# LetPeopleWork - Lighthouse in full > The complete text behind https://letpeople.work for AI assistants and search agents: the summary from llms.txt, the Lighthouse product page with its FAQ, and the guides. Generated from the site's own content on 2026-10-17. The short version is https://letpeople.work/llms.txt. # Part 1: Summary (llms.txt) > LetPeopleWork GmbH builds tools for software delivery teams: Lighthouse, a self-hosted, source-available flow metrics and Monte Carlo forecasting tool, and Sizing Poker, a free browser tool for sizing a backlog without story points. Founded by Peter Zylka-Greger and Benjamin Huser-Berta in Zürich, Switzerland. ## About LetPeopleWork **What does LetPeopleWork do?** LetPeopleWork helps delivery teams answer two questions: when will it be done, and why is delivery slow? Lighthouse answers the first with Monte Carlo forecasts from the team's own throughput, and the second with flow metrics. Sizing Poker sizes a backlog without estimates. **Who builds the tools?** Two practitioners who use them with teams themselves. Benjamin Huser-Berta is the main developer of Lighthouse; Peter Zylka-Greger works with teams and leaders on flow and forecasting. The blog at https://blog.letpeople.work explains how the forecasts work, with worked examples. **Who are the founders?** - Peter Zylka-Greger: Expert in Flow, Kanban, and organizational transformation. Trains teams and leaders in data-driven delivery. - Benjamin Huser-Berta: Software Engineer and Scrum Master. Core maintainer of Lighthouse, focused on reducing waste and building value-focused work environments. ## Core Pages - [Homepage](https://letpeople.work/): Overview of the LetPeopleWork tools, Lighthouse and Sizing Poker, and the company mission - [Lighthouse Product Page](https://letpeople.work/lighthouse/): Full details on Lighthouse features, pricing, downloads, and AI integration - [Sizing Poker](https://letpeople.work/sizing-poker/): Free browser tool that sizes a backlog by asking one question per item instead of estimating it - [AI at LetPeopleWork](https://letpeople.work/ai/): obAIa, open-source Claude Code plugins, and the AI workshop journey - [Delivery Predictability Assessment](https://letpeople.work/assessment/): Free five-minute self-assessment of how predictable your delivery is - [Lighthouse vs ActionableAgile vs Nave](https://letpeople.work/compare/): Side-by-side comparison of hosting, data location, pricing model and integrations, for teams looking for a self-hosted alternative - [Forecasting Jira without story points](https://letpeople.work/jira-forecasting-without-story-points/): How to forecast Jira releases from throughput with Monte Carlo instead of estimating, step by step - [Azure DevOps delivery forecasting](https://letpeople.work/azure-devops-forecasting/): How Lighthouse connects to Azure DevOps and forecasts from your work items, without story points - [llms-full.txt](https://letpeople.work/llms-full.txt): This summary plus the full product page, FAQ and guides in one file ## Sizing Poker - Backlog Sizing Without Estimates **What is Sizing Poker?** A free tool at https://letpeople.work/sizing-poker/ for sizing a backlog without story points or estimates. You set the number of days an item should normally be done in (this is the team's Service Level Expectation), then see one work item at a time and answer a single question: could we finish this within that time? Three answers: Yes (pull it when there is room), Maybe (something has to be arranged first, name it), No (too big, split it). A round of ten items takes minutes. It runs entirely in the browser; nothing is uploaded and no signup is needed. A walkthrough with an example backlog explains each step and what to discuss after each answer. **Who is it for?** Product Owners, Scrum Masters and teams doing backlog refinement who want to replace estimation with a fast fit-or-not decision, and teams using Lighthouse who want to right-size their work against their Service Level Expectation. ## Lighthouse - Flow Metrics & Forecasting Tool **What is Lighthouse?** Lighthouse is a source-available flow metrics and probabilistic forecasting tool. It connects to your ALM system (Jira, Azure DevOps, Linear, or ServiceNow), visualizes how work flows through your teams, and generates Monte Carlo-based delivery forecasts grounded in your actual throughput history. It runs entirely on your own infrastructure — no cloud dependency, no data leaving your environment. **What questions does Lighthouse answer?** - When will these work items be done? Dates with probabilities (for example 50% by 14 March, 85% by 28 March), from a Monte Carlo simulation over the team's past throughput. No story points needed. - How many items will be done by a given date? - What is stuck right now? Work in progress, work item age, blocked and stale items, and items at risk of missing the team's service level expectation (SLE). - Which features make the release? A delivery timeline shows each feature's forecast start and finish, with dependencies read from the tracker. - How is the portfolio doing? Feature forecasts that combine every team working on a feature into one joint probability. - How much should we refine? A refinement forecast plus sizing votes per item. - Did the change work? Reports on every team and portfolio. The first report type, Then & Now, freezes a window from before a change (it may have ended long ago) and compares it with a window that moves with today: cycle time, throughput, WIP and work item age, with a summary of what changed. Community includes one report of each type per team and portfolio; paid editions have no limit. More report types are coming. https://docs.lighthouse.letpeople.work/reports/reports.html **Key features of Lighthouse:** 1. Monte Carlo forecasts for teams ("how many" and "when") and for features, deliveries and portfolios 2. Forecast Reality Check: backtests past forecasts against what the team delivered, per percentile and sampling window 3. Forecasted start dates next to completion dates, and a delivery timeline per delivery 4. Dependencies read from Jira (is blocked by), Azure DevOps (predecessor) and Linear, and taken into account in forecasts on paid editions 5. Flow metrics: cycle time, throughput, work in progress, work item age with pace bands, blocked and stale items, risk of missing the SLE, process behaviour charts, percentiles over time, cumulative time in state 6. Reporting: a Reports tab on every team and portfolio, starting with Then & Now, which compares flow metrics from before a change with today's 7. Connects to Jira Cloud, Jira Server and Data Center, Azure DevOps, Linear, ServiceNow and CSV 8. AI integration via MCP (Model Context Protocol), a command line client and agent skills 9. Self-hosted: desktop app (Windows, macOS, Linux), Docker, or Kubernetes with the official Helm chart 10. Source available on GitHub; free Community edition for up to 3 teams and 1 portfolio **What is AI integration in Lighthouse?