マーケティングダッシュボードは死んだ:AIエージェントが主役になる

The Marketing Dashboard Is Dead: AI Agents Are Taking Over

AI Marketing21分
マーケティングダッシュボードは死んだ:AIエージェントが主役になる

マーケティングダッシュボードは死んだ:AIエージェントが主役になる

AI Marketing

0:0021:20

要約

AI Marketing の司会 Mark Fidelman が、エージェントファーストのメール配信ツール Nitro Send の共同創業者 George Hartley に話を聞いた回。Nitro Send は人間ではなくエージェントを主たるユーザーとして設計され、先月の操作の98%がエージェント経由だったという。CRM が単なるデータベースに近づくという見方、ライフサイクル施策の自動構築、アウトバウンド機能の開発状況、ハルシネーション対策、料金体系も語られた。

この回のポイント

  • ●Nitro Send は Claude や Codex 上で動く「ヘッドレス」なフルスタックのメール配信ツールで、エージェントの使い勝手を先に設計し、UI は後から補ったと説明している。
  • ●George によれば、先月の操作の98%はダッシュボードを見ないエージェント経由で、仮説が裏付けられたという。
  • ●ライフサイクルのメール施策(オンボーディングなど)を、エージェントが数分〜1、2時間で複数フロー構築でき、従来の数週間より大幅に短縮できると述べている。
  • ●CRM や従来型 SaaS は AI アシスタントの背後に押しやられ、クエリ可能なデータベースになるという見方を示した。Mark も高価な CRM に疑問を呈している。
  • ●アウトバウンド(営業メール)はスパム規制などリスクが異なるためクローズドベータ中で、今月末の公開を見込む。SMS と、目標ベースで自律的に最適化する V2 も予定している。
  • ●ハルシネーション対策として、主な生成は利用者側のエージェントが担い、エンリッチメントは別モデルによる検証を行う。新しいモデルでは事実の捏造をあまり見ないとも述べた。

章立て

  1. George の経歴と Nitro Send 誕生

    George は3社目のスタートアップとして Nitro Send を立ち上げた。既存のメールツールは使いにくく、エージェントとの連携も不十分だったことが動機だと語る。

  2. エージェントを主役にした設計

    元の仲間が再集結し、エージェントが使いやすいことを最優先に設計した。先月の活動の98%がエージェント経由だという。

  3. アウトバウンドの課題と提供範囲

    Mark はエージェントでの営業メールの難しさを語る。George は現在のメール種別と、ベータ中のアウトバウンドの状況を説明する。

  4. ライフサイクル施策とV2構想

    エージェントで多数のフローを一括構築できる点や、ニュースレター作成を紹介。目標ベースの自律型マーケティングが V2 の構想だと述べる。

  5. CRM は不要になるのか

    CRM はデータベース化し、エージェントの背後に回るという議論。Mark は高価な CRM への不満と、より能動的な CRM の理想を語る。

  6. エンリッチメントとハルシネーション

    自動エンリッチメントと類似オーディエンス探索を説明し、複数モデルでの検証や新しいモデルの精度について語る。

  7. 提供状況・料金・導入方法

    約2,000ユーザーで6月に開始。無料で5,000通、配信量ベースの課金、DNS のワンクリック設定を紹介した。

解説記事

AI Marketing の今回の放送では、司会の Mark Fidelman が、エージェントファースト型のメール配信ツール Nitro Send の共同創業者 George Hartley と、マーケティングのダッシュボードがエージェントに置き換わる可能性について議論した。George は、人が画面をクリックして操作する時代は急速に終わりに向かうと見ている。

人ではなくエージェントを主たるユーザーにする

Nitro Send は Claude や Codex といったエージェント上で動く「ヘッドレス」なフルスタックのメール配信ツールだ。George は、既存のメールツールは MCP(AIが外部ツールを呼び出すための接続規格)に対応していても機能が限定的で、文章作成の質も低く、結局ブラウザを行き来することになったと不満を語る。そこで、エージェントがいかに使いやすいかを最初に設計し、人間向け UI は後から補うという順序で作ったという。George によれば、先月の操作の98%はダッシュボードを使わないエージェント経由で、この仮説は裏付けられたとしている。

