Lindy vs Copilot Studio vs Agentforce pricing: what $200 a month actually buys

Short answer: $200 a month buys 25,000 Copilot Credits from Microsoft, 40,000 Flex Credits from Salesforce, and somewhere between 18,000 and 35,000 credits from Lindy depending on how you split the seats. Those numbers are not comparable, because one Microsoft agent action costs 5 credits and one Salesforce agent action costs 20. So the same $200 buys 5,000 actions on Copilot Studio and 2,000 on Agentforce, even though the Salesforce credit is the cheapest of the three. Lindy publishes no per action rate at all. Every figure below was read from each vendor on September 5, 2026 from a United States address.

Last updated September 2026. We sell a flat rate AI agent, so we have a stake in this comparison. Check the arithmetic against the vendor pages rather than taking our word for it, and note that all three of these companies repriced during 2026.

The three rate cards, as published today

All three vendors meter agent work in credits, and all three publish a price for a credit. That makes the comparison look easy, and it is the reason most roundups get it wrong.

Read that table left to right and Salesforce wins on price per credit by a comfortable margin. Read it top to bottom and the ranking falls apart, because the row that decides your bill is the one about actions.

What $200 a month actually buys on each

Fix the budget and the comparison gets concrete. Two hundred dollars is a real number for a small team, and it happens to be exactly one Microsoft credit pack, which makes it a fair pivot.

Microsoft Copilot Studio. Two hundred dollars is one pack of 25,000 Copilot Credits, priced tenant-wide rather than per person. A standard agent action costs 5 credits, so the pack covers about 5,000 agent actions, and headcount does not multiply it. Ten people share the same pack. The catch is that the pack sits on top of Microsoft 365 licensing, and Microsoft 365 Copilot itself is $30.00 per user per month paid yearly, so $200 is rarely the whole invoice.

Salesforce Agentforce. Two hundred dollars buys 40,000 Flex Credits at the published rate of $500 per 100,000. Salesforce prices a standard agent action at 20 credits, so those 40,000 credits cover about 2,000 actions. Sixty percent more credits, and less than half the work, than the Microsoft pack. On top of that sits an Agentforce User License at $5 per user per month, which is $50 a month for a team of ten, and underneath both sits a Salesforce edition that costs considerably more than either.

Lindy. Two hundred dollars is where Lindy gets interesting, because the answer depends on how you slice it. Six Plus seats cost $179.94 and carry 18,000 pooled credits. Two Pro seats cost $199.98 and carry 30,000. One Max seat costs $199.99 and carries 35,000. Same budget, nearly twice the credits, and the difference is entirely whether you spread the money across people or concentrate it on one. That is an unusual shape, and it means the first Lindy decision is not which plan but how many people genuinely need to be in the workspace. The full ladder is on our Lindy pricing breakdown.

Why the cheapest credit buys the least work

This is the single thing worth taking away. A credit is not a unit of anything. It is a token each vendor defines however it likes, and the definitions differ by a factor of four between Microsoft and Salesforce alone.

Do the division on the thing you actually buy, which is an action. Microsoft charges 5 credits at $0.008, so an action costs $0.04. Salesforce charges 20 credits at $0.005, so an action costs $0.10. Salesforce sells the cheaper credit and the more expensive action, and it is not close: two and a half times more per action while advertising a rate 37 percent lower per credit. Anybody comparing these two platforms on the credit price alone will reach the wrong answer, confidently.

Lindy sits outside that comparison because it publishes no per action figure. What it publishes instead is bands: an everyday ask costs 2 to 250 credits, deep work costs 250 to 1,000, and a big build costs 1,000 to 2,500. On a Max seat at $0.0057 a credit, an everyday ask therefore costs somewhere between $0.011 and $1.43. The floor is four times cheaper than a Microsoft action and the ceiling is fourteen times more expensive than a Salesforce one. Both are true, and no plan page will tell you which you are going to get.