** Lighthouse exposes your flow data to AI assistants through three integration methods: 1. **CLI** (`@letpeoplework/lighthouse-cli`): Run Lighthouse queries from the terminal, shell scripts, or CI/CD pipelines. Outputs JSON for easy downstream use. 2. **MCP stdio** (`@letpeoplework/lighthouse-mcp-stdio`): Run a local MCP server over stdio. Works with Claude Desktop, VS Code, JetBrains, and any MCP-compatible client. 3. **MCP HTTP** (`@letpeoplework/lighthouse-mcp-http`): Run a persistent MCP server over HTTP. Ideal for shared team setups or self-hosted AI environments. With AI integration, you can ask natural-language questions like: - "Which teams should I focus on this week?" — get a prioritized view of where cycle time is spiking, WIP is creeping, or predictability is dropping - "Will we finish the Q3 release by August?" — run a Monte Carlo forecast in seconds - "What's one thing this team could do to improve their flow?" — get coaching recommendations based on actual metrics AI integration is available as a downloadable skill for compatible clients and works with any tool-calling AI (Claude, GitHub Copilot, and others). **Downloadable AI integration packages:** - Skill bundle (.zip): https://github.com/LetPeopleWork/lighthouse-clients/releases/latest/download/lighthouse-skill.zip - MCP bundle (.mcpb): https://github.com/LetPeopleWork/lighthouse-clients/releases/latest/download/lighthouse-mcp-stdio.mcpb - Full setup guide: https://docs.lighthouse.letpeople.work/aiintegration.html **What are Lighthouse's licensing options?** Three editions. The flow metrics and forecasts are in every edition; paid editions lift the limits and add features such as sign-in, role-based access, rule-based deliveries, data sync mappings and dependency-aware forecasts. In the Lighthouse app and documentation these paid features are called Premium, and a Premium license switches them on: Self-Service and Enterprise are the two ways to buy one, with the same features and different support. All editions are self-hosted with the source public, and your data stays on your own systems whether or not you pay. - Community Edition: Free forever. All core flow metrics and Monte Carlo forecasts, up to 3 teams and 1 portfolio with 1 delivery. - Self-Service: CHF 2,000/year. A Premium license: unlimited teams, portfolios, and deliveries, all paid features, community Slack support. One flat rate, no per-seat pricing. - Enterprise: CHF 10,000/year. The same Premium license as Self-Service, plus prioritised support, 24h acknowledgement on async requests, onboarding calls, and named contacts. **How long does it take to set up Lighthouse?** Plan for about half an hour from download to a first forecast. In a recorded live session, Scrum Master Gábor Bittera did it in under ten minutes: he started a ten minute timer, downloaded Lighthouse, connected his Jira instance with an API token, added his team, and forecast the team's next ten work items, finishing with 1:25 left on the clock. The demo is on https://letpeople.work/lighthouse#first-forecast. Guided onboarding is also available free of charge for every edition. **How do I install Lighthouse?** Available for: - Windows: Installer (.exe, .msi) — code signed - macOS: Disk image (.dmg) and App Bundle (.zip) — signed and notarized - Linux: AppImage and server binaries (.zip) - Docker: `ghcr.io/letpeoplework/lighthouse:latest` - Kubernetes: official Helm chart, see https://docs.lighthouse.letpeople.work/Installation/kubernetes.html Download from GitHub releases: https://github.com/LetPeopleWork/Lighthouse/releases **What systems does Lighthouse integrate with?** Jira Cloud, Jira Server and Jira Data Center (with a personal access token), Azure DevOps, Linear, ServiceNow (team level) and CSV files. OAuth for Jira Cloud and Azure DevOps on paid editions. AI assistants via MCP (any tool-calling AI, including Claude and GitHub Copilot). **Does Lighthouse work with a Jira that is only reachable inside our network?** Yes. Lighthouse runs where you install it (a laptop, a VM, Docker or Kubernetes inside your network) and connects to Jira Server or Data Center directly with a personal access token. Work items stay on your systems. **Won't Jira or Azure DevOps build this in?** For one team in one tool, their built-in charts may be enough. Lighthouse does flow and forecasting only, works the same way across Jira, Azure DevOps, Linear and ServiceNow, combines several teams into one forecast, and checks its own forecasts against what really happened. The source is public, so there is no lock-in. **How accurate are Lighthouse forecasts?** A forecast is a range with probabilities, not a promise. Lighthouse assumes the next weeks look roughly like the last ones, and Forecast Reality Check lets you test that on your own history. In one backtest Peter Zylka-Greger ran on his team, the 85% forecast said 12 items for two weeks and the team finished 16. **How does Lighthouse compare to ActionableAgile and Nave?** All three provide flow metrics (cycle time, throughput, WIP, work item age) and data-driven forecasting. The key differences: - Lighthouse is source available and self-hosted (desktop app or Docker); ActionableAgile (by 55 Degrees) and Nave are closed-source commercial SaaS. - Lighthouse has a free-forever Community edition and flat per-organisation pricing (Self-Service CHF 2,000/year; Enterprise CHF 10,000/year). ActionableAgile is priced per user; Nave is priced per connected board. - Lighthouse connects to Jira, Azure DevOps, Linear, and ServiceNow; ActionableAgile to Jira and Azure DevOps; Nave to Jira, Trello, and Azure DevOps. - Lighthouse has native AI integration via MCP and CLI; that is not a core feature of the others. - With Lighthouse your delivery data stays on your own infrastructure. Full, honest comparison table: https://letpeople.work/compare/ ## Questions people ask when looking for a tool like this **Is there a free Monte Carlo forecasting tool for Jira?** Lighthouse's Community edition is free for up to 3 teams and 1 portfolio. It connects to Jira Cloud and Jira Data Center, runs on your own machine or server, and forecasts "when" and "how many" from the team's throughput. **What is a self-hosted alternative to ActionableAgile or Nave?