Podium Hit $150 Million in Agent ARR Selling an AI Employee as a Plan Add-On
Podium says its AI-agent ARR passed $150 million about two and a half years after it rebuilt around agents. Deployed agents went from 10,000 to 20,000 in nine months while that ARR rose by half.
October 23, 2026
Podium's $150 million in agent ARR is the most useful AI number an incumbent software company has put out this year, because you can see how Podium sold it. Podium already had more than 60,000 small businesses paying it to handle leads and customer texts. It sold them a second line on the same invoice: an AI employee that picks up the after-hours call the owner was missing anyway.
Podium announced the milestone on October 22, about two and a half years after it bet the company on agents that work like employees. More than 20,000 of those agents are now deployed across auto dealers, home services contractors, aesthetics practices and health businesses. Podium's original software, launched in 2014, took eight years to reach the same revenue. CEO Eric Rea added the same day that the company is now profitable and free-cash-flow positive.
In January, Rea posted that agent ARR had passed $100 million in under 24 months, with more than 10,000 agents live and AI revenue up 300% year over year. By September 29 he counted more than 18,000 agents working for customers. By our math, Podium added about $50 million of agent ARR in nine months, or roughly $5.5 million a month.
Podium sold the AI employee to customers it already had
Podium hasn't said how much of the $150 million came from existing customers (I'd love to see that split), so this part is built from how Podium sells rather than from a number it published. The base is on the record. Rea's January post said more than 60,000 local businesses used Podium at the end of 2023, and their biggest problem was staffing, not software. They faced 75% annual turnover, and 30% of their leads arrived after hours. An owner can price that problem without a spreadsheet.
Podium's pricing page lists no prices, sends buyers to sales for a quote and offers the AI Employee as an add-on to any plan. That makes the first sale the easiest one in SaaS, because the buyer already trusts Podium with their leads and is adding a second line to an invoice they already pay. In September, Rea said more than 60,000 businesses feed Podium's agents. Even if all 20,000 agents sat at different businesses, they'd cover about a third of that base by our math, so there's plenty of room left.
The add-on isn't Podium's only motion. Its own job board lists in-office sales development roles in Lahore and Islamabad whose job is to prospect and qualify new leads. Rea's October 19 post says Podium replaced Salesloft with Pitchr, a sales tool it built to decide who reps contact next. So alongside the expansion sale, Podium runs an outbound team aimed at businesses that aren't customers yet. It hasn't said how much of the $150 million that side brings in.
Rea is blunt about what he's selling. On a February episode of the Car Dealership Guy podcast, which Podium sponsored, he told dealers that software companies should "start focusing on labor replacement." On October 19 he went further saying Podium has replaced more than 10 tools it paid for, including Jira and Salesloft, and its own customers "are going to start doing the same math on us," so they "should expect us to help turn leads into paying customers, book more jobs, and make more money." For a contractor who can't keep a dispatcher, an agent that does the dispatcher's job is the product. When Podium shipped Jerry 2.0 in January, Car Dealership Guy reported an average revenue lift of about 30% and 45% better lead conversion for dealers. Both are Podium's own numbers.
The second leg reaches past the base, and Podium only describes it in words. In January it launched an AI operating system for aesthetics with its own electronic medical records. In August it added one for home services, a field service system whose agents book jobs, run membership programs and manage dispatch. Podium says hundreds of contractors have already left their old field service software. Selling an add-on into your base got Podium to $150 million. Pulling a contractor off the system they run their business on is a harder sale, and Podium hasn't put a revenue number on it.
Podium hasn't said how it counts an agent, or a dollar
The $150 million comes without a definition. Podium doesn't say what counts as AI-specific ARR. As RuntimeWire pointed out the day of the release, the announcement doesn't say how much is new AI spending and how much moved over from existing products. That matters more for Podium than for an AI-native startup, because its agents now ship inside the field service and medical-records platforms it sells whole. When an agent comes bundled with an operating system, counting the subscription as agent ARR is an accounting choice, and Podium hasn't told us how it made it.
The unit is fuzzy too. Rea's January post counted more than 10,000 agents in production. Car Dealership Guy's write-up a day later said Podium's agents were live in more than 10,000 businesses. Agents and businesses aren't the same count. If you divide anyway, ARR per deployed agent went from about $10,000 in January to about $7,500 in October. Treat that as rough, since both points are "more than" figures. Cheaper agents, smaller accounts, several agents per business or a looser definition of an agent could each explain the drop, and Podium hasn't said which.
Big incumbents move these definitions too. Salesforce's second-quarter fiscal 2027 results put Agentforce ARR above $1.5 billion, up more than 240%. The same release noted that from that quarter, the metric includes Slackbot and Headless 360. A growth rate quoted across a definition change compares two different things. Podium's profitability claim rests on Rea's word, and as a private company Podium gives no margin, retention or headcount figure.
Adding AI doesn't save most incumbents
The strongest argument against copying Podium is a dataset. The State of AI Report's private-company figures, which we covered in October, showed incumbents that added AI growing slower at the 75th percentile than companies with no AI in the product at all. Paul Graham, whose Y Combinator backed Podium in its winter 2016 batch, wrote on X that Podium remade itself as AI-native "by great effort." Both point the same way. Bolting an agent onto an old product doesn't bend the growth curve. Podium rebuilt the product around agents first, and one company above the line doesn't move the distribution.
What this means for SaaS GTM leaders selling to SMBs
If you run GTM at a SaaS company with a big SMB base, here's what I'd take from Podium. Sell the agent as a job done (the missed call, the after-hours lead, the dispatcher who quit), priced against what the owner pays a person rather than as a feature tier. Sell it first to customers who already trust you with that workflow. And make sure the agent can finish the job, because an add-on that can't is a price increase with extra steps.
Then report agent ARR as its own line, and publish the definition with it. Say whether bundled platform revenue counts, and whether the unit is an agent or a customer. Podium gave the number twice in nine months without either, so its growth can only be read on its own terms.
If you're the buyer, ask one question before treating a vendor's agent ARR as new spend. Is the agent its own line on my contract, or folded into the platform price? And what does my base plan cost at renewal if I drop it?
Our bet is that Podium announces $200 million in agent ARR before the end of July 2027, the date its January-to-October pace implies.
When an agent comes bundled with an operating system, counting the subscription as agent ARR is an accounting choice, and Podium hasn't told us how it made it.
Before treating a vendor's agent ARR as new spend, ask whether the agent is its own line on your contract and what your base plan costs at renewal if you drop it.
