Why the category splits four ways, and which one you need

Search for an AI sales tool and four different products come back on one page. A notetaker. A revenue intelligence platform. A CRM. And something calling itself a workspace, which is the kind we build. The screenshots look alike. The purchases are not alike at all.

They are not really competing on quality. They answer different questions, and each answers its own question well and the other three badly. Buying the wrong kind does not feel like a weak feature list. It feels like paying every month for a good answer to a question you did not have.

Several sections below end with us telling you to buy something that is not ours.

Four questions, not four feature lists

Start with the object each product is built around. A notetaker is built around the meeting: one conversation, made readable afterwards. Revenue intelligence is built around the pipeline in aggregate: all the conversations at once, read so a leader can see what is happening across them. A CRM is built around the record: the deal, the contact attached to it, and where each one stands. A workspace is built around what comes after the call: the quote, the document, the invoice, the task nobody wrote down.

A notetaker's job finishes when the summary lands

Fathom, Fireflies and Otter are good products and this is not a setup for a punchline. Fathom is unusually generous: a real free tier, a clean product, summaries that land fast. For one person who wants to stop taking notes, we would recommend it ourselves. Otter is the general-purpose version of the same job, and general is a strength rather than a compromise: if most of what you record is not a sales call, buy that.

Fireflies is the mature, broadly integrated one. If your CRM is HubSpot or Salesforce and native sync matters to you more than anything else here, buy Fireflies and stop thinking about it. We have no HubSpot integration and no Salesforce integration: both are reachable from our side through the REST API and webhook steps, by hand, which is a worse answer than a native sync.

The category limit is structural rather than a gap someone forgot to fill. The job finishes when the summary is delivered. It lands in your CRM as a paragraph, so the assistant there reasons about a paragraph rather than about the call. And your price list lives somewhere the notetaker cannot see, which is why a summary can never tell you the quote was wrong.

Revenue intelligence is for whoever has to defend a forecast

Gong is the most established product in this space and very good at what it was built for: reading a whole sales organisation's conversations and telling leadership what is happening across the pipeline.

The buying test is simple. Can you name the number you are being asked to defend, and are there enough sellers for a pattern to exist across them? Conversation analytics over a handful of reps is anecdotes with a chart on top. Over dozens, the depth is real and hard to replicate.

There is a second reason to buy it that has nothing to do with features. Some purchases have to clear procurement, and the posture of an established vendor is the product. On that axis we are the wrong answer: we hold no SOC 2 and no HIPAA.

What revenue intelligence does not do is the selling. It reads it. It sits alongside a CRM, a quoting tool and a billing system, and assumes you already have all three.

A CRM tells you where the deal is, not what was said

Two quite different products get called CRM. Pipedrive is one of the clearest pipeline CRMs ever built, and small teams adopt it because the mental model is obvious on day one. If nobody knows which deals are where, that is the answer, and you do not have to buy a suite to get it.

HubSpot is a platform you configure, and the configuring is the price of what it then does: marketing, sales, service and a CMS in one account, a large marketplace, and agencies who already know it. Buy it if you will need all of that, if someone's job is to own the CRM, or if you depend on partners who work in it.

The category limit is the same in both. A card at a late stage, and a record of what the client pushed back on, what you quoted and what you still owe them, are different kinds of knowledge, and only one of them survives the seller leaving.

Which brings up the arrangement most teams end up with, and it deserves saying plainly: a CRM plus a notetaker is a good default. Each piece is best in class and you can replace one without replacing the others. Most companies should probably do this. What it costs is the seams: two subscriptions become four, each with its own seat count, and each assistant sees only its own tool.

The workspace is a different bet, and depth is the price

The fourth kind starts from one wager: that a single AI reading everything beats three assistants each reading a slice.

In practice the call, the price list, the proposal, the invoice and the tasks are one product rather than four integrated ones. The copilot quotes the real number during the call because your price list is data in the same workspace, not a PDF nobody can find. The follow-up is drafted from what was said and the products you quoted, and what is owed sits on the client record beside the call it came from.

Here is the cost, in the same words our comparison pages use. Best-in-class depth in each individual area: no. A specialist will beat us in its specialty, because that is what a specialist is. Swap one vendor without touching the others: no. You are buying the joins, and joins are the part you cannot unbundle later.

Two limits from the other essays belong beside that trade. Autopilot is off until you turn it on, and nothing in it touches your money at all, which the essay on how much rope it gets walks lock by lock. And this is a European product: the analysis comes back in the language the call happened in, the contract is governed by Lithuanian law, and the product sets no advertising or analytics cookies, which the essay on selling from Europe goes through in full.

How to tell which one you are

If you want a record of what was said and nothing after it, buy a notetaker. If most of what you record is not a sales call, buy general transcription. Both are narrower purchases than a workspace, and the narrower purchase asks less of you. Fathom has a free tier, so you can find out whether the summary was the thing you wanted before you pay anyone.

If someone is asking what is happening across the pipeline, and there are enough sellers for the question to have an answer, buy revenue intelligence and keep the CRM you have. If nobody knows which deals are where, buy the CRM: the suite if you also need marketing, service and a CMS, the pipeline board if not.

If the paperwork after the call is where your week goes, and you have not assembled the stack yet, the workspace is worth a look. That last clause carries weight: if you already own the pieces and they work, the seams you have are the ones you know. And if your process is genuinely unlike anyone else's, building it yourself is not a mistake. Just cost the second year rather than the first: a call room that survives hotel wifi, an OAuth grant per calendar vendor, an AI budget somebody has to meter.

Each of these products has a page here with the row-by-row version, dated, and a section on when to buy theirs. Read the one for the kind you have decided you are. If that turns out to be this one, signing up builds the workspace, switches the copilot on and puts credits in it, with no call to sit through.

Questions this raises

I already pay for a notetaker. Does a workspace mean dropping it?

Usually yes, and if that feels wrong it is a signal worth listening to. Two products that both record and summarise your calls is two subscriptions doing one job, and the specialist is better at that job. The real question is whether the summary is the thing you are buying. If it is, keep the notetaker and ignore the rest of this. If what you needed was the quote, the proposal and the invoice that follow it, the summary was never the purchase.

How do I know I need revenue intelligence rather than one of the other three?

You can name the number you are being asked to defend, and you have enough sellers for a pattern to exist across them. Conversation analytics over a handful of reps is anecdotes with a chart on top. If leadership is asking what is happening across the pipeline rather than what happened on Tuesday's call, that is the buy, and it is not the one we make.

Is one workspace cheaper than the stack?

One bill and one seat count instead of three or four. That is the part we can state plainly, and it is where a saving would have to come from if there is one. We are not going to print anyone else's prices to prove it. What we will name is the trade. You give up best-in-class depth in each area, and you give up replacing one vendor without touching the others. If the money is close, decide on that instead.