** Lighthouse. It covers the same flow metrics (cycle time, throughput, WIP, work item age) and Monte Carlo forecasting, runs on your own infrastructure, and has flat per-organisation pricing. Comparison: https://letpeople.work/compare/ **How can we forecast delivery dates in Azure DevOps without story points?** Lighthouse reads finished work items from Azure DevOps and forecasts from throughput, so no estimates are needed. It also reads Predecessor links as dependencies. **How do we forecast a feature that several teams work on?** Lighthouse combines the simulations of every contributing team into one joint probability, which is more realistic than taking the slowest team's date. ## AI at LetPeopleWork **How does LetPeopleWork use AI?** We run our own company with AI and call the setup obAIa: an AI-augmented "big room" where decisions get pressure-tested by three perspectives (commercial, marketing, go-to-market) and are kept with their reasoning. It holds our work, pipeline, and dates in one place and surfaces what needs attention. We teach AI the way we use it every day, not from a slide deck. **What open-source AI plugins does LetPeopleWork provide?** Two Claude-Code-native plugins, free and open source: - LetPeopleWorkShop (https://github.com/LetPeopleWork/LetPeopleWorkShop): Design, prepare, and learn from workshops, all in markdown. Agents: designer, executor, feedback. Includes a facilitation-practices library (Liberating Structures, Training from the BACK of the Room). Install: `/plugin marketplace add LetPeopleWork/LetPeopleWorkShop` then `/plugin install let-people-workshop@letpeoplework`. - LetPeopleGrow (https://github.com/LetPeopleWork/LetPeopleGrow): Prepare for, capture, and learn from 1:1s. Coaching support, never a performance dossier. Agents: prep, capture, growth, patterns. Includes a conversation-practices library (GROW, SBI, NVC, career conversations). **What AI workshops does LetPeopleWork offer?** A four-step AI journey for businesses. Each workshop is a private 1-day session for one organisation, on-site or remote, bookable on its own or as the full journey. Pilot price: CHF 2,500 per team (covers the entire team, not a seat), subject to change once the pilot cohorts fill up. 1. The bAsIcs (literacy): what agents, commands, and skills are, and why the chat box is not the whole of AI 2. BIYA: Build It Yourself Agent: build your first working agent hands-on 3. From Prototype to Product: build a real, working product yourself with tools like Lovable and Claude Code 4. Filling the Gap: The AI-Powered Business: run parts of your business with AI, the obAIa way A separate "AI for Engineers" track (software engineering with agents) is coming soon. ## Assessment and Onboarding **Is there help with getting started?** - Free Delivery Predictability Assessment (https://letpeople.work/assessment/): a five-minute, six-question self-assessment that scores how you measure flow and how you forecast, with a concrete next step. No email required. - Lighthouse Setup & Onboarding: free of charge, for every edition. We help you get Lighthouse running, connect Jira, Azure DevOps, Linear, or ServiceNow, and walk your team through the tool. Contact contact@letpeople.work with subject "Lighthouse onboarding". - Premium trial: a 30-day license for the Self-Service edition, sent by email. No account and no credit card. Request it on https://letpeople.work/lighthouse/ **How can I contact LetPeopleWork?** - General inquiries: contact@letpeople.work - Tool support: support@letpeople.work - Premium licensing: licensing@letpeople.work - Community Slack: https://join.slack.com/t/let-people-work/shared_invite/zt-38df4z4sy-iqJEo6S8kmIgIfsgsV0J1A ## Company Information **Where is LetPeopleWork located?** LetPeopleWork GmbH, Mühlackerstrasse 108, 8046 Zürich, Switzerland. **What are LetPeopleWork's channels?** - Website: https://letpeople.work - LinkedIn: https://www.linkedin.com/company/let-people-work/ - GitHub: https://github.com/LetPeopleWork - YouTube (Lighthouse Live and sessions): https://www.youtube.com/@LetPeopleWork/videos - Meetup (community events): https://www.meetup.com/lighthouselive/ - Community Slack: https://join.slack.com/t/let-people-work/shared_invite/zt-38df4z4sy-iqJEo6S8kmIgIfsgsV0J1A - Lighthouse Documentation: https://docs.lighthouse.letpeople.work **What is LetPeopleWork's mission?** To let people work: fewer guessed dates, fewer status meetings, more finished work. The data to improve delivery is already in your tracker. Lighthouse shows it, and Sizing Poker helps you act on it in refinement. ## Technical Details **What technology stack is Lighthouse built on?** .NET Core / C# backend, React + TypeScript frontend. SQLite for local data storage. Cross-platform (Windows, macOS, Linux, Docker). **Is Lighthouse open source?** No - Lighthouse is source available. The source is public on GitHub and you can read, audit, run and modify it, but it is not open source under the Open Source Initiative definition: competing products and hosted resale are restricted. Releases up to v26.9.9.9 remain MIT licensed. https://github.com/LetPeopleWork/Lighthouse **Where does Lighthouse store data?** In its own database on your infrastructure: SQLite by default, and PostgreSQL is supported (the Helm chart bundles it). Lighthouse sends nothing to LetPeopleWork unless someone turns on usage data, which is opt-in and never contains work items, names or queries. **What is MCP?** MCP (Model Context Protocol) is an open standard for connecting AI assistants to external data sources. Lighthouse implements MCP so that AI tools like Claude and GitHub Copilot can query your flow data directly — no copy-paste, no exports. ## Keywords & Topics Flow metrics, kanban, delivery forecasting, Monte Carlo simulation, throughput, cycle time, work item age, work in progress, predictability, Jira integration, Azure DevOps integration, Linear integration, ServiceNow integration, AI integration, MCP, Model Context Protocol, agentic workflows, probabilistic forecasting, team performance, organizational improvement, source available tools, self-hosted tools, self-hosted solutions, Lighthouse tool, Sizing Poker, right sizing, no estimates, LetPeopleWork, obAIa, AI workshops, AI for business, Claude Code plugins, LetPeopleWorkShop, LetPeopleGrow, AI-powered business, delivery predictability assessment. ## Documentation & Resources - [Lighthouse Docs](https://docs.lighthouse.letpeople.work): Complete guide to