数週間かかる施策を数分〜数時間に

最も反響が大きいのは、顧客ライフサイクルのメール施策だという。通常、オンボーディングから継続利用・アップグレードに至る流れの設計、テスト、最適化は大仕事で、ホワイトボードに描き、翌四半期に数字を見て改善するという進め方になる。George は、エージェントが利用者の ICP(理想的な顧客像)や既存データ、過去の配信を読み、アプリ内の重要な局面ごとに多数のフローを作り、イベントに応じて停止や遷移まで設定できると説明する。エージェントが HTML で作ったものを Nitro Send 側で再構築し、受信箱で綺麗に表示されるようにする。これにより、従来数週間かかった作業が数分から1、2時間で済み、画面上でも構造を確認できるという。ニュースレターも、既存のものを指して「再デザインして」、あるいは新機能を X の投稿から文章化して配信するといった指示だけで作れるとしている。

CRM はデータベースになるのか

CRM との関係について George は、主要な CRM にはコネクタで連携し、置き換えは狙っていないと述べる。ただし個人的には、CRM を含む従来型 SaaS の多くはエージェントの背後に抽象化され、CRM は問い合わせ可能なデータベースのようになると見ている。利用者はエージェントの中で過ごす時間が増えるため、他社のダッシュボードをわざわざ見に行く理由が薄れるという考えだ。最近は Claude や Codex のデスクトップ上に編集可能なメール作成ウィジェットを置けるようになり、UI に行かず編集できるようになったことも例に挙げた。

Mark は、エージェントが顧客と話すなら CRM の更新もエージェントがやるべきで、Salesforce や HubSpot は高価すぎると主張する。さらに、LinkedIn の投稿や企業の開示資料から営業の好機を見つけて知らせるような、能動的な CRM を望むと語った。George も、目標を理解して「この人たちにフォローアップが必要、下書き済み」と提案するのが2026年の CRM だと応じている。

エンリッチメントと精度への姿勢

Nitro Send では、新規購読者のメールアドレスから複数のソースで情報を補完し、ICP に合うか、どのフローに入れるかを自動で判断する。アウトバウンドでは、優良顧客のデータから類似オーディエンスを探し、プロスペクティング用 API で見込み客を探す設計だという。ハルシネーション(事実と異なる生成)については、作成の大半は利用者側のエージェントの責任で、同社側のエンリッチメントでは別のモデルで結果を検証すると説明した。Mark はデータベースへの AI 問い合わせでの誤りを懸念したが、George は、取り込まれたデータは正しいと仮定しており、最近のモデルが事実を捏造する場面はあまり見ていないと答えた。これは本人の実感に基づく見解で、客観的な検証結果が示されたわけではない。

提供状況とロードマップ

6月に開始し、約2,000ユーザーを抱える。アウトバウンドはスパム規制などリスクが異なるため慎重を期し、現在はクローズドベータで、今月末の公開を見込む。次に SMS を予定し、数か月内に、目標を設定するとエージェントが自律的に施策を提案・最適化する V2 を目指す。無料アカウントで5,000通まで送れ、料金は連絡先数でなく配信量に応じる。DNS のワンクリック設定で自社ドメインから送信できる。

まとめ

今回の議論は、SaaS の主たる利用者が人からエージェントへ移るなら、製品設計も UI 中心から API・エージェント中心に変わるという問いを示している。98%という数字は同社の自己申告で、ユーザー層も限定的である点には注意が要る。それでも、既存ツールの連携が MCP 対応にとどまる日本の SaaS 事業者や、マーケティング担当者にとっては、自社の業務がエージェント経由で完結するかを見直すきっかけになるだろう。

文字起こし(英語・自動生成)

