If your team has a pile of AI tools, a dashboard nobody opens, and a nagging feeling that you're paying for software the work has already outgrown, you're not alone. I hear some version of "is this helping our bottom line or just adding another layer to manage?" from leaders every week, and I've asked it myself during our own planning at StringCan.
You'll walk away knowing how one web agency went from fearing AI would replace it to shipping in a week what used to take six, why your next reporting setup might not need a login screen at all, and where you still can't take your hands off the wheel without getting burned.
This post is based on Episode 78 of Revenue Rewired | Will AI Replace the Dashboard and Web Development? Jon Coss of Wawai Weighs In.
If you'd rather listen than read, find the full episode on Apple Podcasts, YouTube, Spotify, or Amazon. It's worth your time.
I met Jon Coss through a cold LinkedIn message a couple of years ago, and I replied because his note clearly wasn't copy and paste. Jon founded Wawai Partners, a web development company that builds and supports sites for small businesses and marketing agencies, and he runs that US business from Malang, Indonesia, with a team that spans two cultures. He started as a social worker in a Waco emergency room, so he came to tech the long way around, and that shows in how he talks about it.
For a while, Jon was nervous. The word on the street was that offshore web shops would be the first thing AI replaced, and his team is known for being affordable. Instead of waiting to find out, they doubled down on using it, and he says the shift from fear to excitement happened within the last six months. My co-host Jay and I went through the same arc here, scared by all the talk of agencies disappearing overnight, until we realized the reason we get great work out of these tools is the 20-plus years of experience we feed into them.
Jon's bottom line answer is speed. A build that used to eat six weeks now wraps in one, with the same people and the same eye for design behind it. The part I didn't expect is that they're producing roughly the same number of sites as before, and the reclaimed time is going into new products for their clients rather than a bigger production quota.
Jon calls himself non-technical, and he has a great story about tearing down an old motorcycle with a buddy who's a car mechanic and then realizing he had no idea how to put it back together. Today he edits live client sites by chatting with Claude Code. He names the GitHub repo, points to a page, drops in a screenshot, asks for a text change, and reviews it on the staging site before anything goes live.
He was quick to say that not just anyone can walk in and pull this off. His developer, Ekki, architected the repo and the connections so the AI has, in Jon's words, railroad tracks to run on. I think that's the piece many leaders skip, because the demo looks like magic while the reason it works is that somebody did the unglamorous systems design first.
The same setup turned into a learning tool for him. While on vacation, Jon spent a couple of hours quizzing Claude about his own codebase, including how an AWS serverless system works and how the pieces connect. He pointed out that you'd never ask your developers to sit down and explain every little thing, but you don't feel bad asking a machine, and a well-organized system lets non-technical people learn from it directly.
This part of the episode got a little ranty, in the best way. Jon reminded me how people lost their minds when the iPhone 7 dropped the headphone jack, and today only a select few phones still have one. He thinks static dashboards that live behind a software login are next. His business data sits in Airtable, hooked up through an MCP connector, and two people on his team barely touch the app. Rico had AI build him a custom view of his own tasks, while Ekki skips the dashboard entirely, asking for his priorities each morning and checking items off by telling Claude what got done.
Wawai is taking the same idea into its own product. Their bulk email sender, Wawai Mail, has no dashboard to log into. You connect it to your AI, write and design the email right in the chat, send it, and the next day you can ask for a report on bounces, opens, and delivery for a 20,000-person send.
I immediately thought about our Q4 quarterly planning. Jay wants the headline first and the facts to fill in his questions, while I need to see exactly how we got to where we are, and our team has learned to present data to each of us differently. Every dashboard project we've done in Agency Analytics has meant building several views for different stakeholders, and group dashboards force concessions where sales sees one thing and ops sees another. A view built for each person, pulling from one shared pool of data and the same company goals, would save a lot of time spent decoding what matters.
Jon also uses a Stripe connection, set to view only because he's not brave enough to give it write access, and that turned out to be useful during a payment dispute with a client. He asked for every payment that client had ever made, and it laid out which month was missing and what had been paid since. For another client who has ADHD and dyslexia, Jon has AI rewrite anything he sends so it's easier to read.
Here's my uncomfortable story. I gave AI a data set that included our client list, and it returned a report with client names that were completely made up. When I asked where those came from, it basically shrugged and said it just put them in there. Jon responded that you still have to care, and you can't trust it blindly or get lazy with it.
We spent a fair amount of time on AI slop. My take is that slop starts at the top, because teams copy whatever their leaders tolerate. Jon brought up Zen and the Art of Motorcycle Maintenance and how slippery quality is to define, since he has a client who's a design perfectionist and another only cares that the button is bright and the message lands. He's not willing to call something bad just because AI helped build it, but he does think reputation and taste will matter more every year, and he pays a local Indonesian artist for the illustrations on his own site because that signal matters to him.
His practical advice for revenue leaders this quarter was concrete. Stop letting the AI connected to your inbox write emails you send without reading, because he recently got one with three em dashes in it and knew instantly. He also wants people to quit watching their AI work. If you spent 15 minutes prompting and the build ran for two hours, the real cost was 15 minutes of your time, and that frees you to run several agents on different tasks at once.
I'll add one warning from personal experience. I was juggling several big projects in parallel, fired off a long reply, and started arguing with the AI about why it forgot everything so fast before realizing I was in the wrong chat. Running work in parallel is a skill, and it takes a few reps before it stops tripping you up.
Q: Is AI really making web development faster, or is that hype?
A: For Wawai Partners, it's real. Jon's team turns sites around in about a week now instead of six, and they've put the saved hours into building new software rather than cranking out more sites.
Q: Do I need to be technical to edit my website through AI?
A: Jon isn't, and he makes copy changes himself, then checks them on staging before they publish. The catch is that his developer set up the repo first, giving it a clean system to work inside.
Q: What does replacing a dashboard with AI look like day to day?
A: You connect a data source like Airtable or Stripe and ask for what you need in plain language. On Jon's team, one developer never logs in anywhere and gets his to-do list from a quick conversation each morning.
Q: How do I keep AI from making things up in my reports?
A: Check its output against the source every time, especially names and numbers. I've had it invent clients out of thin air, so a human review step isn't optional.
Q: Will Google penalize content written with AI?
A: Nobody fully knows yet, and Jon worries that an expert who spends three hours writing and uses AI for a 5% polish gets lumped in with someone who pasted a one-line prompt. Your best protection is keeping your own expertise and voice in the final draft.
At StringCan Interactive, we help mid-market teams connect sales, marketing, and operations so revenue stops leaking through the cracks between them. These days, that includes figuring out where AI belongs in your reporting and workflows, and where it needs a person standing behind it.
If this conversation with Jon sparked ideas about your own dashboards or your team's AI habits, listen to Episode 78 and then reach out to us at StringCan to talk through what that could look like for your business.