setup, features, and AI integration - [GitHub Repository](https://github.com/LetPeopleWork/Lighthouse): Source code, issues, release notes - [GitHub Releases](https://github.com/LetPeopleWork/Lighthouse/releases): Latest downloads for all platforms - [AI Integration Guide](https://docs.lighthouse.letpeople.work/aiintegration.html): How to connect Lighthouse to Claude, Copilot, and other AI tools - [LetPeopleWorkShop Plugin](https://github.com/LetPeopleWork/LetPeopleWorkShop): Open-source Claude Code plugin for designing and running workshops - [LetPeopleGrow Plugin](https://github.com/LetPeopleWork/LetPeopleGrow): Open-source Claude Code plugin for 1:1s and people development # Part 2: The Lighthouse product page URL: https://letpeople.work/lighthouse/ ## Lighthouse: flow metrics and Monte Carlo forecasts for software delivery Know how likely your date is before you promise it. Lighthouse is a self-hosted, source-available tool that turns your team's delivery history into probabilistic forecasts and flow metrics, for one team or a whole portfolio. It runs on your laptop or inside your network and connects to Jira Cloud and Data Center, Azure DevOps, Linear and ServiceNow. It is a self-hosted alternative to ActionableAgile and Nave, and the Community edition is free for up to 3 teams. ### Questions Lighthouse answers #### When will it be done? Ask about 10 items and you get dates with their odds, for example 50% by 14 March and 85% by 28 March. Ask about a date and you get how many items will probably make it. Lighthouse runs a Monte Carlo simulation over the team's past throughput, so nobody has to estimate. It assumes the next weeks look roughly like the last ones; holidays can be blocked out. #### What's stuck right now? Work in progress, work item age, blocked and stale items, each red, amber or green. Items at risk of missing the team's service level expectation (SLE) are listed while there is still time to act. #### Which features make the release? The delivery timeline shows one bar per feature, from the forecast start to the forecast finish, with dependencies from your tracker drawn as lines. Features that will not start before the target date are marked. #### How is the portfolio doing? The same flow metrics one level up, and a forecast for every feature. When several teams work on one feature, Lighthouse combines their simulations into one probability, which is lower than any single team's odds because all of them have to make it. #### How much should we refine? Lighthouse forecasts how many items the team will pull before the refinement after next, and the team sizes each item with one question: could we finish it within our SLE? #### Did the change work? Every team and portfolio has a Reports tab. The first report type, Then & Now, freezes a window from before a change, even one that ended long ago, and compares it with a window that moves with today: cycle time, throughput, WIP and work item age, with a summary of what changed. Reports can be edited and deleted, and they show on the Overview. The Community edition includes one report of each type per team and portfolio; Self-Service and Enterprise have no limit. More report types are coming. [How reports work](https://docs.lighthouse.letpeople.work/reports/reports.html). ### Forecasts you can check Forecast Reality Check forecasts weeks that already happened, as if they had not, and compares the result with what the team really finished, for every percentile and sampling window. When Peter Zylka-Greger backtested his own team, the 85% forecast said 12 items for two weeks; the team finished 16. ### Flow metrics - Cycle time scatterplot and percentiles, named cycle times - Throughput run chart and total throughput - Work in progress and work item age, with pace bands - Blocked items over time and stale items - Risk of missing the service level expectation (SLE) - Process behaviour charts, percentiles over time, cumulative time in state - Feature size percentiles and a predictability score ### Where it runs and what it connects to - Desktop app for Windows, macOS and Linux, a Docker container, or Kubernetes with the official Helm chart. - Jira Cloud, and Jira Server or Data Center with a personal access token. Azure DevOps, Linear, ServiceNow and CSV. OAuth for Jira Cloud and Azure DevOps on paid editions. - Your work items stay in Lighthouse's own database on your systems. Usage data to LetPeopleWork is off unless someone turns it on, and it never contains work items, names or queries. - The source is public on [GitHub](https://github.com/LetPeopleWork/Lighthouse) under the Lighthouse Source Available License. - AI assistants such as Claude or Copilot can read the metrics and run forecasts through Lighthouse's MCP server. There is also a command line client, lh. ### Editions and pricing - Community: free, up to 3 teams and 1 portfolio with 1 delivery, all core flow metrics and Monte Carlo forecasts. - Self-Service: CHF 2,000 a year, unlimited teams and portfolios, sign-in and role-based access, rule-based deliveries, data sync mappings. - Enterprise: CHF 10,000 a year, everything in Self-Service plus priority support and onboarding calls. - In the app and the documentation the paid features are called Premium. Self-Service and Enterprise both include all of them; Enterprise adds the support. - Try Premium free for 30 days: the license comes by email, with no account and no credit card. Compared with ActionableAgile and Nave: [https://letpeople.work/compare/](https://letpeople.work/compare/) ### Questions people ask us #### I need lots of data first. About ten finished work items are enough for a first useful forecast. Most teams have that after two weeks. #### Our tracker is messy. I don't have clean data. Messy is normal. Lighthouse pulls from Jira, Azure DevOps, Linear, ServiceNow, or CSV, and the math holds up against noisy data. The signal you need is when items started and when they finished, and that survives even in trackers that have been through three reorgs. #### IT will block it. Our Jira isn't even reachable from the internet. Then run Lighthouse inside your network too: on your laptop, a VM, Docker or Kubernetes. It connects to Jira Server and Data Center with a personal access token. Your work items stay on your systems, and nothing comes to us unless someone turns on usage data (which holds no content). For a first try, your own laptop is enough. No ticket to IT. #### Won't Jira or Azure DevOps just build this in? Maybe. For one team in one tool, its charts may be all you need. Lighthouse only does flow