I first saw a Plaud at OpenSearchCon, on the table in front of Bianca Lewis. I was setting up my phone and lav mics to record the interview, and she just gestured at the little square device and said, "oh, it's fine, Plaud can do it, I'll share with you." Good tooling makes a difference for working journalists – we get to be present in the room with the person we're talking to, instead of half-buried in the mechanics of capture.

Two years on, Plaud is making an enterprise pitch with something of the same quality. "System of record" is the phrase enterprise vendors reach for when they want to mean destination – the source of truth, the place data lives, the system everyone funnels into. Plaud uses the term too. But when Brandon Lutz, Head of Marketing EMEA at Plaud, says it, the word he's emphasizing is of – as in of conversation, the bit of enterprise work that's never had a system of record at all.

A lot of enterprise software sells itself on the implicit promise that it'll become the place everything else flows into. Plaud positions itself as wanting to be the place one specific kind of input – the spoken word in meetings, calls, and rooms – finally has a structured home, and to feed whatever destination the buyer has already chosen.

Plaud sells small audio devices paired with software that captures, transcribes, and structures conversations. The company hit $100 million in annual recurring revenue (ARR) this month, two years after passing $1 million. It has around two million users globally. On 1 June it launched Plaud Team, a workspace-tier plan that lets organizations centralize accounts, billing, and devices.

The Team plan is the point at which Plaud stops being a personal productivity tool and starts being something IT has to vet, Lutz explains:

Now suddenly you have two different systems of record. Whereas with Plaud desktop we can now say it's one ecosystem, right? No matter where you have a conversation whether it's on a virtual call, in person on a phone call, one system of record backed by extremely powerful software capabilities that don't just transcribe.

Conversation – in-person, phone, virtual – has historically been split across whatever happened to be in the room. Off-the-shelf bots caught the Zoom call. A notebook (or a tool like Otter on a phone) may have caught the corridor discussion. The text thread after the meeting caught the actual decision, but nothing tied them together and you have to go hunting for each. Plaud is offering to be the single captured record across all three.

Most enterprise vendors who claim system-of-record status do it to position themselves as the central place a customer has to migrate into. Lutz spends a substantial chunk of the conversation describing the opposite – Plaud's captured material is designed, deliberately, to flow into the systems the customer already has.

How do I capture conversations and have a centralized system of record for them? How do I transform those insights into documents or artifacts that are actually meaningful to the business and are within our ways of working, and then the most important part is, how do I then set up this information to be flown or sort of flowed into systems of record that the business uses right – our CRMs, our emails, and it has to be from start to finish extremely airtight.

There are two senses of "system of record" in this case – the centralized one Plaud is creating for conversation, and the existing ones it's feeding. Lutz names Notion and customer relationship management (CRM) tools elsewhere in the conversation. The end state he describes isn't a buyer choosing Plaud instead of their existing tools, but choosing Plaud because of their existing tools – because they want a capture layer that respects where the rest of the work already lives.

That's less of a land-grab and avoids the biggest objection enterprise software typically faces, which is migration. Nobody's being asked to move their Salesforce data to Plaud. The premise is: we'll be the front end for the conversation; you decide where it ends up afterwards.

The technical element making this possible is MCP, the Model Context Protocol, an open standard Anthropic released last year that lets AI assistants talk to external systems. Plaud has built MCP support into its product, which means a captured Plaud transcript can be pulled into an MCP-compatible client like Claude and routed onward without bespoke integration work. Lutz uses it himself, and mentions his own non-technical background before describing how he used it to clean up a personal admin problem.

With Claude MCP, I was able to just go into Claude and ask, 'Hey, can you pull the audio transcripts from my personal account and migrate it into my professional?' It was doing it in the background. I didn't have to manually do this. It saved me about thirty minutes of work.

Rather than building bespoke connectors to each enterprise tool one by one, Plaud is exposing an open interface any MCP-compatible client can talk to – the layer that turns "we feed your CRM" from marketing language into a working integration, which implies a roadmap instead of a product feature.

