiting on EngineersFor Answers About Your Product

Understand your product's codebase without reading code. Get answers in plain English. Write better specs, resolve support tickets faster, and ship with confidence.

Plain EnglishNo coding requiredInstant answersEvidence trails

Free and open source. Business and Enterprise plans available.

What changes for your team
Before
Days waiting for developer answers
Incomplete specs missing edge cases
Hours triaging support tickets manually
After
Self-serve answers in seconds
Engineering-grade specs with test cases
Automated ticket investigation and triage
Built for non-technical users
Instant Accurate Plain English

Why non-technical teams choose Probe

Understand Without Code

Ask questions about your product in plain English. Get accurate answers grounded in the actual codebase — no technical knowledge required.

Write Better Specs

Generate specs with constraints, dependencies, and edge cases automatically pulled from the codebase. Engineers can execute them immediately.

Resolve Tickets Faster

Investigate customer issues without waiting on engineering. Understand what changed, which components are affected, and what the fix should look like.

Keep Docs Fresh

Documentation that stays synchronized with code. When features change, you know immediately and can update docs before customers notice.

The Real Problem

You shouldn't need a CS degree to understand your product.

But every question lands you in a Slack thread waiting for engineers who are buried in their own work. Days pass. Specs get delayed. Customers wait for answers. All because the knowledge is locked in code.

01

The Developer Queue

You need to write a spec. But first you need context: How does auth work? What are the edge cases? Which services would this affect? Engineers are swamped. You wait three days for a 5-minute Slack call. Your sprint planning slips. The feature ships a week late because you couldn't get answers.

02

The Support Bottleneck

Customer reports a bug. You need to know: Is this actually a bug? Which component is broken? What changed recently? You file a ticket with engineering. They assign it. Someone investigates. Three days later: "We need more info." The customer is frustrated. The ticket bounces back and forth for a week.

03

The Documentation Drift

You write docs based on what you think the feature does. Engineering ships something slightly different. Customers follow your docs. Things break. Engineers fix the docs in code comments that you never see. Six months later, your documentation is 30% accurate and nobody trusts it.

What non-technical teams actually ask Probe

Real questions from product managers, support engineers, QA, and technical writers. Self-serve answers without engineering bottlenecks.