and forecasting, works the same way across Jira, Azure DevOps, Linear and ServiceNow, and checks its own forecasts against what really happened. A feature worked by three teams gets one forecast for all three. And the source is public, so you're not locked in, to us or to your tracker. #### My team's not mature enough for this. Probabilistic forecasting works on whatever flow you have today. Starting early helps, because the forecast shows you what to improve first. #### I need management buy-in first. Not to try it. Maybe to roll it out. Try it on your own, produce a forecast you couldn't have produced before, then decide whether to share it. The cheapest way to show the value is to run it yourself. #### I'm just an IC, Scrum Master, or coach. This isn't my call. Most of Lighthouse's early users are exactly that: ICs, Scrum Masters, and coaches running it for their own team. It installs on your laptop, and nobody else needs to know until you have something worth showing. #### It looks complicated. Plan for half an hour from download to a first forecast. Gábor Bittera did it live in under ten minutes, which you can watch further up this page. Pre-loaded sample data lets you look around before you connect anything. #### Do you help with setup and onboarding? Yes, and it's free. We help you get Lighthouse running, connect Jira, Azure DevOps, Linear, or ServiceNow, and walk your team through the tool, at no cost and for every edition. Write to contact@letpeople.work with the subject "Lighthouse onboarding". Enterprise licenses also include structured onboarding calls. #### I tried something like this before and it didn't stick. Was it ActionableAgile, Nave, or Jira's own forecasting? Most of these tools die in real organisations for boring reasons: IT blocks the SaaS, the budget gets cut, the champion leaves. Lighthouse runs on your own systems and is free to start, so none of that hits the same way. #### Is this even for me? Maybe not if you are one team with nothing bigger than the next sprint to plan; the charts in your tracker may be enough. Or if your project is nearly done, because a forecast can't help after the fact. Fixed deadlines are different: when the date can't move, the useful question is how much will fit by then, and Lighthouse answers that. Downloads: [https://github.com/LetPeopleWork/Lighthouse/releases](https://github.com/LetPeopleWork/Lighthouse/releases). Documentation: [https://docs.lighthouse.letpeople.work](https://docs.lighthouse.letpeople.work). Blog: [blog.letpeople.work](https://blog.letpeople.work). LetPeopleWork GmbH, Switzerland. Built by Peter Zylka-Greger and Benjamin Huser-Berta. Contact: contact@letpeople.work. Machine-readable summary: [https://letpeople.work/llms.txt](https://letpeople.work/llms.txt) # Part 3: Guides ## How to forecast Jira releases without story points URL: https://letpeople.work/jira-forecasting-without-story-points/ (updated 2026-10-17) Someone asks when the release will be done. You could add up the story points and divide by velocity. Or you could count what the team really finished over the last weeks and let a Monte Carlo simulation give you dates with their odds. This guide walks through the second way, and says where it falls short as it goes. ### Why story points make a weak forecast Story points were meant to start a conversation about a piece of work. As an input for a forecast they cause trouble. When Peter plotted his teams' story points against how long the items really took, he found almost no relationship: items estimated at 2 points took anywhere from a few days to almost 160 days ([Reality check through scatter plotting](https://blog.letpeople.work/p/reality-check-through-scatter-plotting)). Estimates are noisy as well. Two teams estimating similar features can land far apart, and once teams get compared on their numbers, a 5 has a way of turning into an 8 ([The hidden cost of noise](https://blog.letpeople.work/p/the-hidden-cost-of-noise-in-your), [Why you should compare metrics of different teams](https://blog.letpeople.work/p/why-you-should-compare-metrics-of-different-teams-with-each-other)). Divide one noisy number by another and you get a date that looks precise. If your team finds points useful for talking through a piece of work, keep them for that. You don't need them to forecast. ### What you need instead: finished items and their dates Throughput is the number of items your team finishes per day. Jira already records it, because every issue that reaches Done has a date. There is no extra field to fill in and no estimation meeting. Each item counts as one, whatever its size. That sounds crude, and it works well enough as long as items are kept reasonably small (more on that further down). Holidays and sick days are already in that history, so you don't have to plan capacity on top of it. ### How a Monte Carlo forecast works Take the last 30 days and how many items the team finished on each of them. To forecast, the simulation builds one possible future: for every day ahead it picks one of those past days at random and uses its count. It keeps adding until the remaining items are done, or until the target date is reached. Then it builds the next future. Lighthouse does this 10,000 times, and the share of futures that make it is your probability. That gives you two kinds of answers: - **When will these items be done?** You get dates with their odds, for example 50% by one date and 85% by a later one. - **How many items will be done by a date?** You get counts with their odds. Benji's [step-by-step guide to Monte Carlo simulations](https://blog.letpeople.work/p/an-introduction-and-step-by-step-guide-to-monte-carlo-simulations) builds the same thing in a spreadsheet, if you want to see every step. ### Which date to promise The 50% date is a coin flip: half of the simulated futures finish later. That's fine for planning inside the team. For a customer or a steering committee, Peter's advice is short: use the 85% date and sleep well. When missing would be expensive, use 95%. An 85% date still misses roughly one time in seven. Say so when you share it. A range with its odds makes for a better conversation than a single date that everyone suspects is a guess ([Dear stakeholder, here's why I'm giving you a range](https://blog.letpeople.work/p/dear-stakeholder-heres-why-im-giving)). ### Does it hold up? Check it against your own past Benji compared both methods on a real team with six developers. Their average from the previous month said 37.5 items in the next 30 days. The Monte Carlo forecast said 38 