Honest answers to the questions we get asked most. If yours isn't here, email us.
No, and the distinction matters. LogicReader does two things:
Rule of thumb: use LogicReader to understand what your P6 schedule is saying. Use P6 (or MSP, Asta) for the dates and resources that go in the contract.
Yes — the early-access toolkit imports P6 XER, computes critical path, lets you compare revisions, isolate scopes, and export to GraphML for yEd. That's a real workflow you can run on your own schedule today.
Smoothly to ~25,000 activities on a modest laptop. Up to 50,000 with a small first-load pause. Above that, expect to need a discrete GPU or to load by sub-programme.
Nothing to start. The free tier is £0 for as long as you use it and give us feedback, with two limits only: export back out to Primavera P6, and the size of the project. Collaboration is free on every tier. One paid tier, LR Plus — the Products page has the price and what it adds. Subscribe opens before the end of 2026, once the commercial licence for our visualisation engine is in place.
You do not need one. Create your free account with an email address and the product is yours: £0 for as long as you use it and give us feedback. No card, no time bomb. We ask for honest feedback instead of money — we read it, we assess it, and the product changes because of it.
The live demo runs on a single server in Europe. Your schedule data is held on that one server. It is not replicated, not used for training any model, and not shared. With the planned desktop version, your file stays on your own machine and never crosses the network.
Yes. CosmosPM Ltd is registered with the UK Information Commissioner's Office (ICO ZC143102). Our Privacy Policy describes what we collect and why. Data processing agreements are in place with each of our sub-processors: Hostinger, Anthropic, Brevo and Microsoft, and PostHog when it is enabled. Paddle, the reseller that sells our paid plans, is not a sub-processor: it processes order and payment data as an independent controller, under its own privacy notice. Paddle never receives your schedule data.
No. Every date, float and critical path is computed by LogicReader's own deterministic engine — never by AI. Some AI features can add schedule content when you ask them to: the AI Draft Builder turns a scope document into a first-draft schedule, and the logic-type classifier suggests a logic type for each link. Both run only when you invoke them, both are off by default, and both mark what they add as AI-sourced so you can review it. You are asked before anything is sent, and your answer holds for that session. You can bring your own API key or account (Anthropic, OpenAI, Mistral, or a local model).
No. The AI features we have or plan are read-only assistants — they help you understand your schedule. Nothing you load is used to train or fine-tune a model. The AI vendors we use (currently Anthropic) are bound by contract to the same.
Yes — for enterprise pilots we offer an on-premise option inside your own perimeter. Pilots usually start on the hosted demo, then move on-premise before any production schedule data is loaded. Talk to us about your environment.
XER from P6 7.0 (2010) onwards. Older XERs may parse but are not supported.
Today the import is Primavera P6 (XER). Native connectors for Microsoft Project, Asta and others are on the roadmap and follow real demand — if you need one, ask us.
Yes — the web app runs in any modern browser, Mac included. A desktop version is planned for Mac, Windows and Linux.
Evangelos Kourentzis — chartered mechanical engineer, twenty-seven years in UK infrastructure, P6 specialist, founder of CosmosPM Ltd. Currently building the product hands-on, with input from a small group of pilot users.
CosmosPM is self-funded today and built deliberately with a small group of pilot users. We're a small company and say so plainly.
Honest answer: that depends on whether enough projects choose it for it to stand on its own commercially — which is true of any small product. Two things hedge the question. One: a desktop version you keep means a worst-case scenario still leaves you with a usable copy. Two: every export is in open formats (GraphML, CSV) so your data is never trapped.