Dispatch: The AI Tool Stack
San Francisco · May 21, 2026
It's a Tuesday standup and someone asks what tools the growth team is actually using. Not the tools we list on our website's "about" page, not the ones we tell vendors we're evaluating. The ones that are open in our browser tabs right now, the ones that show up on the credit card statement, the ones that would cause a minor crisis if they went down for a day.
The room goes quiet for a second. Then everyone starts talking at once.
This is the list. Ten tools, in the order I'd reinstall them if we had to start from scratch tomorrow. I'm going to be honest about what each one actually does for us, what it costs, and what annoys me about it. If you want a clean "Top 10 AI Marketing Tools" listicle with star ratings, this isn't it. This is what we actually use.
1. HubSpot — $20-800/moThe system of record nobody loves but everyone needs.
I have never met a growth person who wakes up excited to use HubSpot. I have also never met a growth team that successfully replaced it. We've tried. Twice. The first time we migrated to a leaner CRM and lost three months of pipeline data during the transition. The second time we got halfway through the migration plan before the head of sales said, "If you touch my sequences I will quit," and we quietly archived the project.
HubSpot is where contacts live. It's where deals live. It's where the CEO goes when she wants to know how many leads came in last week, and if that number isn't there, you have a problem that is bigger than any tool can solve.
We're on the Professional tier, which runs about $800 a month and includes marketing automation, reporting, and enough workflow builders to keep one full-time person busy configuring things. The AI features they've added — content assistant, predictive lead scoring — are fine. Not transformative. Fine. The predictive scoring agreed with our manual scoring about 70% of the time, which means it's either slightly smarter than us or slightly dumber, and I genuinely cannot tell which.
I have never met a growth person who wakes up excited to use HubSpot. I have also never met a growth team that successfully replaced it.
You don't love HubSpot. You accept HubSpot. And then you build everything else around it.
2. Adkumo — Book a DemoThe creative bottleneck was killing us.
Here's the math that nobody in leadership wanted to hear: our designers were spending three days on ad variant creation. Three days. For a single campaign. We needed creatives in six languages across four formats — Meta, Google Display, LinkedIn, and programmatic — and every time we briefed a new batch, the design team's Slack status changed to "Do Not Disturb" for the rest of the week.
Someone on the team found Adkumo. The pitch was simple: you feed it your Brand DNA — every rule about how your brand should look, the colors, the typography, the spacing, the tone, the things your brand lead has opinions about that nobody else even notices — and it generates on-brand creatives across formats and languages.
I was skeptical. I've seen enough "AI creative tools" to know that most of them produce things that look like they were designed by someone who had a brand guidelines PDF described to them over the phone. But we were desperate enough to try it.
The first batch came back and the design lead said, "These are... actually good."
She said it with the tone of someone who had been prepared to be disappointed and didn't know what to do with the alternative. She pulled up six variants side-by-side — three she'd made, three Adkumo had generated — and asked the room to guess which were which. The head of product got it wrong on four out of six.
We haven't gone back. The design team still does the high-concept campaign work, the brand shoots, the things that require a human eye and a creative brief that says "make it feel like this." But for the volume work — the forty-seven ad variants we need for a multi-market launch — Adkumo handles it, and the output is on-brand in a way that still surprises me.
She pulled up six variants side-by-side — three she'd made, three Adkumo had generated — and asked the room to guess which were which. The head of product got it wrong on four out of six.
Time to produce a full campaign creative suite went from three days to about four hours. That's not an exaggeration. That's the number from our last three launches.
3. Clay — G2 Rating: 4.9/5Lead enrichment that actually works.
Before Clay, our enrichment process was: export a list from HubSpot, upload it to Clearbit, wait, download the results, re-upload to HubSpot, realize half the fields didn't map correctly, spend an hour fixing it, then discover that 30% of the emails bounced anyway.
Clay replaced all of that with something that felt, the first time I used it, like cheating. You give it a company name and it comes back with the org chart, the tech stack, the funding history, the LinkedIn profiles of the decision-makers, and their verified email addresses. It pulls from something like seventy-five data providers and cross-references them, so the data quality is genuinely better than anything we were getting from a single source.