items at 50%, 35 at 70%, 31 at 85% and 27 at 95%. The team finished 30, one fewer than the 85% forecast and well short of what the average promised ([The flaw of averages](https://blog.letpeople.work/p/the-flaw-of-averages-comparing-monte-carlo-simulations-with-estimates-based-on-averages-267528ae4f47)). When Peter backtested his own team, the 85% forecast said 12 items for two weeks. They finished 16 ([How Lighthouse changed the way I work](https://blog.letpeople.work/p/how-lighthouse-changed-the-way-i)). You can run the same check on your data. Lighthouse's Forecast Reality Check forecasts weeks that already happened, as if they hadn't, and compares the result with what the team really finished, for each percentile and history window. ### What about items of very different sizes? Counting items doesn't need them all to be the same size. It needs them small enough that one giant ticket can't swallow a month. The common way to get there is a service level expectation (SLE): the team agrees that 85% of its items should be done within, say, 10 days. Before anyone starts an item, the team asks one question. Can we finish this within our SLE? If not, split it first. [Sizing Poker](https://letpeople.work/sizing-poker/) asks that question for a whole backlog in a few minutes, in the browser and for free. Epics that have no child items yet get a size from your history. Lighthouse can assume, for example, as many items as 85% of your past features needed. In Benji's own team, 85% of features had 7 child items or fewer ([Low WIP, small features, high freedom](https://blog.letpeople.work/p/low-wip-small-features-high-freedom)). ### Where it falls short - **It assumes the next weeks look roughly like the last ones.** A reorganisation, a new team or a holiday season changes that. Pick a history window that resembles what's coming: a team that formed a month ago shouldn't use three months of history. - **It can't see work that doesn't exist yet.** In one case study a release grew from 29 items to 63, with 34 added along the way, 26 of them bugs. Forecasting the original 29 said 9 April, and the release shipped on 25 April. Re-run with all 63 items, the 85% forecast said 26 April. So re-forecast whenever the scope moves ([Monte Carlo simulations, accuracy and unplanned work](https://blog.letpeople.work/p/monte-carlo-simulations-accuracy-and-unplanned-work-a-case-study-9dab4d36bd02)). - **Several teams on one feature lower the odds.** All of them have to make it. Five teams that are each 85% sure give you a joint date that is close to a coin flip ([Forecasting across several teams](https://blog.letpeople.work/p/an-introduction-and-step-by-step)). - **Spiky throughput gives wide ranges.** That's useful information in its own right: predictability is something a team can work on. ### What Jira does on its own Jira Plans, part of Jira Premium and Enterprise on Cloud and called Advanced Roadmaps on Data Center, schedules work from estimates, team capacity and velocity, in story points or hours. Items without an estimate are left out of the capacity ([capacity and velocity in Plans](https://support.atlassian.com/jira-software-cloud/docs/what-are-capacity-and-velocity-in-advanced-roadmaps/)). The velocity chart can count issues instead of points, and the control chart shows cycle time with an average. A forecast with its odds isn't part of Jira, as far as Atlassian's documentation shows. Marketplace apps such as ActionableAgile add Monte Carlo charts inside Jira. Lighthouse does it from outside, on your own systems ([how they compare](https://letpeople.work/compare/)). ### Doing it in Jira with Lighthouse A spreadsheet is fine for a one-off forecast. Lighthouse keeps forecasting against your Jira as items finish: 1. Install Lighthouse on your laptop or a server, as a desktop app, a Docker container or on Kubernetes. 2. Connect Jira. Jira Cloud uses your username and an API token. Jira Server and Data Center, version 8 or later, use a personal access token, so a Jira that is only reachable inside your network works as well. 3. Add a team with a JQL query, for example project = "SHOP" AND labels = "checkout". Then say which issue types count and which states mean doing and done. 4. Ask when a set of items will be done, or how many will be done by a date. Epics are forecast too, including epics that several teams work on. Lighthouse doesn't use story points for its forecasts. If you keep estimating anyway, it can show how your estimates compare with how long items really took. On paid editions it can also write the 85% date back into a Jira field, so people see it where they already look. About two weeks of finished work is usually enough for a first forecast. The [documentation explains how Lighthouse forecasts](https://docs.lighthouse.letpeople.work/concepts/howlighthouseforecasts.html) in more detail. ### Questions people ask **Can I forecast in Jira without story points?** Yes. Count the items your team finishes per day and run a Monte Carlo simulation over that history. You get dates with their odds, for example the date you are 85% likely to make. Jira already records when each issue reached Done, so you need no extra field and no estimation meeting. **How much history do I need?** About two weeks of finished work is usually enough for a first forecast. Use a history window that looks like the weeks ahead, typically 30 to 90 days, and forecast again as new items finish. **Which percentile should I commit to?** The 85% date for anything you promise outside the team. The 50% date is a coin flip, which is fine for planning inside the team. Use 95% when missing the date would be expensive. **What if our items are very different in size?** Agree on a service level expectation, for example that 85% of items are done within 10 days, and split anything that won't fit before you start it. Counting items then works, because no single item is big enough to swallow the forecast. **Does Jira have Monte Carlo forecasting built in?** Not as far as Atlassian's documentation shows. Jira Plans, called Advanced Roadmaps on Data Center, schedules work from estimates, team capacity and velocity in story points or hours, and leaves items without an estimate out of the capacity. Marketplace apps such as ActionableAgile, or a separate tool such as Lighthouse, add Monte Carlo forecasts. **Does this work with Jira Data Center?