A purely software-based capture tool – a meeting bot, a phone app – can't credibly make this same pitch. Lutz spends time discussing the hardware aspects:

Phones were designed to record or not even record necessarily, but to have phone calls. They weren't designed to capture three to five meters of a conference room, focus on individual speakers, right?

There are reliability points too. Plaud captures over Bluetooth rather than depending on WiFi, so dropped connections don't lose the conversation. Battery life on the dedicated device is uncontested ground. This is what you'd expect from a company defending its hardware investment - although it's relatable when you've been in more conference rooms where the WiFi signal being shared by several thousands of delegates turns out to be pretty thin and two hours later, you're still waiting for the transcript to be generated by your application of choice.

Lutz argues that a dedicated, visible device changes social dynamics intentionally:

It's so crucial that you ask the person across the way from you, 'Hey, I am really excited to have this conversation with you. I am just wondering, so I can be fully engaged and present. Do you mind if I record this conversation?' You want to convey signals of trust building of honesty, right?

If Plaud's job is to be the captured layer that feeds existing enterprise systems, the captured material has to be defensible. Records nobody knew were being made are a liability. Records everyone in the room agreed to are an asset. The hardware is what makes consent something a user does in the room, rather than something the customer signs in a terms-of-service box.

Lutz notes that European customers' data sits in Plaud's Frankfurt cloud infrastructure rather than the US, and that consent is one of the two most common questions Plaud gets in EMEA. The other is data security.

Plaud's vertical mix skews enterprise. Roughly 20% of users are in tech, about 12% in healthcare, and 15% in consulting, with manufacturing, education, financial, and legal filling out the rest. The demographic split is where things get more interesting.

Today most of our consumers – like 53% of our customers stay self-reported – are about director plus. So they're already sort of in a leadership position. 73% of them are between 40 to 59 years old.

This demographic suggests senior professionals already managing complex tool stacks, which is exactly the audience the interoperability argument is designed for. A director of operations who's already running Notion, Salesforce, and a working email-and-calendar setup isn't likely to want yet another destination, but will be more open to something that stops losing the content of the conversations they spend most of their week in.

The Team plan that launched on 1 June is built for that buyer specifically. UK pricing during the early-bird period (until 31 August 2026) is £16 per user per month on annual billing, rising to £20 at list price. It adds centralized billing, device management, and workspace controls. Collaboration features are scheduled for later this year – the product available today is closer to multi-device admin than a true team workspace. The press release is more generous with the word "collaboration" than the current functionality justifies.

I'd want to see more proof before I called this approach bulletproof. The CRM and email workflows Lutz describes are still in the works, and the team collaboration layer is explicitly still in development. 

However, Plaud is admitting the limits of its own claim to be the center of anything. The "system of record" phrase has been used so loosely for so long that it now means roughly "the thing that wants to be in charge." Plaud wants to own the input layer for a category nobody else owns, and then plug into whatever the customer has already built. I don't see that angle very often, and it makes the company's approach more palatable than the average AI tool that arrives with a migration plan attached.

The consent angle is the most important one. I record my own interviews and I always ask first – in person if my bot isn't online, saying "are you happy for me to record so I don't have to scribble?" A bot announcing itself by name in a Zoom corner, or pinging a "bing bong" as it joins, isn't actually asking anyone anything. The social cost of objecting to a notification is high enough that most people don't bother – which has been normalizing surveillance-by-default in enterprise meetings for years. Plaud's visible hardware doesn't fix this on its own. The device sitting on the table makes the question askable, but it doesn't ask the question. The person doing the recording still has to. The device is right there, and "I forgot" stops being a defensible answer. For video and phone calls when using Plaud, there's no "bing bong" from a bot, which makes asking for consent upfront even more important. Captured data matters more when it was actually invited rather than assumed – and the invitation is on the person, not the tool.