Question
How does user authentication work in our app? Walk me through the flow from login button to successful login.
Authentication flow: 1. User enters credentials (LoginForm.tsx) 2. Frontend sends POST to /api/auth/login (auth.controller.ts) 3. Controller validates credentials against User table 4. If valid, generates JWT token with 24h expiration 5. Token stored in httpOnly cookie 6. Subsequent requests include token in Authorization header 7. Middleware validates token on protected routes...
Question
Customer says their payment failed but they were charged. What could cause this? What data should I ask for?
This is a known edge case in payment processing. Root cause: If the Stripe webhook arrives after the HTTP request times out, the UI shows failure but the payment succeeds. Located in payment.service.ts lines 145-167. Ask customer for: Order ID, exact time of attempt, error message they saw. Check: Stripe dashboard for successful charge, database for order status. Workaround: Refresh order page — status should update once webhook processed.
Investigation
We changed the pricing tiers last week. What features are affected? Which parts of the UI need updating?
Pricing tier logic found in subscription.model.ts. Feature checks based on tier: API rate limits (api-gateway/rate-limiter.ts), export limits (export.service.ts lines 89-102), user seats (team.service.ts), storage quota (storage.service.ts). UI components referencing tiers: PricingPage.tsx, UpgradeModal.tsx, FeatureGate.tsx (used in 14 places), SettingsSidebar.tsx. All need audit if tier names or features changed.
Question
I'm writing a spec for bulk CSV import. What validation rules does the system already enforce? What edge cases should I document?
Current validation in csv-import.service.ts: Max file size 10MB, max 50k rows, required columns: email, name. Email validation: RFC 5322 compliant. Edge cases to document: 1. Duplicate emails (takes first, logs warning) 2. Invalid UTF-8 chars (import fails, line number reported) 3. Empty rows (skipped silently) 4. Column header case sensitivity (case-insensitive match) 5. Timeout after 5 min for large files. Recommend: Add progress indicator, chunk processing for files >10k rows.
Investigation
Customer reports export feature is broken. It worked yesterday. What changed in the last 24 hours?
Export feature changes in last 24h: PR #1247 merged yesterday 14:30 UTC. Changes: Added new column "created_date" to CSV export, updated database query to join created_at timestamp, added date formatting util. Potential issue: Query timeout on large datasets — no pagination on new query (export.service.ts line 234). Quick fix: Add LIMIT clause. Deployed to prod without testing on large datasets. Recommend: Rollback PR #1247 or hotfix with query optimization.
Digest
Generate release notes for v2.4.0 in customer-friendly language. Focus on what changed from their perspective.
Release Notes v2.4.0 (Customer-Facing): New: Bulk user import via CSV — upload up to 50,000 users at once from Settings → Team. Improved: Dashboard load time reduced by 40% for large accounts. Improved: Export now includes creation date for all records. Fixed: Payment error messages now show clearer instructions. Fixed: Session timeout extended from 1 hour to 24 hours. Technical: Upgraded to Node 20, improved database query performance, added request retries for external APIs.
Question
How does user authentication work in our app? Walk me through the flow from login button to successful login.
Authentication flow: 1. User enters credentials (LoginForm.tsx) 2. Frontend sends POST to /api/auth/login (auth.controller.ts) 3. Controller validates credentials against User table 4. If valid, generates JWT token with 24h expiration 5. Token stored in httpOnly cookie 6. Subsequent requests include token in Authorization header 7. Middleware validates token on protected routes...
Question
Customer says their payment failed but they were charged. What could cause this? What data should I ask for?
This is a known edge case in payment processing. Root cause: If the Stripe webhook arrives after the HTTP request times out, the UI shows failure but the payment succeeds. Located in payment.service.ts lines 145-167. Ask customer for: Order ID, exact time of attempt, error message they saw. Check: Stripe dashboard for successful charge, database for order status. Workaround: Refresh order page — status should update once webhook processed.
Investigation
We changed the pricing tiers last week. What features are affected? Which parts of the UI need updating?
Pricing tier logic found in subscription.model.ts. Feature checks based on tier: API rate limits (api-gateway/rate-limiter.ts), export limits (export.service.ts lines 89-102), user seats (team.service.ts), storage quota (storage.service.ts). UI components referencing tiers: PricingPage.tsx, UpgradeModal.tsx, FeatureGate.tsx (used in 14 places), SettingsSidebar.tsx. All need audit if tier names or features changed.
Question
I'm writing a spec for bulk CSV import. What validation rules does the system already enforce? What edge cases should I document?