Hello, everyone. Welcome to AI Marketing. Joining me today is George Hartley. I think I pronounced that correctly. George, we're going to talk about how AI is replacing the traditional marketing dashboard, which is hard enough to put together. Hopefully, you know, AI can help make that easier. I know I use AI to put together partial dashboards. I'm very interested to see what you're doing. So before we jump into that, why don't you briefly introduce yourself and we'll get going. Yeah, sure. So we're building at the moment NitroSend, which is sort of a headless full-stack email marketing tool that runs in your agents at Claude and Codex. But this is my third startup, so I've been doing this for 13 years. I've built an e-commerce marketplace, which is Australia's biggest online art marketplace. and I built an email automation in e-commerce, an email automation platform for e-commerce called Smartemail. We built that and grew it and sold that in 2022.

And, yeah, this is kind of the third rodeo. We started it this year, so a newer one, but pretty exciting to be building in 2026 when kind of it feels pretty limitless with what you can ship and what you can automate. and, you know, just building an AI tool itself is, yeah, it's just, yeah, it's a wild time to be doing it. So, yeah, happy to chat. On the one hand, it's easier to build, but on the other, it's like, well, you've got to build something even more spectacular because somebody else can put together using AI as well. So let's jump into that. What made you want to create, you know, a replacement for the traditional marketing dashboard? What led you to that? We needed it. So... What was wrong with the old tools? They sucked to use from core. And I think if you're a content marketer or someone who does a lot of email, you probably felt the same kind of pain that I did when I was trying to set up, yeah, a lot of email automation earlier this year for a book brand I was doing and someone else's side as well.

I was, you know, I'm pretty claw-pilled and, well, Codex as well, But, you know, I love being able to, you know, get your SEO running from your agent, get outbound going, kind of get admin going, get like paid social, you know, ripping that with your agent. You know, it's an incredible feeling to be able to get that going really well. And email, I came to do email and it just really, really was a poor experience. Like, I tried a couple of the existing platform tools that, you know, do email marketing. And even the ones that had an MCP connection for Claudio Codex, it was, yeah, really constrained with what you could do. And the composing wasn't that good. And I was constantly in and out of the browser anyway. And it just felt such a second-class thing. I'm like, this is not, you know, this is not how it should be. So, yeah, we just thought we could do better. We've built email tools before. We're like, we can do this. So, got the band back together, really. The old CTO of Smart Mail joined, and my brother, who we did Blue Thumb together, joined.

And then the guy who was running Smart Mail after us for that fund, he joined. So, yeah, it's a bunch of old dogs kind of having another go at the same problem. But from the user is actually the agent. So the UX is the first thing we build is, you know, how does an agent use this best and assume that a human's not going to look at the UI. And then we kind of backfill for the UI as well. if you want to see it or do it in the UI whenever you can. But, you know, the hypothesis was that agents will be doing all this stuff very soon. Yeah. And it's proven right for us, you know. 98% of everything done now last month in 9% was an agent in not looking at a dashboard. So it's working for that use case. Okay. Yeah, I mean, I use an agent for emails, mainly prospecting emails. Yep. And reaching out to reporters and other things. And what I can't figure out is why they can't get that process right. I mean, Gemini doesn't even, to my knowledge, doesn't have a Gmail agent. And most of my emails are in Gmail, even if they're professional.

And I haven't even seen one from them. If somebody knows one, please send me a note on it. But, yeah, it really hasn't been a great experience for me, even though I can do it through one of my agents on OpenAI and even Claude. and the best I could do is get it into the drafts. That's about it. Then I have to approve it from the drafts. But any kind of follow-up or what I call rebumps or anything like that is a nightmare to get set up. So very curious as to see what your vision is for this and how it works. So kind of walk us through what it is that you've got set up and how soon people can get their hands on it. Yeah, the prospecting outbound piece is an interesting one. We haven't shipped that live to everyone. We have it. It's in beta. But to me, that's an obvious thing that needs improving. Exactly that problem you just described, everyone has that basically that does outbound. And I think there's an incentive if you're a big incumbent like Google or whatever, an