** Yes. Lighthouse connects to Jira Server and Data Center from version 8 with a personal access token. Because you run Lighthouse yourself, it can sit in the same network as your Jira. ## Delivery forecasting for Azure DevOps URL: https://letpeople.work/azure-devops-forecasting/ (updated 2026-10-17) Your backlog lives in Azure DevOps and someone wants a date. This page shows how to forecast from the work items your teams have already closed, with the odds attached, and how Lighthouse connects to Azure DevOps Services or an Azure DevOps Server inside your network. ### What Azure DevOps offers on its own The Forecast option on a backlog draws lines for the coming sprints from a velocity you type in. It sums Effort or Story Points, so it depends on how well the team estimates, and Microsoft's own documentation says it doesn't account for dependencies or blockers ([Forecast your product backlog](https://learn.microsoft.com/en-us/azure/devops/boards/sprints/forecast?view=azure-devops)). The Velocity widget can count work items instead of points, and Analytics adds a cumulative flow diagram and cycle time charts with averages. What you won't find in Microsoft's documentation is a forecast that gives you a date with its odds. For that you need a Monte Carlo simulation over your throughput. ### Forecasting from closed work items Throughput is the number of work items a team closes per day. Azure DevOps already knows it, because every item records when it reached its done state. A Monte Carlo simulation builds thousands of possible futures from those past days and counts how many of them make it. You get answers like "85% likely by this date" for a set of items, or "85% likely to finish at least this many" by a date. No estimates are involved. Benji wrote about running these simulations against Azure DevOps with a nightly script back in 2023 ([Running Monte Carlo simulations in Azure DevOps](https://blog.letpeople.work/p/running-monte-carlo-simulations-in-azure-devops-728746ef15be)). The method in detail, with which date to promise and where it falls short, is in our guide to [forecasting without story points](https://letpeople.work/jira-forecasting-without-story-points/). It applies to Azure DevOps just the same. ### How Lighthouse connects to Azure DevOps - **Connection.** You give Lighthouse your organisation URL and a personal access token. Paid editions can sign in with OAuth through a Microsoft Entra ID app registration instead, for Azure DevOps Services. - **Azure DevOps Server.** Lighthouse runs on your own systems, so it can sit in the same network as an Azure DevOps Server and connect with a personal access token. Nothing has to be opened up to the internet. - **Teams.** A team is a WIQL query, usually by area path, for example [System.TeamProject] = "Shop" AND [System.AreaPath] = "Shop\Checkout". You choose which work item types count, such as User Story, Product Backlog Item and Bug, and map your states to to do, doing and done. A board wizard can read the query, types and states from an existing board for you. - **Blocked work.** Rules on any field, a tag for example, mark items as blocked. They feed the blocked and stale item views. The work items Lighthouse reads are stored in its own database on your systems. The [Azure DevOps setup guide](https://docs.lighthouse.letpeople.work/concepts/worktrackingsystems/azuredevops.html) in the documentation has the details. ### Features and epics across teams A portfolio is a second WIQL query for your features or epics. Lighthouse follows the parent links from team items to their feature and forecasts every feature in backlog order, continuously. When several teams work on one feature, it simulates each team and combines the results into one probability, which is never higher than the least likely team's, because all of them have to make it. Features without child items yet get a size from a field you choose, or from your history, for example as many items as 85% of your past features needed. Lighthouse flags those features, so nobody mistakes an assumed size for a real one. Predecessor links show up as dependencies. ### Forecasts back in Azure DevOps On paid editions, Lighthouse can write forecasts into a field on the work item, for example the 85% completion date, or the item's age. People then see the forecast where they already look, without opening another tool, and Azure DevOps doesn't send notifications for those updates. ### Other options for Azure DevOps Lighthouse isn't the only way to get Monte Carlo forecasts into Azure DevOps. On the Visual Studio Marketplace you'll also find ActionableAgile Analytics for Azure DevOps Services, Agile Analytics by Baytek, Honest Cheetah by Benjamin Day Consulting, and FlowViz by Nicolas Brown, a Power BI template for flow metrics. Lighthouse is a separate application that you run yourself. That lets it bring teams on Jira, Linear or ServiceNow into the same portfolio forecast as your Azure DevOps teams. For ActionableAgile and Nave there is a [side-by-side comparison](https://letpeople.work/compare/). ### Questions people ask **Can Azure DevOps forecast delivery without story points?** Not with its own Forecast tool, which works from a velocity in Effort or Story Points. You can forecast from throughput instead: count the work items each team closes per day and run a Monte Carlo simulation over that history. Lighthouse does this against Azure DevOps and gives you dates with their odds. **Does Lighthouse work with Azure DevOps Server on-premises?** Yes. Lighthouse connects to Azure DevOps Server with a personal access token. Because you run Lighthouse yourself, it can sit in the same network as the server, so nothing has to be reachable from the internet. **How do I tell Lighthouse which work items belong to a team?** With a WIQL query, usually by team project and area path, plus the work item types that count and how your states map to to do, doing and done. A board wizard can take all three from an existing board. **Can Lighthouse write the forecast back to Azure DevOps?** Yes, on paid editions. Lighthouse can write values such as the 85% completion date into a field on the work item, so people see the forecast in Azure DevOps. **What does it cost?