Current validation in csv-import.service.ts: Max file size 10MB, max 50k rows, required columns: email, name. Email validation: RFC 5322 compliant. Edge cases to document: 1. Duplicate emails (takes first, logs warning) 2. Invalid UTF-8 chars (import fails, line number reported) 3. Empty rows (skipped silently) 4. Column header case sensitivity (case-insensitive match) 5. Timeout after 5 min for large files. Recommend: Add progress indicator, chunk processing for files >10k rows.
Investigation
Customer reports export feature is broken. It worked yesterday. What changed in the last 24 hours?
Export feature changes in last 24h: PR #1247 merged yesterday 14:30 UTC. Changes: Added new column "created_date" to CSV export, updated database query to join created_at timestamp, added date formatting util. Potential issue: Query timeout on large datasets — no pagination on new query (export.service.ts line 234). Quick fix: Add LIMIT clause. Deployed to prod without testing on large datasets. Recommend: Rollback PR #1247 or hotfix with query optimization.
Digest
Generate release notes for v2.4.0 in customer-friendly language. Focus on what changed from their perspective.
Release Notes v2.4.0 (Customer-Facing): New: Bulk user import via CSV — upload up to 50,000 users at once from Settings → Team. Improved: Dashboard load time reduced by 40% for large accounts. Improved: Export now includes creation date for all records. Fixed: Payment error messages now show clearer instructions. Fixed: Session timeout extended from 1 hour to 24 hours. Technical: Upgraded to Node 20, improved database query performance, added request retries for external APIs.
Automation
@Probe when a new support ticket mentions "payment" or "billing": gather context from payment service, recent payment-related code changes, and known issues. Add findings as internal comment.
Ticket #4521 created with keywords "payment failed". Context gathered: Recent changes — PR #1247 modified payment webhook handler 2 days ago. Known issues — 3 similar tickets this week (all related to webhook timing). Payment service status — healthy, no errors in logs. Stripe API status — operational. Added internal comment with: webhook change details, link to related tickets, recommended questions to ask customer, temporary workaround while fix is in progress.
Automation
@Probe daily at 9am: check for PRs merged yesterday that affect user-facing features. Generate a summary for the product team to review for doc updates.
Daily Product Changelog (yesterday): 3 user-facing changes. 1. Added "Export to PDF" button (PR #1289) — affects Settings → Reports page. Docs impact: Update reports documentation, add PDF export section. 2. Changed password requirements (PR #1291) — now requires 12 chars min, was 8. Docs impact: Update security docs, onboarding guide, FAQ. 3. Fixed timezone display bug (PR #1294) — user-reported issue. Docs impact: None, bug fix only. Posted summary to #product-updates.
Automation
Read the PRD for "Advanced Filters" feature, analyze the codebase for constraints and dependencies, and generate an implementation spec with edge cases and test scenarios.
Spec generated from PRD-2024-08. Implementation analysis: Filtering logic should extend existing search.service.ts. Database: Add indexes on filterable columns (user_created_at, account_tier, status). Constraints found: Current API returns max 1000 results — add pagination for filter results. Dependencies: Filter UI components can reuse existing FilterBar.tsx. Edge cases identified: 1. Multiple filters ANDed or ORed? 2. Empty filter results handling 3. Filter persistence across sessions 4. Performance with 100k+ records. Test scenarios: 12 cases generated including valid filters, invalid inputs, pagination, and performance tests.
Automation
When a customer reports a bug via email, extract the issue description, check if it's a known issue, search for similar past tickets, and draft a response with status and workaround.
Customer email processed (Sarah Chen, Acme Corp). Issue: "Can't export more than 100 records." Known issue: Yes — export limit introduced in v2.3.0 for performance. Located in export.service.ts line 67. Similar tickets: 8 in last month, all resolved with upgrade to Pro plan. Workaround: API export supports unlimited records, provide API docs link. Draft response generated: "Thanks for reaching out. This is a current limit on the Basic plan for performance reasons. Pro plan removes this limit, or you can use our API for unlimited exports. Here's how: [API docs link]. Let me know if you need help upgrading."
Automation
After each release, compare the release commit range with documentation files. Flag docs that are likely outdated based on code changes.
Release v2.4.0 doc audit complete. Code changes analyzed against docs/. Outdated docs detected: 1. docs/authentication.md — describes 1h session timeout, now 24h (changed in PR #1285). 2. docs/csv-import.md — missing new file size limit of 10MB (added in PR #1278). 3. docs/api-rate-limits.md — tier limits changed, docs still reference old tiers (changed in subscription.model.ts). Fresh docs: dashboard.md, webhooks.md, getting-started.md. Created DOC-401, DOC-402, DOC-403 with specific changes needed and code references.