incentive to not really let other agents have a really good time using Gmail like that would be kind of distracting away value from them So I think that probably I assume the answer why it is a clunky thing right now So we are shipping a solution to that. That's by far the most requested thing. Overwhelmingly, our users want to do full-stack email, and full-stack includes smart prospecting as well. But we need to ship it properly because it's a different type of email, and there's other You know, there's different risks and spam laws and stuff that, you know, we just need to make sure that we do it properly. What we launch with and do is pretty much any other type of email you can imagine for a business or B2B or marketing. So transactional emails that you need to trigger if people are signing up and doing things with your app or site or SaaS or whatever. Onboarding flows. Because, you know, like I said, I've done a lot of startups and that life cycle journey of getting a first subscriber and nurturing them into a customer and then a repeat customer or a kind of, you know, an upgraded account or whatever.

Normally, that setting it up, testing, ongoing optimizing is just, it's a huge piece of work normally. Yeah. You know, you map it out on a wall or a whiteboard and then you go, you know, next quarter you kind of look at the stats and try and optimize. and, yeah, it's just everyone takes a deep breath before they even wade into there. But we just thought if you can connect a smart agent to us, you can get Astra or Opus 5.5, whatever, just to look at your ICP, look at your current data, look at whatever you sent, and just say, okay, cool, let's set up our best, you know, what are the magic moments in our app or store or whatever we're doing? Let's set up the flows for those, you know, for the right customer to hit those, you know, magic moments and go hard. Like, set up as many flows as you need, stop them when the events stop, do it all. And so, you know, what you can do now, because you can just, you can get all that in your agent, you can then push it through to NitroSend, and in NitroSend you still see the UI. So you can actually see, oh, I've set up 12 flows, and, you know, they each have, you know,

two steps to 20 steps, whatever, with splits and event triggers that will stop or move to the next one. You can see it all if you need to get your head around it, usually. But you can just one-shot it. Your agent does it. it looks like it looks on brand and as good as the agent can design. It goes all the way through to us and then we kind of, we rebuild it. You know, it comes in HTML, we rebuild it in JML so that it kind of renders really nicely on an inbox. And it's like minutes to an hour or two to get an amazing life cycle flow set up going and ready to be optimized versus weeks. And so that is the most magical thing I can think of for an email marketer who wants the results but you just don't want to waste your life in there. So yeah, we shipped that. That's been the most used part of the app and kind of the biggest wow. You know, when someone who's done a bit of email marketing goes like, oh, okay, they really clicked, you know. But also you can do like really great newsletters as well. You know, you can point it at what you used to do, say redesign this, make it better, you know. I've shipped a new feature, look at my X account, write me the newsletter, send it to the, you know, to make an engaged

segment or whoever. You can just do that with a prompt. So yeah, all that type of email you can do with NitroSend. And the last piece of it is exactly what I described, the prospecting, the outbound, which has like a ton of kind of promise for AI really personalising, getting really great results, finding the right ICP, enriching all that stuff. But yeah, we just want to make sure we get it right before we ship it. But really, this is kind of just the V1 that we're doing. Like what we're actually building in the background is kind of goal-based marketing, agentic marketing. So, you know, when you onboard, you say, what's your goal for this business? What's your goal for, you know, the site that you're connecting? Whatever, you know, I'm a content creator and what I want is, you know, to grow my audience or something. Or, you know, I have a SaaS and we want to convert people. You know, you get the goal and then you let the agent rip. And, yeah, it says, well, we think you should set up this, this, this and this. These are the segments that look right. Connect more data. We'll confirm. and like an actual autonomous agent,

it will do it proactively for you and optimize for you and, you know, hit that goal. So that is actually the V2 that we're working on and building. We just need to ship kind of what I would say the base level of full stack, which is at the outbound piece. And then, yeah, we get into the agentic V2, which is pretty exciting too. Yeah, sounds like it. Now, a lot of workflows are being done in CRMs now. Are you planning on replacing CRMs or tying in them? We connect. So, yeah, we have connectors to, like, the popular CRMs to pull in the data. And, like, we, yeah, we're not really playing it. Like, we're not trying to replace the CRM. We sort of have that. Like, you have, you know, enriched contacts and segments and the data flows in and you can see it all. So, you can do, the lifecycle stuff kind of is very similar, I would say. but in my mind CRMs traditional SaaS marketing tools most SaaS is getting abstracted away and pushed down I I just think you just spend go to spend more time in an agent because it smarter better Why leave and look at someone else silly dashboard