The row that matters more than any rate: what happens when you run out

All three sell credits that expire monthly and none of them roll over. What differs is the behaviour at zero, and the three vendors have chosen three opposite postures.

Lindy pauses. It stops credit-using actions until the next cycle, does not charge an overage, and even interrupts individual tasks mid-run when one starts consuming far more than usual, asking in the same Slack thread whether to continue. Microsoft disables. Cross 125 percent of prepaid tenant capacity and custom agents stop working rather than generating an invoice. Salesforce bills. Consumption is charged in arrears, so the first sign of a problem is the bill.

Which of those you prefer depends entirely on what you are running. If the agent is customer-facing, being switched off mid-month is worse than an unexpected charge, and Salesforce's posture is the safer one. If it is internal work, an invoice you did not agree to is the worse outcome. Either way it is a contract decision, not a pricing one, and it deserves more attention than the half cent difference in the rate.

Which one is actually cheaper for a ten person team

Only one of these three prices is a whole number. Lindy's $200 is the entire bill: no platform licence underneath, no seat fee on top, nothing else to buy. Microsoft's $200 pack requires Microsoft 365 licensing, and at $30.00 per user per month paid yearly for Microsoft 365 Copilot, a ten person team is looking at $300 a month before a single credit is spent. Salesforce's $200 of Flex Credits requires a $5 per user Agentforce licence, which is $50, and a Salesforce edition beneath that, which for ten users is a four figure monthly commitment on its own.

So the honest ranking for a small team with no existing platform is Lindy first by a wide margin, and the other two are not really competing for that buyer. Where Microsoft and Salesforce win is when the platform is already paid for. If your company already runs Microsoft 365, the marginal cost of agents is genuinely $200 a pack and the tenant-wide pooling makes it cheap per person. If you already run Salesforce, the same logic applies. We worked both of those through separately on Copilot Studio pricing and Agentforce pricing, and there is a working estimator on the Agentforce pricing calculator.

Where a metered agent quietly becomes the expensive option

Metered pricing is fair for work that varies. It is a bad deal for work that does not. A scheduled routine that runs every weekday morning, a report rebuilt from the same three sources every Monday, a follow up sequence that fires on the same trigger every time: these consume credits forever and they never get cheaper, because they never get easier. Three years of a Monday report is three years of credits for a job whose shape has not changed since week one.

That is the case for pricing the predictable half flat. Ours is $149 a month with no meter, no pool and no band, and the AI agent pricing page sets out exactly what that covers. It is not a replacement for what Lindy does in a Slack channel or what Agentforce does inside a Salesforce org, and we would not pretend otherwise. It is an answer to a narrower question: what should the recurring work cost, given it is the same work every month.

A note on research workloads, which distort every estimate

The line item that blows up credit forecasts more than any other is research. Lindy prices researching a competitor and writing the report at 250 to 1,000 credits, and a big build assembled from live data at up to 2,500. Those are the jobs that turn a comfortable month into a paused workspace, and they are also the jobs teams add last, after the budget is set. If a large part of what you want is pulling structured information off the public web at volume rather than reasoning about it, that is a different tool and a different cost base: a purpose-built AI web scraper handles bulk extraction far more cheaply than paying an agent platform's credit rate to browse. Keep the agent for the judgement and the writing, and it stops being the most expensive part of your stack.

How to choose without guessing

Run it in this order. First, work out whether you already pay for Microsoft 365 or Salesforce, because that answer decides most of the comparison before any rate card is involved. Second, count the work by type rather than by volume, splitting it into things that vary and things that repeat, and price only the varying half against a meter. Third, take whichever meter you are considering and multiply your worst month, not your average one, because both Microsoft and Lindy stop working when the pool is empty and an average is a poor guide to that. Fourth, ask what the platform underneath costs, since on two of these three it is the larger number.

If you want the same exercise done for a human assistant rather than a software one, the published US retainer rates are laid out on virtual assistant cost, and the head to head between an agent and a hire is on AI vs hiring a virtual assistant.