Question 1

How do I tell whether a tool is free or part of the licence?

Accepted Answer

Every article carries a badge, and the badge is not a description — it is read from the entitlement set the app enforces, so it cannot disagree with what happens when you press the button. 31 of the 53 tools are free with a free account (3 downloads a day, with a small PDFMacro line on each page of a saved PDF); the other 22 are included in the one lifetime licence. A few free tools host a licensed capability inside them — verified redaction inside the free Redact, table extraction inside free Extract — and those are named on the free tool's own article rather than left to be discovered at the gate.

Question 2

Where does the text in these articles come from?

Accepted Answer

The same place the in-app assistant answers from. Each article's summary, explanation, alternative names and example questions are read out of the assistant's knowledge base at build time, and each Free or Lifetime badge out of the entitlement set that enforces it. Nothing on this page is written for this page, which is why the help centre and the assistant cannot tell you two different things about the same tool.

Question 3

How do I get help for the tool I am using right now?

Accepted Answer

Press the Help control in the Workspace header while a tool panel is open and you land on that tool's article — the Redact panel opens the Redact article. With no panel open it opens this index instead. The Workflows builder has the same control, pointed at the branch, retry and variables reference, and the Workflows gallery at the Workflows overview. All of them open this one page; none of them keeps a second copy of the text.

Question 4

Are the screenshots in the walkthroughs current?

Accepted Answer

They are real captures of the running app — not mockups, not drawings — taken on the free tier with nobody signed in, so the padlocks in them are the ones you would see. But they are taken by hand when a guide is written or revised, not regenerated on every release. A test proves the image on the page is the exact file that was captured and that no guide names a control the app no longer has; nothing proves the picture matches this week's build. Trust the numbered steps over the pixels, and tell us if a panel has moved.

Question 5

Do the AI features need an internet connection?

Accepted Answer

46 of the 53 tools never touch the network once the page has loaded. 3 download a small on-device model the first time you use them and run offline afterwards — the weights are cached in your browser's storage on this device. Chat with PDF, Summarize, Write with AI and Translate run on your own AI: with a key they need a connection to your provider, and with a local model they need your Ollama or LM Studio running.

Question 6

My own Ollama server will not connect — what do I fix?

Accepted Answer

Almost always the origin. A browser will not let this page read another origin's server unless that server says it may, and Ollama refuses unknown origins by default — the refusal reaches page JavaScript looking identical to “nothing is listening”. Set OLLAMA_ORIGINS to this app's origin, restart Ollama, and press Test connection again. LM Studio has a CORS toggle in its server settings instead. The connection test names which of the two problems you actually have.

Question 7

Should I connect a key or a local model?

Accepted Answer

Either works with every AI feature. A key with OpenAI, Anthropic or Google gives you the most capable models; the text you ask about goes to that provider under your account, and they bill you for it. A local model in Ollama or LM Studio is free and keeps everything on your machine, and how capable it is depends on your machine. For privileged or sensitive material, a local model is the one where nothing leaves.