So I kind of think CRM just becomes like a database you can query. You won't be looking at much. I don't know. I assume people like Adio and stuff, people mostly I think use from their agents. So, you know, and we have the same thing for what we're building. You know, I know that like I can see from our UI, people aren't in there much. and that's the plan. The hardest part for us is, okay, if you have a lot of complexity with your customer nurture journey or whatever, it is kind of hard to show all that back to someone in an agent right now, and that's why there is a UI to go and have a look at. But I still think that'll be sold pretty quick. Like, even just a few weeks ago, they shipped the ability for us to put an editable kind of composer widget in Claw Desktop and Codex desktop. And so that's awesome. So now, you know, you draft your next email, it actually appears in your agent on your desktop and you can edit it from there. You don't have to go to our UI. And I think, you know, that stuff is just going to keep getting pumped into the agent.

So if you're a CRM, you've got to assume that, you know, that people aren't going to go there. So, yeah. Yeah, I always thought, I don't know why an agentic system couldn't just post all the relevant information and updates to a big spreadsheet, you know, Google spreadsheet. But the front end would be something that you're building. So you're not having to sort through a bunch of cells on the spreadsheet to find what you're looking for. But it's presented in such a way on a good, decent UI, whatever it is that you're looking for or trying to do. So I always wondered why somebody hasn't created that yet so you could skip the CRM entirely because they're getting so damn expensive. The good ones are. I mean, Salesforce, HubSpot, it's ridiculous. And the agentic system built from the ground up would perform better because my agents go out and talk to people. You know, the agents I have deployed go out and talk to people. I don't want to have to update the CRM. I want the agent to update the CRM. And I don't know why I'd have to use a Salesforce or HubSpot in order to do that.

What are your thoughts? Yeah, I don't know why you would either. As long as I didn't have to work in the spreadsheet, I can get all the information by query or by asking or by, you know, looking at a better front end, which could easily be produced. in the solution. Yeah, on the fly in a way that you wanted it. And it should be proactive and agentic. If the point of your CRM is to kind of nurture leads to close or something, it should be telling you, right, these people need bumps. I've drafted them. Do you want to send them? We've got a live one here. Make sure you're ready for the booking link. It should be completely proactive knowing your goal. That is what a CRM of 2026 should be. I want it more proactive. I want it looking at this person's career on LinkedIn, seeing their posts, seeing the company posts, and trying to determine via, to be rudimentary about it, a lead score, and saying, hey, this person you've been talking to for a while, look at all this new activity that's come up.

I read it says S1. I read some 8Ks. and this is the time to approach them with your solution or something like that. You know, that would be on the B2B side and then obviously on the B2C side, you'd work a little bit differently. But I'm hoping that's where the people designing these solutions are heading towards. I mean, yeah, I don't know. Who knows what their incentive is? But yeah, it's like, yeah, it's getting, and for example, building the outbound piece for us, where you collect a new subscriber, you can auto-enrich it. So you have waterfall enrichment across a few different places, and you've just got their email or signed up for your newsletter or something. You know who they are. Are they in your ICP? Do they sort of match whoever your customers actually are? What flow do they automatically want to go into because of that? That stuff is kind of baked into Nitro's end because that's the stuff we dog food we need, right? And it's the same with Outbound. You know, with Outbound, we have the data on who are your best customers right now