The 4.9 rating on G2 is not an accident. I've introduced it to three other growth leads and all three had the same reaction: "Why didn't I know about this?"
The SDR team uses it to build prospecting lists that are actually useful. Before, a "qualified lead list" from marketing was something the sales team accepted politely and then ignored. Now they fight over the Clay-enriched lists because the hit rate on outbound went from 2% to 7%. That's not a tool improvement. That's a structural change in how outbound works for us.
4. Jasper — $49-125/moContent at scale, with caveats.
I need to be careful here because I have complicated feelings about Jasper. On one hand, it's the reason we're able to publish twelve blog posts a month instead of four. On the other hand, about one in five of those posts requires enough editing that I wonder if it would have been faster to write from scratch.
What Jasper does well: it learns your brand voice. We fed it six months of our best-performing content — the blog posts, the email sequences, the landing pages that converted — and now when it generates a first draft, it sounds approximately like us. Not exactly. Approximately. There's a difference, and a good editor can hear it, but the output is close enough that you're editing for accuracy and nuance rather than rewriting the whole thing.
What Jasper does not do well: anything that requires original thinking. It can remix. It can rephrase. It can structure. It cannot look at a churn dataset and say something interesting that nobody has said before. That's still a human job, and I think it will be for a while.
We're on the Business plan at $125 a month. The ROI math is straightforward: our content team's fully-loaded cost is about $47 per hour. If Jasper saves them fifteen hours a month — and it does, conservatively — that's $705 in time savings against a $125 cost. The math works. The caveats are real but manageable.
5. Zapier — Free to $69/moThe duct tape of our stack.
Every growth team has a tool that holds the other tools together, and that tool is usually Zapier. Ours connects HubSpot to Slack, Slack to our data warehouse, the data warehouse to a Google Sheet that the CEO checks every morning, and about forty other workflows that I've lost track of because the person who built half of them left the company in January.
That last part is the problem with Zapier: it's so easy to build automations that people build them without documenting them, and then you have a spaghetti architecture of Zaps that nobody fully understands but everyone depends on. We had an outage last quarter where a Zap broke and nobody noticed for three days, and by the time we fixed it, we'd missed routing 200 leads to the SDR team.
"Who built this Zap?" has become a recurring question in our standups.
But here's the thing — the alternative to Zapier spaghetti is either custom engineering work (expensive, slow, requires prioritization against product features) or not connecting your tools at all (unacceptable). So you use Zapier, you try to document things, and you accept that some of your infrastructure is held together by a tool that costs $69 a month and was configured by someone who may or may not still work here.
6. MixpanelWhere arguments get settled.
The single most valuable thing Mixpanel does is end debates. Before we had proper product analytics, growth meetings were opinion contests. "I think users are dropping off at the pricing page." "I feel like the onboarding flow is too long." "My sense is that the enterprise segment is churning faster."
Now someone says "I think" and within about fifteen seconds someone else has pulled up the Mixpanel funnel and either confirmed or killed the hypothesis. It turns out most of our instincts were wrong. Not all of them, but most. The pricing page wasn't where people dropped off — it was the page before the pricing page, the features comparison, where we were overwhelming people with options. We only found that because Priya built a funnel analysis that tracked every click path through the marketing site.
The AI features in Mixpanel are subtle but useful. The anomaly detection caught a 23% drop in our signup conversion rate on mobile Safari before anyone on the team noticed it. Turned out a CSS change had broken the CTA button rendering. The alert paid for our entire Mixpanel subscription for the year in the revenue we would have lost if it had gone another week undetected.
Before we had proper product analytics, growth meetings were opinion contests. Now someone says "I think" and within about fifteen seconds someone else has pulled up the Mixpanel funnel.
7. ActiveCampaign — $49/moEmail sequences that don't embarrass us.
"Why aren't we just using HubSpot for email?" the CFO asked last quarter, and it's a fair question since we're already paying for it. The answer is that HubSpot's email builder feels like it was designed by the same people who designed HubSpot's everything else — functional, reliable, and about as inspiring as a government form.