Investigation
Three customers reported slow dashboard loading this week. Is this a new issue? What changed recently that could affect dashboard performance?
Dashboard performance investigation: Deployment timeline — v2.3.5 deployed Monday 9am. Code changes affecting dashboard: PR #1233 added real-time activity feed to dashboard (loads last 100 events per page load, no caching). Database query analysis: New query runs on every dashboard load, joins 3 tables, no indexes. Impact: Accounts with >10k events see 3-5s load time (was <1s). Customer reports match timing — all after Monday deploy. Recommendation: Add caching layer for activity feed, limit to last 50 events, add database index. Estimated fix: 2-3 hours dev time.
Automation
@Probe when a new support ticket mentions "payment" or "billing": gather context from payment service, recent payment-related code changes, and known issues. Add findings as internal comment.
Ticket #4521 created with keywords "payment failed". Context gathered: Recent changes — PR #1247 modified payment webhook handler 2 days ago. Known issues — 3 similar tickets this week (all related to webhook timing). Payment service status — healthy, no errors in logs. Stripe API status — operational. Added internal comment with: webhook change details, link to related tickets, recommended questions to ask customer, temporary workaround while fix is in progress.
Automation
@Probe daily at 9am: check for PRs merged yesterday that affect user-facing features. Generate a summary for the product team to review for doc updates.
Daily Product Changelog (yesterday): 3 user-facing changes. 1. Added "Export to PDF" button (PR #1289) — affects Settings → Reports page. Docs impact: Update reports documentation, add PDF export section. 2. Changed password requirements (PR #1291) — now requires 12 chars min, was 8. Docs impact: Update security docs, onboarding guide, FAQ. 3. Fixed timezone display bug (PR #1294) — user-reported issue. Docs impact: None, bug fix only. Posted summary to #product-updates.
Automation
Read the PRD for "Advanced Filters" feature, analyze the codebase for constraints and dependencies, and generate an implementation spec with edge cases and test scenarios.
Spec generated from PRD-2024-08. Implementation analysis: Filtering logic should extend existing search.service.ts. Database: Add indexes on filterable columns (user_created_at, account_tier, status). Constraints found: Current API returns max 1000 results — add pagination for filter results. Dependencies: Filter UI components can reuse existing FilterBar.tsx. Edge cases identified: 1. Multiple filters ANDed or ORed? 2. Empty filter results handling 3. Filter persistence across sessions 4. Performance with 100k+ records. Test scenarios: 12 cases generated including valid filters, invalid inputs, pagination, and performance tests.
Automation
When a customer reports a bug via email, extract the issue description, check if it's a known issue, search for similar past tickets, and draft a response with status and workaround.
Customer email processed (Sarah Chen, Acme Corp). Issue: "Can't export more than 100 records." Known issue: Yes — export limit introduced in v2.3.0 for performance. Located in export.service.ts line 67. Similar tickets: 8 in last month, all resolved with upgrade to Pro plan. Workaround: API export supports unlimited records, provide API docs link. Draft response generated: "Thanks for reaching out. This is a current limit on the Basic plan for performance reasons. Pro plan removes this limit, or you can use our API for unlimited exports. Here's how: [API docs link]. Let me know if you need help upgrading."
Automation
After each release, compare the release commit range with documentation files. Flag docs that are likely outdated based on code changes.
Release v2.4.0 doc audit complete. Code changes analyzed against docs/. Outdated docs detected: 1. docs/authentication.md — describes 1h session timeout, now 24h (changed in PR #1285). 2. docs/csv-import.md — missing new file size limit of 10MB (added in PR #1278). 3. docs/api-rate-limits.md — tier limits changed, docs still reference old tiers (changed in subscription.model.ts). Fresh docs: dashboard.md, webhooks.md, getting-started.md. Created DOC-401, DOC-402, DOC-403 with specific changes needed and code references.
Investigation
Three customers reported slow dashboard loading this week. Is this a new issue? What changed recently that could affect dashboard performance?
Dashboard performance investigation: Deployment timeline — v2.3.5 deployed Monday 9am. Code changes affecting dashboard: PR #1233 added real-time activity feed to dashboard (loads last 100 events per page load, no caching). Database query analysis: New query runs on every dashboard load, joins 3 tables, no indexes. Impact: Accounts with >10k events see 3-5s load time (was <1s). Customer reports match timing — all after Monday deploy. Recommendation: Add caching layer for activity feed, limit to last 50 events, add database index. Estimated fix: 2-3 hours dev time.