and who you're sending to. Do we want to find a lookalike audience as best we can prospecting based on those things? And off you go and use those prospecting APIs. So, yeah, it just seems to me that why build it with sticky tape and clawed and a bunch of APIs yourself and have to keep rebuilding and everyone has to? Why not just bake it into a product that is useful? Yeah, yeah. I mean, so I'll throw it back to you. Are we going to see some of these things in version 2? Oh, no, V1. The stuff I just described are outbound. The problem is we just need to put it in. Will I need a CRM? I use your V1. I don't think so. We don't have one. We just have a database in NitroSend. Okay. But we're not trying to tell people that you can throw away HubSpot or Salesforce if you want. We sort of, we just plug into them, but yeah, they're just kind of like a source of truth for the data if things are already piped there and then we can kind of do the rest How are you accounting for hallucinations Well it depends I guess what you mean But like we you know with composing and kind of setting up

most of that work is done from an external agent in with RMTP. So most of the heavy lifting is done by the user's agent. And so it's sort of on them, really. you know, like however you best get an accurate result for, you know, composing content and email, that's what comes through to NitroSend. And then with our enrichment and those things, we just run a couple of different models. So we just have a stack of kind of, yeah, checking its work with a separate model, et cetera. Okay. And so my question is more towards, okay, once the data is in, you have it in your database and I want to query it. Yeah. and look for anyone that has got revenues of over $10 million a year and maybe in a certain vertical and maybe in a certain state or province, depending on where you're at. How do you account for hallucinations in that kind of query, AI query? A database SQL query would just be dead on, right?

Yeah. If the data's in there, we assume it's correct, right? So it's come from your data source. so it's come from a connected CRM or, you know, a connected other app or whatever, you put it in as a contact or, you know, this is a prospect or whatever. We assume that's correct. Yeah, well, let's say it's clean. Let's say it's all clean data in your database. How do you account for, you know, some sort of hallucination? Like if I query it for something and then how do you stop it from coming up with some wild, crazy results from my query? I would say the kind of the newer models, I haven't really seen them go off track on stuff like that. Yeah, that's my sort of high-level take is, yeah, the newer models, I just don't see them making up facts and changing data back to the user. You know, if you're running, I don't know, the lowest level, some open source something, maybe they still do that but yeah I think generally you know our biggest chunk of users

are on Cord and then ChatGPT and Codex and yeah in general and they're sort of business users really so in general they're running like at least mid intelligence models I just haven't seen it, didn't get it wrong Yep, okay and then I guess you know in closing when do you think you'll be shipping do you have beta testers now? When do you think you'll be shipping the software? when can people check it out? Yeah. Well, so we're live and in-market with, you know, the full-stack email marketing tool, and so we've got 2,000 users. Yeah, we launched in June, and that's growing quickly, which is great. So you can get on there and start using that straight, you know, right now. And the outbound piece is live in beta for, like, a closed beta, so a handful of users. And internally, we're dogfooding it too. So that should be shipped, I would say, by the end of this month, so next three weeks. It should be live, which will be great.

And then, yeah, I'd say that is kind of the full stack of email. SMS is coming too. Got a bunch of requests for that, but... Yeah, SMS would be great. Yeah, yeah. And it sort of plugs in nicely to the flows and stuff, so that'll be coming next. And we are working on the full agentic V2 as well, so coming months for that one. All right. Well, we appreciate it. How do people get a hold of you? what's your website yeah let us know yeah yeah so we're nitrocent.com or just ask your agent to install Nitrocent and yeah otherwise you know I spend too much time on X my personal is GT Hartley on there and we you know we tend to put product announcements and stuff on our Nitrocent X and LinkedIn and yeah otherwise just jump in with your agent and have a poke around so you could use it without we're just using an agent in Collagen yeah yeah Yeah, absolutely. Ask Claude to install it. So we have a free account too to get going, sort of 5,000 free emails, and then after that there's tiered pricing. And we sort of...