** The Community edition is free for up to 3 teams and 1 portfolio, with Monte Carlo forecasts and flow metrics. A Premium license, sold as Self-Service or Enterprise, adds unlimited teams and portfolios, OAuth sign-in and writing forecasts back to Azure DevOps. ## Lighthouse vs ActionableAgile vs Nave URL: https://letpeople.work/compare/ (updated 2026-10-17) All three show cycle time, throughput, work in progress and work item age, and all three run Monte Carlo forecasts from your throughput instead of estimates. They differ in where they run, where your data ends up and how they charge. **The short version:** ActionableAgile and Nave are established commercial products. ActionableAgile charges per user and Nave per connected board. Lighthouse is the self-hosted, source-available alternative with a free edition and a flat yearly licence. If your tracker only lives inside your network, or you work with Azure DevOps Server, Linear or ServiceNow, that difference decides most of it. We build Lighthouse, so read our framing with that in mind. The facts in the table come from the vendors' own pages, linked below it. ### Side by side | | Lighthouse | ActionableAgile | Nave | | --- | --- | --- | --- | | Made by | LetPeopleWork GmbH, Switzerland | 55 Degrees AB, Sweden | Nave | | Where it runs | On your systems: desktop app for Windows, macOS and Linux, Docker, or Kubernetes with the official Helm chart | Standalone web app, apps for Jira Cloud and Jira Data Center, and an extension for Azure DevOps Services | Web app, with an app for Jira Cloud and Data Center | | Where your data ends up | In Lighthouse's own database on your systems. Usage data to us is off unless someone turns it on, and never contains work items. | The Data Center app runs inside your own Jira. The other editions run in the vendor's or Atlassian's cloud; the vendor says it processes work data without exporting it. | Nave's cloud, hosted in the EU. A Jira Data Center has to be reachable from the internet, or you upload exports. | | Trackers | Jira Cloud, Jira Server and Data Center, Azure DevOps Services and Server, Linear, ServiceNow, CSV | Jira Cloud, Jira Data Center, Azure DevOps Services, Trello, file upload | Jira, Azure DevOps (Server through exports), Trello, Asana, CSV and JSON | | Forecasting | Monte Carlo "when" and "how many" per team, continuous forecasts for every feature in a portfolio, features shared by several teams, and a backtest of its own forecasts | Monte Carlo "when" and "how many". Multi-team views come with Portfolio Forecaster, a separate product for Jira Cloud. | Monte Carlo completion dates and "how many" at 50, 85 and 95% | | Source code | Public on GitHub under a source-available licence | Not public | Not public | | Free to use | Community edition, free for up to 3 teams and 1 portfolio with 1 delivery. Premium trial for 30 days, no sign-up or credit card. | 30-day trial with all features for up to 3 users. No free edition. | 14-day trial, no credit card. No free edition. | | Pricing model | Flat yearly Premium licence, not per user: Self-Service CHF 2,000 a year, or Enterprise CHF 10,000 a year with priority support | Per user, monthly or yearly | Per connected board, with unlimited users | Checked in October 2026 against the vendors' public pages. Prices and features change, so check the details at [55degrees.se](https://www.55degrees.se/pricing) and [getnave.com](https://getnave.com/pricing). Spotted something wrong? Tell us at [contact@letpeople.work](mailto:contact@letpeople.work) and we'll fix it. ### When ActionableAgile or Nave fit better If you want a managed web app with nothing to install, both are mature and well regarded. ActionableAgile grew out of Daniel Vacanti's work on flow metrics, and its Jira apps put the charts right inside Jira. If your work lives in Trello, both of them support it, and Nave also supports Asana. Lighthouse supports neither. ### When Lighthouse fits better Lighthouse is the better fit when the data has to stay inside your network, for example with a Jira Data Center or an Azure DevOps Server that isn't reachable from the internet. It also fits when per-user pricing gets expensive across many teams, or when you want to try the full core for free without a procurement round. One installation can connect to Jira, Azure DevOps, Linear and ServiceNow at the same time, so teams on different trackers end up in one portfolio forecast. Lighthouse also has an MCP server, so AI assistants such as Claude or Copilot can query your flow data, and its Forecast Reality Check tests the forecasts against weeks that already happened. Because the source is public, your security team can read what it does with your data. ### Looking for a self-hosted alternative to ActionableAgile? If your Jira or Azure DevOps sits inside the company network, a cloud tool first needs a way in. If the licence is per user, the bill grows with every team that starts using it. Lighthouse runs on a laptop or a server next to your tracker, so you can check its forecasts on your own data before anyone has to approve a new vendor. See [how forecasting without story points works in Jira](https://letpeople.work/jira-forecasting-without-story-points/), or [how Lighthouse connects to Azure DevOps](https://letpeople.work/azure-devops-forecasting/). ### Questions people ask **Is there a self-hosted alternative to ActionableAgile and Nave?** Yes. Lighthouse runs on your own systems, as a desktop app, a Docker container or on Kubernetes, and keeps your work items in its own database there. It connects to Jira Cloud and Data Center, Azure DevOps Services and Server, Linear and ServiceNow. The Community edition is free for up to 3 teams; paid editions are a flat yearly licence instead of per-user pricing. ActionableAgile's Jira Data Center app also runs inside your own Jira. **How does Lighthouse compare to ActionableAgile?** Both show flow metrics and run Monte Carlo forecasts for when work will be done and how much will be done by a date. ActionableAgile is sold per user as a web app, as Jira Cloud and Data Center apps and as an Azure DevOps Services extension. Lighthouse is self-hosted and source-available and free for up to 3 teams. Portfolio forecasts and features shared by several teams are part of the same product, and it also connects to Linear and ServiceNow. **How does Lighthouse compare to Nave?** Both show flow metrics and run Monte Carlo forecasts from throughput. Nave is a web app hosted in the EU, priced per connected board, and supports Trello and Asana as well as Jira and Azure DevOps. Lighthouse runs on your own systems, so it also reaches a Jira or Azure DevOps Server that is not on the internet, and it uses a flat yearly licence with a free edition. **Does Lighthouse charge per user?** No. The Community edition is free for up to 3 teams and 1 portfolio with 1 delivery. The paid features are called Premium in the app and the documentation, and two licences include them: Self-Service at CHF 2,000 a year, and Enterprise at CHF 10,000 a year, which adds priority support. Neither has limits on teams or portfolios.