# Bilanc

> Engineering intelligence for the AI SDLC. Bilanc reads every line of code your team ships, from prompt to production — measuring engineering output, the return on every AI dollar, and where delivery is stuck.

Bilanc is engineering intelligence for the AI SDLC. We read every line of code your team ships, from prompt to production, showing leaders who's actually shipping, where delivery is stuck, and undeniable proof of what AI contributes. Built for engineering leaders navigating the agentic era.

## Pages

- [Home](/): Overview of Bilanc — engineering intelligence for the AI SDLC.
- [AI measurement & ROI](/product/ai-roi): Adoption, acceptance, attribution and cost per AI tool.
- [Engineering intelligence](/product/engineering-intelligence): Complexity-scored PRs, cycle time, rework and dashboards.
- [Enterprise & self-hosting](/product/enterprise): RBAC, audit logs, SSO/SAML and VPC deployment.
- [PostHook](/product/posthook): Open-source CLI for per-line AI attribution, measured on-device.
- [DX Surveys](/product/surveys): DORA-grade developer surveys with AI-written results reports.
- [MCP server](/product/mcp): Your engineering data in any MCP client — Claude, Cursor and beyond.
- [Pricing](/pricing): Simple per-engineer pricing and Enterprise plan details.
- [Blog](/blog): Insights on engineering productivity, AI coding tool ROI, and DORA metrics.
- [News UK case study](/blog/news-uk-measuring-engineering-productivity-in-the-ai-era): How News UK's SVP Technology uses Bilanc to reconcile engineering output with cost and run AI rollouts as controlled experiments (video + transcript).
- [Beyond Lines of Code](/blog/beyond-lines-of-code-measuring-engineering-productivity-with-ai): How we measure engineering productivity with AI.
- [Measuring AI's Impact](/blog/measuring-ai-impact-on-engineering-productivity): Framework for measuring AI tool impact on engineering teams.
- [LLMs as Feature Extractors](/blog/llms-as-feature-extractors-for-classical-ml): Using LLMs to feed classical ML pipelines.

## Product

Bilanc connects read-only to your version control (GitHub, GitLab, Bitbucket, Azure DevOps), AI coding tools, tickets, CI and Slack, then analyses every pull request with AI. It measures engineering output, traces AI-generated code from prompt to production, and prices the whole thing against salaries, seats and tokens.

### Key Features

- **AI Impact Measurement**: Quantify adoption, acceptance and merged AI output for every AI coding tool across your engineering organisation
- **Pull Request Analytics**: Every PR is scored for complexity, effort, and quality — giving managers a ground-truth view of engineering output
- **Bottleneck Detection**: Cycle time decomposed into coding, pickup and review; surface slow review cycles and blocked engineers before they compound
- **Automated Reporting**: AI-generated reports and scheduled briefs for engineering leaders, delivered automatically
- **Cost Analytics**: Cost per PR and per complexity point, built from salaries, seats and token spend

### PostHook

- Open-source CLI (github.com/Bilanc/posthook) that runs on engineers' laptops and attributes AI-written code line by line
- Native hooks for Claude Code, Cursor and Codex CLI, plus a git shadow for every other tool; attribution travels with the repo via git notes
- The AI Code Attribution Funnel: Generated → Committed → Merged → Retained — measured on-device, not estimated from vendor APIs
- Local-first: solo mode keeps everything in local SQLite; team mode is one shared, revocable install link with identity from git config

### DX Surveys

- Built on DORA's actual research instruments: State of Software Development (DORA 2025), Impact of Generative AI, and AI Capabilities templates
- 7 question types including reverse-coded matrix rows and non-scorable "I don't know" handling
- Favorability scored 0–100 in heatmaps across Teams, Levels, Locations, Squads, Departments and Managers
- An AI-generated results report is produced when the survey closes

### MCP Server

- Live at https://api.bilanc.co/mcp — point any MCP client at it (Claude Code, claude.ai, Claude Desktop, Cursor)
- 6 tools: read-only SQL over 21 analytics tables plus code search/read over a tenant code sandbox
- Read-only, OAuth 2.1, per-user Postgres roles with row-level security — results scoped to what the caller may see

### Pricing

$25 per engineer per month. Simple, per-seat pricing. No annual commitments required.

## Company

- **Founded**: 2023
- **Website**: https://bilanc.co
- **Contact**: hello@bilanc.co
- **Backed by**: Y Combinator

## Use Cases

- Prove what AI contributes to the board with per-line attribution from generated to retained
- Roll out coding agents as a measured experiment with balanced waves and a control group
- Unblock delivery and review bottlenecks with decomposed cycle time
- Run DORA-grade developer experience surveys with AI-written reports
- Query your engineering data from your own AI agent over MCP
- Put a price on every PR from salaries, seats and token spend

## Target Audience

Engineering leaders at software companies (10–500+ engineers) who are adopting AI coding tools and need to measure their impact objectively.

## Integrations

- GitHub, GitLab, Bitbucket, Azure DevOps
- Claude Code, Cursor, GitHub Copilot, Codex, Augment Code, PostHook
- Jira, Linear, GitHub Issues, Azure Boards
- Slack, CircleCI, Incident.io

## Key Differentiators

Unlike DORA metrics dashboards or activity trackers, Bilanc uses LLMs to read the actual code in every pull request — understanding effort, complexity, and quality at the semantic level. With PostHook, attribution starts on the laptop where the AI code is written, so productivity scores reflect real shipped work, not vendor adoption charts.

## Links

- Homepage: https://bilanc.co
- Pricing: https://bilanc.co/pricing
- Blog: https://bilanc.co/blog
- PostHook (open source): https://github.com/Bilanc/posthook
- LinkedIn: https://www.linkedin.com/company/bilanc
- Twitter/X: https://twitter.com/biaborhein
- Y Combinator: https://www.ycombinator.com/companies/bilanc