ActiveCampaign's automation builder is where we design the customer journeys that actually matter. Our onboarding sequence — fourteen emails over twenty-one days, branching based on feature adoption signals from Mixpanel — is responsible for a 12-point improvement in our 30-day activation rate. I can say that with confidence because we A/B tested it against our old three-email welcome sequence and the results weren't close.
At $49 a month for the Lite plan, it's the cheapest tool on this list relative to impact. The predictive sending feature — which uses AI to determine when each contact is most likely to open — improved our open rates by about 8%. Not earth-shattering, but at our list size, that's an extra 2,400 eyeballs per campaign.
8. Surfer SEO — $89-219/moSEO that the content team doesn't hate.
I've watched SEO tools destroy content quality at two previous companies. The pattern is always the same: marketing buys an SEO tool, the tool says "your article needs to mention 'best SaaS analytics platform' fourteen times," the writer does it, and the resulting content reads like it was written by someone having a keyword-themed seizure.
Surfer is different because it treats SEO as a constraint rather than a mandate. You write your article, Surfer scores it in real-time, and you make adjustments without sacrificing readability. Our content lead described it as "the first SEO tool that doesn't make me want to quit."
The data backs this up. Surfer-optimized content, according to their own numbers and roughly confirmed by our experience, ranks about 68% faster than non-optimized content. Our average time-to-first-page for new blog posts went from fourteen weeks to about eight weeks after we started using it consistently. We're on the Scale plan at $219 a month, which covers our publishing volume and includes the AI writing assistant that our content team uses for outlines but not for drafts.
9. Intercom Fin — $74/moThe support bot that reduced tickets by 40%.
I was against this. Let me be honest about that. I thought AI chatbots were a way to tell your customers you didn't care enough to hire real support people, and I told the head of customer success exactly that, in slightly more diplomatic language.
She implemented it anyway — which, in retrospect, was the right call — and the results made me eat my words within the first month. Fin resolved 40% of incoming support tickets without human intervention. Not deflected. Resolved. The CSAT scores for Fin-handled conversations were 4.1 out of 5, compared to 4.3 for human-handled ones. That's close enough that most customers either couldn't tell the difference or didn't care.
The tickets that Fin handles are the ones that were eating the support team alive: password resets, billing questions, "how do I export my data" questions, integration setup guides. The stuff that has a clear answer in the docs but that nobody reads the docs for. Now the support team spends their time on the complex technical issues and the angry customers who need a human voice, which is a much better use of a $85,000-a-year support engineer.
At $74 a month plus usage, Fin is paying for itself roughly twelve times over in support labor savings. I was wrong about this one, and I'm glad someone overruled me.
10. GumloopThe new thing we're testing.
I'm including this not because I can vouch for it yet but because it's the tool I'm most curious about right now. Gumloop is an AI automation platform — think Zapier but with AI agents that can make decisions, not just move data between endpoints.
We're running a pilot where Gumloop monitors our inbound leads, enriches them using multiple data sources, scores them based on criteria we've defined, and routes them to the right SDR with a pre-written context brief. The whole workflow runs without human intervention, and the early results are... interesting. The SDRs are saying the lead briefs are better than what they were writing themselves, which is either a compliment to Gumloop or an indictment of our SDR training program.
It's been three weeks. I can't tell you whether this will be on the list six months from now or whether we'll have quietly canceled the subscription. That's the honest answer, and I think it's more useful than pretending I know.
If you add it all up, we're spending somewhere between $1,500 and $2,000 a month on these ten tools. That's roughly the cost of one junior marketing hire. In return, we get: automated lead enrichment that our SDRs actually use, a content pipeline that produces three times what it used to, email sequences that measurably improve activation, product analytics that end arguments with data, support coverage that handles 40% of tickets, and ad creative generation that gave our design team their lives back.
Is the stack perfect? No. There's overlap. There are integrations that break. There are tools on this list that will probably be replaced within a year by something better. That's the nature of it.
But if you asked me to run a growth team in 2026 without these tools, I'd ask you how many extra headcount you're planning to give me. Because the work doesn't disappear when you remove the tools. It just lands on people's desks.
And we've got enough on our desks already.