AI & Tools

Three Free Tools for Better Calls (No Signup, Nothing Uploaded)

Uma Mahesh Bandaru Uma Mahesh Bandaru August 16, 2026 4 min read
Three Free Tools for Better Calls (No Signup, Nothing Uploaded)

We build a desktop assistant that listens to live calls. In building it we kept running into three questions that are surprisingly hard to answer and that nobody seems to measure, so we built a small free tool for each and put them at craqly.com/tools.

All three run entirely in your browser. No account, no email, nothing uploaded. That is not generosity — it is what makes them free to run at any volume, and it is why there is no reason to put a limit or a login in front of them.

1. What your interviewer actually sees

Pick your meeting app, your operating system and what you are sharing, and the screen share checker gives you a straight answer on what lands in the capture: overlay windows, your cursor, notification banners, the Dock or taskbar, the screen-recording indicator, and what a second monitor does.

The part we care most about is the labelling. Every claim carries one of four markers — traceable to our own source code, tested by us, documented platform behaviour, or not verified by us. The last one is on there because we do not own a Windows machine, so we have not put the Windows exclusion path in front of a real capture ourselves. Saying that is more useful than implying a test we did not run.

It also spells out what nothing can hide. The operating system draws your mouse cursor into the capture, so an invisible window does not make the pointer moving over it invisible. That single fact catches more people than any software setting.

2. What that meeting actually costs

The meeting cost calculator builds the room from role presets, applies a loaded-cost multiplier — because base salary is not what an employer pays for an hour — and gives you the cost of one occurrence and of a year of it.

Two decisions in it are worth explaining. The salary bands are our own estimates, and it says so rather than attaching a source that does not exist; every figure is editable, and replacing them with your payroll numbers is the point. And each currency has its own local band rather than an exchange-rate conversion of the dollar figure, because a US salary converted into rupees or euros describes nobody.

There is a “could this have been an email?” verdict, and it shows every rule it fired rather than hiding the logic, so you can disagree with it.

3. How much of the call was you

The talk ratio analyzer measures talk share, longest unbroken stretch, turn count and average turn length, and draws the whole call as a waveform. Record straight into the page to practise a pitch, or load a recording you already have.

The audio never leaves your browser, and you can check rather than trust us: open developer tools, watch the Network tab, and run an analysis. Nothing goes out. That also sidesteps the consent problem that comes with uploading a recording of a real customer call to a third party.

It is honest about its main limit. A single mixed track cannot reliably be split into two speakers by loudness alone, so with mixed audio it reports speech versus silence — which is exactly right if you record only yourself. When a file genuinely has two different channels it analyses each separately and says so, and it checks whether the channels actually differ first, because mono duplicated across a stereo pair would otherwise invent a 50/50 split out of nothing.

Why they all run in the browser

It started as a cost decision and turned into a design principle. A tool that processes everything locally cannot be rate-limited, cannot leak what you put into it, and cannot quietly become a paid tier because the bill got uncomfortable. It also means the privacy claim on the talk ratio tool is verifiable rather than a promise — the measurement code has no network access in it at all.

What they do not do

They measure after the fact. That is the honest gap between a free tool and the product: the analyzer tells you your longest monologue once the call is over, whereas the Craqly desktop app tracks talk ratio live with genuine dual-channel separation and flags a monologue while it is still happening, which is the moment the information is worth anything.

If the free version tells you what you needed to know, that is a good outcome. They are useful on their own, which is the only reason to build them.

Frequently asked questions

Are these tools actually free?

Yes, with nothing to sign up for. All three run in your browser, so they cost nothing per use and there is no reason to put a limit or a login in front of them. No email address, no trial, no card.

Is anything I enter or record sent to Craqly?

No. The screen share checker and the meeting cost calculator are pure client-side logic with no server call at all, and the talk ratio analyzer processes audio in the browser tab — recordings are reduced to loudness numbers as they arrive, and files are decoded locally. You can confirm it in the Network tab of your browser’s developer tools while you use them.

Do they work on a phone?

Yes, all three are built mobile-first. The talk ratio analyzer can record through your phone’s microphone, though a call recording from a desktop generally gives a cleaner result, and browsers throttle background tabs so keep the page in front while recording.

Can I use the results at work?

That is what they are for. The meeting cost calculator produces a shareable link that carries your inputs, so the figure arrives with its assumptions attached and whoever owns the invite can check the maths rather than argue with a number in a message.

Uma Mahesh Bandaru

Written by

Uma Mahesh Bandaru

Writes about live interviews, sales calls and meetings, and how real-time AI assistance changes each of them.

Scroll to Top