Intercom Is Called Fin Now And Salesforce Is Buying It

Two Bits Of News In About Six Weeks

If you have been busy running a business and not reading tech press, here is what you missed.

In May 2026, Intercom renamed itself Fin. Not a product line, the whole company. Fin was the name of their AI agent, the AI agent became the main event, and so the company took the name of its most successful child.

Then on June 15, 2026, Salesforce announced a definitive agreement to buy them for somewhere around 3.6 billion dollars. Expected to close in Salesforce fiscal 2027.

So the tool you installed to answer questions on your pricing page is now a differently named company being absorbed into the largest CRM vendor on earth. Which is quite a journey for a chat bubble.

Does This Actually Affect You Tomorrow

Short answer, no. Longer answer, also mostly no, but keep an eye on it.

Signed is not closed. Deals this size take months and involve regulators who move at the pace of continental drift. Your widget will keep working. Your login will keep working. Nobody is going to email you a migration deadline next week.

I want to say that clearly because there is a genre of blog post right now that is basically panic dressed up as analysis, usually written by a competitor who would love you to switch to them today. I am also a competitor, so treat me with the same suspicion. But the honest read is that the acquisition itself is not an emergency.

What acquisitions do change, eventually, is priorities. Salesforce did not spend 3.6 billion dollars to get better at serving three person startups. Their customer is the enterprise. Over the next couple of years, that is where the roadmap attention goes, and small accounts usually feel that as a slow drift rather than a sudden shock. Features you liked get deprioritised. Support gets a bit more distant. Pricing gets restructured in a way that happens to suit larger contracts.

We literally just watched this play out with Drift, which got acquired, acquired again, and switched off in March 2026. I am not predicting the same ending here, and Fin is a far bigger and healthier business than Drift was. But the pattern is worth knowing.

The Thing You Should Actually Be Looking At

Forget the acquisition for a minute. There is a live issue with the pricing model that matters much more to a small team, and it exists regardless of who owns the company.

Fin AI charges 99 cents per resolution.

Not per seat. Not per month. Per time the AI handles something. And on top of that you are still paying seats, which run 29 dollars each on Essential annual, 85 on Advanced, and 132 on Expert.

Here is a real shaped example that gets quoted a lot. Eight agents on the Advanced plan, with 2,100 Fin resolutions in a month, comes to roughly 2,960 dollars a month. That is around 35,500 dollars a year to answer customer questions.

And What Exactly Is A Resolution

This is the bit that I think is genuinely worth being annoyed about.

You would assume a resolution means the AI solved the problem. Customer asked a thing, AI answered correctly, customer went away happy, 99 cents well spent.

But the definition is broader than that. A resolution also counts when the customer confirms they are done or simply stops replying without asking for more help.

Stops replying. Think about how often people stop replying to a chat. They get distracted. Their kid walks in. They find the answer somewhere else out of frustration. They close the tab because the bot was unhelpful and they gave up.

Under that definition, a conversation where your AI failed so thoroughly that the customer abandoned it can still show up on your invoice as a success. I am not saying that is deliberate. I am saying it is a measurement problem that happens to point in the vendor's favour, and you should know about it before you turn the feature on.

The Bill Shock Stories

Go read the pricing threads on Reddit and the reviews on Trustpilot and you will find a consistent theme.

One team reported going from around 4,000 dollars a month to 9,000. Another, on a legacy grandfathered plan, got moved to the new model and watched their bill go from 119 dollars to 854. A third projected a jump from 1,200 to 10,000 a month once they modelled their volume properly.

The quote that stuck with me, from a Reddit thread, was roughly: the tech is great, but the pricing is a joke.

And I think that is a fair summary. Fin is genuinely good software. The AI is legitimately one of the better ones. The problem is not quality. The problem is that the meter runs faster the better it works, so the reward for successfully automating your support is a bigger invoice.

That is a structurally odd incentive. You are paying more precisely when you need humans less.

If You Do Decide To Leave, Start The Export Early

This is the practical warning, and it applies whether you leave next month or in two years.

Getting your conversation history out of Intercom is more annoying than it should be. The data export in the interface gives you CSVs containing metadata. Not message content. The actual text of your conversations requires going through the Conversations API, listing conversations and then fetching each one individually, or setting up an S3 export. There is a manual per conversation export too, which is fine for five conversations and absurd for five thousand.

The CSV export also only covers the past two years.

None of this is malicious, but it does mean "I will export it when I leave" is a worse plan than it sounds. If your support history has any value to you at all, whether for training an AI later or just for looking things up, pull it now while you are calm and have time, not during a migration week.

Zendesk is worse on this front, incidentally. Their exports have to be enabled by contacting support first, and help centre articles have no export interface at all. So this is an industry habit, not one company being difficult.