The other thing is we don't agree with these pricing models where you pay on your contact number, you know, because there's no real cost there. So we just price on volume. So we sort of think it's a fairer way to go. You know, if you send more, your price goes up. So, yeah, we have big people who've jumped on recently who might be paying a ton on a competitor because they have a list of a million subs, but it's a fairer way to go as they scale up to sending. And then how do they link their email to your system? Well, yeah, we still have a one-click setup for your DNS, and away you go you can send from your Apex domain. So if your domain was kind of mark.com, it's the one-click, and then you can be sending hi from Mark or whatever. So you'll go from your business email. Okay, all right. Great. Thank you very much for being on the show. Thanks for the explanation. And please do check that out. You've got to link the information in the show notes. And, again, really appreciate you, George, being on the show. Yeah, thanks for having me, Matt.

Good to chat.

番組の概要欄(原文)

Sign up free for NitroSend via this link Marketing dashboards were built for humans to click through. George Hartley thinks AI agents will make most of that clicking unnecessary. In this episode of AI Marketing, Mark Fidelman talks with George Hartley, co-founder of Nitro Send, about the shift from traditional marketing software to agent-first systems that work directly through Claude, Codex and other AI assistants. George explains why his team designed Nitro Send with the agent as the primary user and the human interface as the backup. The result: 98% of the activity on the platform last month was completed through AI agents rather than its dashboard. The conversation goes beyond email. Mark and George discuss what happens when CRMs become little more than databases, agents monitor buying signals automatically, follow-ups write themselves and marketers stop spending half their day bouncing between browser tabs. What you'll learn Why traditional marketing dashboards may become largely irrelevant How AI agents can build complete email lifecycle campaigns in minutes instead of weeks Why most CRM software could eventually become a background data layer How agents can enrich leads, identify buying signals and draft follow-ups Why outbound email requires different safeguards than newsletters and transactional email How Nitro Send handles customer data and reduces the risk of inaccurate results What goal-based marketing could look like when agents optimize campaigns autonomously Why George believes much of traditional SaaS will be pushed behind the AI assistant Episode highlights 00:31 George's background and the launch of Nitro Send 01:51 Why existing email marketing tools were painful to use 03:52 Building software where the AI agent is the primary user 04:16 Why 98% of Nitro Send activity now happens through agents 05:41 The problems with AI-powered prospecting and follow-up 06:58 Building complete customer lifecycle campaigns with an agent 09:47 Creating and redesigning newsletters through a single prompt 10:38 The move toward autonomous, goal-based marketing 11:43 Will AI agents replace traditional CRMs? 14:10 Why expensive CRM dashboards may become unnecessary 17:00 Lead enrichment, lookalike audiences and automated prospecting 18:57 How Nitro Send approaches AI hallucinations and data accuracy 22:01 Product availability, outbound beta and the roadmap 23:19 How to try Nitro Send 24:39 Connecting your business domain and email About George Hartley George Hartley is a repeat technology founder and co-founder of Nitro Send, an agent-first email marketing platform designed to run through tools such as Claude and Codex. He previously helped build Australia's largest online art marketplace and co-founded SmartrMail, an ecommerce email automation company that was acquired in 2022. Try Nitro Send Sign up free for NitroSend via this link or ask your AI agent to install Nitro Send. The platform offers a free account that includes up to 5,000 emails. George is also on X at @GTHartley. Get AgentShock AgentShock: Surviving and Profiting from the Rise of AI Agents and Robots, by Jas Dhillon and Mark Fidelman, explains what happens as AI moves from answering questions to completing work, directing workflows and running entire business functions. Learn more and get the book at Agentized.com/book.

出典・この記事について

番組
AI Marketing(Mark Fidelman)
元の回
The Marketing Dashboard Is Dead: AI Agents Are Taking Over
配信日
要約作成日
公式サイト
www.fanaticsmedia.com

この要約・解説記事は、元の音声を文字起こしした内容をもとに AI(Claude)が日本語で作成したものです。 固有名詞や数値に誤りを含む場合があるため、正確な内容は元の音声でご確認ください。制作方法について

X でシェア

関連エピソード