Three things that change everything

01

Plain English Codebase Understanding

Ask any question about your product and get accurate, contextual answers in plain English — no coding required.

  • "How does password reset work?"
  • "What validations exist for email addresses?"
  • "Which features are gated by subscription tier?"
  • "What happens when a webhook fails?"

Probe reads code semantically and translates it into plain English explanations. You get accurate answers grounded in the actual implementation, not guesses or outdated documentation. Perfect for product managers writing specs, support engineers investigating issues, and QA building test plans.

No technical knowledge neededAsk questions in plain English, get answers in plain English
Grounded in actual codeAnswers include file references and line numbers for verification
Cross-feature understandingTraces how features connect and depend on each other
02

Automated Spec Generation

Turn product ideas into engineering-ready specs with constraints, edge cases, and test scenarios automatically identified.

Great specs require understanding what already exists, what constraints apply, and what could go wrong. That knowledge lives in code, and extracting it usually requires bothering engineers for days.

Probe analyzes the codebase to generate complete specs. It identifies existing validation rules, finds similar features to reference, spots potential edge cases, and suggests test scenarios. Engineers can execute your specs immediately instead of sending them back with twenty clarifying questions.

Constraint discoveryAutomatically identify technical constraints and limitations
Edge case detectionSpot edge cases by analyzing similar features and error handling
Test scenario generationGenerate test cases covering positive, negative, and edge scenarios
03

Intelligent Support Triage

Investigate customer issues without waiting on engineering. Understand what changed, what's affected, and what to do about it.

Customer reports an issue. You need to know: Is it a bug? Is it a known issue? What changed recently? Which component is involved? This typically means filing an engineering ticket and waiting days for someone to investigate.

Probe investigates for you. It searches for recent code changes, checks for similar past issues, identifies affected components, and suggests workarounds. You can respond to customers faster, escalate to engineering with full context, or resolve issues yourself without engineering time.

Change correlationLink customer issues to recent code deployments and changes
Similar issue detectionFind past tickets and bug reports with related symptoms
Workaround identificationDiscover temporary fixes and alternative approaches from code

Workflow packs for non-technical teams

Pre-built automation workflows that replace manual processes. Customize per team. Version like code. Improve over time.

Product Management

PRD to Implementation Spec

Transform product requirements into engineering-ready specs with technical constraints, edge cases, dependencies, acceptance criteria, and test scenarios — all pulled automatically from your codebase.

  • Technical constraints identified
  • Edge cases and error scenarios
  • Dependency and integration points
  • Test case recommendations
Customer Support

Ticket Investigation

When a customer reports an issue, automatically gather context: recent code changes, affected components, similar past tickets, and known workarounds. Respond faster or escalate with full context.

  • Recent change timeline
  • Component analysis
  • Similar issue history
  • Workaround suggestions
Quality Assurance

Test Plan Generation

Generate comprehensive test plans from feature specs and code context. Identifies edge cases, integration points, regression risks, and acceptance criteria based on actual implementation.

  • Positive test scenarios
  • Negative test scenarios
  • Edge case coverage
  • Regression test recommendations
Documentation

Doc Freshness Monitoring

After each release, compare code changes with documentation files. Automatically flag docs that are likely outdated and suggest specific updates based on what changed in the code.

  • Outdated doc detection
  • Specific change recommendations
  • Code reference links
  • Priority ranking by user impact

Built for production use

Plain English Interface

No coding required. Ask questions in natural language and get answers you can actually understand. Built for product managers, support engineers, and QA teams.

Evidence Trails

Every answer includes references to source files and line numbers. Verify information yourself or share with engineering for validation. No black box magic.

Secure by Default

On-premises deployment option keeps code within your infrastructure. Choose your LLM provider, including self-hosted models. Full control over data security.

Integration Ready

Works with your existing tools: Jira, Confluence, Zendesk, Slack, GitHub. Workflow automation fits into how your team already works. No process overhaul required.

Open Source vs Enterprise

Start with the open-source core to evaluate the technology. Scale to enterprise when you need workflow automation and integrations.

Probe Open Source

Free forever

The core code intelligence engine. Perfect for evaluating the technology on a single project or for individual users exploring a codebase.

  • Single-project code understanding — Ask questions about one repository at a time
  • Plain English answers — Get explanations without reading code
  • No indexing required — Works instantly on any codebase, runs locally
  • MCP integration — Use with Claude Code, Cursor, or any MCP-compatible tool
  • Any LLM provider — Claude, GPT, open-source models — your choice
  • Privacy-first — Everything runs locally, no data sent to external servers

Probe Enterprise

Contact for pricing

Everything in Open Source, plus multi-project architecture, workflow automation, and integrations with your existing tools.