Who Should Absolutely Stay

I am not going to tell everyone to leave, because for a lot of companies Fin is the right call.

Stay if you have a support team of ten or more people who need real queue management, shift scheduling, SLA reporting and workforce planning. Stay if you are already deep in Salesforce, because this acquisition will probably make that integration much better over time. Stay if your support volume is high enough that AI deflection genuinely saves you salaries, and you have done the maths and it works.

Those are real situations and the product is strong for them.

Who Should At Least Look Around

Reassess if any of these describe you.

• You are under about eight people and mostly need "somebody sees the message and replies."
• Your monthly bill has moved in a direction you did not plan for.
• You cannot predict next month's invoice within about 20 percent.
• You are paying for seats belonging to people who open the dashboard twice a week.
• Your AI resolutions cost more than the revenue from the conversations they handle.

That last one is the real test and almost nobody runs it. Take your monthly AI spend, divide by the number of conversations, and compare that to your average order value or the value of a lead. If you sell a 40 dollar product and you are paying 99 cents per AI conversation across hundreds of chats, the numbers get uncomfortable quickly.

What The Alternative Looks Like

My bias is about to be extremely visible, so brace yourself.

The other model is flat. You pay a fixed amount per month, the AI is included, the seats are unlimited, and your invoice is the same number in December as it was in June regardless of how well anything worked. Ours is 12 dollars a month for Pro and 29 for Agency, and the replies land in Telegram so there is no dashboard to staff.

Is it as powerful as Fin? No, and I am not going to claim it is. Fin has a decade of engineering, deep reporting, workforce management and an enormous integration ecosystem. If you need those things, buy those things.

But an awful lot of companies are paying enterprise prices for a chat bubble and a bot, because they started small on a cheap plan and the model changed underneath them. If that is you, the honest move is to work out what you actually use, then price that, rather than paying for a platform because you happened to sign up in 2021.

Do The Seat Audit While You Are In There

One quick job that usually pays for itself immediately, and it has nothing to do with the acquisition.

Pull the list of everyone with a paid seat. Now check when each of them last actually replied to a conversation. Not logged in, replied. Most teams find at least one or two seats belonging to somebody who left, changed roles, or only ever needed to read.

At 85 dollars a month on Advanced, two dormant seats is over two thousand dollars a year for nothing. And seats are the one part of the bill you can fix in about ten minutes without changing anything else about how you work.

Worth checking whether the lighter seat types cover some of those people too. Higher plans include a number of free limited seats, and plenty of the people you are paying full price for only need to look at conversations rather than own them.

What To Do This Month

Nothing dramatic. Three small things.

Export your conversation history while you are not in a hurry. Set a calendar reminder to actually check your invoice for the next three months, since pricing changes tend to arrive quietly. And spend twenty minutes writing down which features you genuinely use, because that list is almost always shorter than people expect and it is the only sensible basis for any decision later.

What Acquisitions Actually Change, In Order

Having watched a few of these, the changes tend to arrive in a fairly predictable sequence. Useful to know so you can spot which stage you are in rather than reacting to headlines.

Months one to six. Almost nothing visible. Public reassurance that nothing is changing. This is usually true, because the deal has not even closed and the teams are busy with integration planning rather than product work.

Months six to eighteen. The roadmap quietly reorients toward the acquirer's customers. Integrations with the parent company get built first. Features that only mattered to smaller accounts stop getting attention. Nothing is removed, it just stops improving.

Months eighteen and beyond. Pricing and packaging get restructured, usually in a way that suits larger contracts. Cheaper legacy plans get retired and existing customers are migrated onto current pricing. This is the stage where people are genuinely surprised by their invoice, because the change is presented as a simplification.

None of this is villainous. It is just what happens when a company optimises for its new owner's customer base. But if you are a five person team on an old plan, the eighteen month mark is where you want to be paying attention, not the announcement day.

Questions Worth Asking Your Account Manager

If you have a rep, and on the higher plans you will, there are a few things worth getting in writing while everyone is being reassuring.

• Is my current pricing guaranteed through my renewal date, and what happens at renewal?
• Are there plans to retire the plan I am on?
• What is the notice period for pricing changes?
• Can I get a resolution cap or a spend alert so a bad month cannot produce a surprise bill?

That last one is the most practically useful and hardly anyone asks for it. A hard cap or an alert at a threshold turns an unpredictable metered bill into something you can plan around. Vendors will often agree to alerts even when they will not agree to caps, and an alert at least means you find out in week two rather than on the invoice.

Get the answers by email rather than on a call. Not because anyone is lying, but because account managers change jobs and email survives them.

The acquisition is not a fire. It is a reason to look up from the invoice and check you are still buying the thing you meant to buy.