Updated May 2026. This is a substantial rewrite of a post originally published in July 2024. Two of the three tools have changed entirely. The take is the same: most AI is mid, a small amount is excellent at one thing, and that's where I spend my time.
When I first wrote this post in mid-2024, the AI conversation was about whether the bubble was about to pop. The Nasdaq had just shed a trillion dollars in a week, analysts were calling top, and half my LinkedIn feed was certain ChatGPT was a fad. The other half was certain it was the printing press.
Two years later, the answer is neither — and also both. The bubble didn't pop. Capital just kept stacking. Microsoft, Meta, and BlackRock have committed nearly $200 billion combined to AI data centers. But most of the AI tools people were excited about in 2024 are gone, pivoted, or quietly worse than they used to be. The hype rotated from chatbots to agents to MCPs to agentic workflows to "AI employees" — each one a 60-day news cycle, mostly a rebrand of something that already existed.
So here's where I've landed: AI is a tool category like any other. Some of it is genuinely useful. Most of it is mid. A small amount is excellent at exactly one thing — and that's where I spend my time.
This post is an update to the original. Of the three tools I wrote about in 2024, only one is still in my regular rotation. The other two slots have changed. Here's what's actually transforming my day in 2026, and what to skip.
Quick definitional setup, because the category has become a mess.
"AI workflow tools" used to mean Zapier or Power Automate with an AI step bolted on. Today, the term covers at least four distinct things:
This post is mostly about the last two categories. I'll cover the first two in a separate piece on workflow automation tools for B2B marketers — they're a different conversation.
If you've wasted too much time this week punching prompts into MidJourney or asking ChatGPT to do your taxes, the three tools below are the ones I keep coming back to. They have one thing in common: each does one thing extremely well.
This is the most consequential change to my workflow in the last 18 months, full stop.
The HubSpot CLI lets you work on HubSpot themes, modules, and templates from your local machine. Pair it with Claude Code — Anthropic's terminal-based coding agent — and you get a development setup where you describe what you want in plain English and watch the files change.
I have built custom HubSpot modules for clients, rebuilt my entire website, and built two new apps using this stack. Things I genuinely could not have shipped myself a year ago. The workflow looks like this: open a project folder, describe the change, Claude Code edits the relevant files, runs commands, fixes its own errors, and pushes to HubSpot's design manager. I review, accept or revise, and ship.
For an under-resourced founder or a fractional CMO juggling multiple client environments, this changes the math on what's possible without hiring a dev shop. A few of the things I've shipped on this stack:
If you want a deeper walkthrough of plugging Claude into HubSpot more broadly — content drafting, contact enrichment, and so on — I covered that workflow in Funnel Vision Issue 3.
I work in Canva most of the time. It's faster, the templates are better for marketing assets, and my team can collaborate without anyone needing a $60/month Creative Cloud seat. For 90% of what I produce — social graphics, slide decks, newsletter headers — Canva is the right tool.
But there's one job Photoshop still does better than anything else: fixing a bad image.
Generative Expand, Generative Fill, and Photoshop's object-removal tools — all powered by Adobe Firefly — are in a different league from any of Canva's undercooked app library. If a client sends me a stock photo with a watermark, a product shot with a distracting bystander in the background, or an image that's the wrong aspect ratio for a hero section, Photoshop handles it cleanly.
The Adobe vs. Canva split in my workflow:
I recently launched the Funnel Vision Podcast. It was something I'd wanted to do for over a year but kept putting off — the editing alone seemed like enough work to kill the project before it started.
Then I was invited to be a guest on someone else's podcast, and the host used Riverside. The guest experience was so smooth that I signed up for a free trial the next day. After about an hour of poking around, I realized: these are the best internal AI tools I've seen in any consumer-facing product.
For each episode, Riverside handles:
What's notable is that Riverside doesn't try to do everything. It's not a CRM. It's not a content calendar. It's not "an AI agent for podcasting." It records, cleans, and exports — and it does each of those things excellently.
Half the comments on the original version of this post asked some variation of "what about agents?" — and back in 2024, the honest answer was "they don't really work yet."
That's changed. Agents work now, in narrow domains, when scoped well. I've written about a few that actually deliver:
The pattern across the agentic tools that actually work in 2026 is the same as the pattern across the productivity tools in this post: narrow scope, one job, done well. The "AI employee that does everything" pitch is still mostly vapor. The "AI that does one specific thing inside an existing workflow" pitch is delivering real value.
So what did the 2024 post still get right? The skepticism, mostly. Specifically:
The three tools above earn their place in my stack because they save me real hours, do one job extremely well, and have stayed useful over time.
Depends entirely on what you're trying to do. For development and technical work inside your marketing stack, Claude Code paired with the HubSpot CLI is the highest-leverage tool I've found. For creative production, Adobe's Firefly-powered tools are still the best for image cleanup, while Canva is the right default for everything else. For content production workflows, Riverside is the standout for audio and video. There's no single "best AI tool" — the right answer is matching a tool with a narrow purpose to a specific job you actually do.
Yes, but the role has changed. In 2024, ChatGPT was the default for almost any AI task. In 2026, it competes with Claude, Gemini, and a dozen specialized tools — and for most workflow use cases, one of those specialized tools beats it. ChatGPT is still excellent for ad-hoc reasoning, drafting, and quick research. It's less essential as a workflow integration than it used to be.
An AI workflow tool helps you do a specific task faster — generating an image, transcribing audio, drafting an email. You're still in the driver's seat. An AI agent takes actions on your behalf across multiple tools, often without step-by-step instructions. You give it a goal; it figures out the steps. Agents are powerful in narrow domains but still require careful scoping. If you're using AI to help you do your job, you want workflow tools. If you're using AI to take a job off your plate, you want an agent.
No. AI replaces specific tasks within a marketing function, not the function itself. The teams getting the most value out of AI in 2026 are the ones using it to eliminate the busywork that used to consume their best people — so those people can spend more time on strategy, customer conversations, and creative work that AI is still bad at. If you're trying to replace your marketing team with AI, you don't understand what your marketing team does.
Less than you'd think. The three tools in this post total roughly $80 to $200 a month, depending on tier — Claude API usage scales with actual work, Adobe Creative Cloud is around $60/month, Riverside ranges from free to about $40 monthly. The bigger cost is time spent evaluating and switching tools. Pick a small number of products that do one thing well, learn them deeply, and stop chasing the next launch.
I help B2B startups and growth-stage teams set up the kind of AI-powered automation and analytics workflows covered in this post. If you're trying to figure out which tools are worth your team's time — and which are noise — let's talk.
For weekly walkthroughs of specific AI workflows worth setting up, the Funnel Vision newsletter ships every Thursday.