  • Multi-repository architecture — Query across your entire system of services, not just one project
  • Cross-service understanding — Understand how features connect across the entire product
  • Workflow automation — Automated spec generation, ticket triage, test planning, and doc monitoring
  • Jira integration — Pull ticket context, specs, and acceptance criteria into answers
  • Zendesk integration — Connect support tickets to code for faster issue resolution
  • Confluence integration — Keep documentation synchronized with code changes
  • Change correlation — Link customer issues to recent deployments and code changes
  • Similar issue detection — Find related past tickets and known workarounds automatically
  • Slack/Teams integration — Ask questions from where you already work
  • On-premises deployment — Runs entirely in your infrastructure for maximum security

How to evaluate Probe

Two-phase approach: validate the core technology with open source, then pilot enterprise features with your actual workflows.

Phase 1

Individual Evaluation

~30 minutes

Try Probe yourself on your codebase. Ask questions about features, investigate how things work, and see how accurate the answers are. No engineering help required.

~10 min

Ask Questions About Your Product

Install Probe and point it at your repository. Ask questions about features, authentication, workflows, or anything you've always wondered about but never had time to dig into code for.

You get: Plain English explanations of how your product works, grounded in actual code with references you can verify.
Quick start guide →
~15 min

Generate a Test Spec

Give Probe a feature description and ask it to generate test scenarios. Compare what it produces to what you would write manually. See if it catches edge cases you missed.

You get: A comprehensive test plan with positive, negative, and edge case scenarios based on actual implementation details from the code.
AI chat guide →
~20 min

Investigate a Past Issue

Pick a customer issue you investigated before. Ask Probe to analyze it: What changed? Which components were involved? What similar issues existed? Compare to what you learned the hard way.

You get: Full investigation context that would have taken days of back-and-forth with engineering, delivered in seconds.
AI chat guide →
Phase 2

Team Pilot

2-4 weeks

Once you've validated core capabilities, run a team pilot to test workflow automation and integrations with your existing tools.

1
Connect your systems

Integrate Probe with your existing tools: GitHub, Jira, Confluence, Zendesk, Slack. Map your repositories and configure access permissions.

2
Deploy workflow automations

Set up automated workflows for your most time-consuming processes: ticket triage, spec generation, test planning, or doc monitoring.

3
Customize per team

Configure workflows to match your team's terminology, quality standards, and processes. Templates are starting points, not rigid requirements.

4
Measure the difference

Track time waiting for developer answers, spec iteration cycles, support ticket resolution time, and documentation freshness. Compare to pre-pilot baselines.

Success criteria: Measurable reduction in time waiting on engineering. Fewer spec rewrites due to missing context. Faster support ticket resolution. Documentation that stays fresh automatically.

Want to discuss how a pilot would work for your team?

Schedule a Conversation

What non-technical teams ask us

Do I need to know how to code to use this?

No. Probe is designed for non-technical users. You ask questions in plain English and get answers in plain English. The system handles reading and understanding code — you just interact with the insights it provides.

How is this different from just asking ChatGPT or reading documentation?

ChatGPT has no access to your actual codebase and often hallucinates answers. Documentation goes stale the moment code changes. Probe reads your actual code in real-time, giving you accurate answers with references to specific files and line numbers. You can verify every answer.

Will this replace the need to talk to engineers?

No, but it will make those conversations much more productive. Instead of "how does auth work?", you'll ask "I see auth uses JWT tokens with 24-hour expiration — what happens when a token expires during an active session?" Engineers spend less time explaining basics and more time on strategic design discussions.

What if the answers are wrong?

Every answer includes references to source code locations. You can verify information yourself or share with engineers for validation. The system is transparent about what it knows and where that knowledge comes from. When something is unclear or complex, it says so instead of guessing.

Ready to stop waiting on engineers for answers?

Let's talk about how Probe can help your team understand the codebase, write better specs, and resolve customer issues faster. We'll show you how it works on your actual product.