Case study: PlanLog
UK care providers were juggling 4 separate tools for planning, scheduling, medication and compliance.
We built one platform that replaces all four. We co-own it, so we live with every decision.
Need your ops in one place? DM me or visit https://t.co/kIuhv1Vfmg
If your team keeps ignoring a tool you paid for, ask them what they'd use instead.
Sounds obvious, but most owners just buy the next tool.
The answer is usually a spreadsheet they already trust. Build from that.
Every week my feed says AI will replace dev teams.
Cool. Make sure the AI can:
> say no to a bad feature
> notice the client meant something else
> pick up the call when it breaks
> explain the invoice
Still waiting.
Before you let an AI answer your business phone, call it yourself and test these 7 things.
I've been building Sitee Agent, our AI phone receptionist at SiteFusion, and it launches next week. Along the way I wrote down what I'd check if I were a business owner trying any AI receptionist, ours included.
1. Ask it something your website doesn't cover. A good one admits it doesn't know and takes a message.
2. Call at 11 PM and say it's urgent. Find out who on your team gets the call passed to them, and how quickly.
3. Hang up halfway through. See if you still get a summary of what the caller said.
4. Ask to book a job. It should show up as a job your team can see, without anyone retyping it from a voicemail.
5. Play back the recording and read the transcript. If you can't check what it told your customer, you can't trust it.
6. Give your staff their own logins. The person covering urgent calls doesn't need to see your sales numbers.
7. Read the questions callers asked last week. That list shows you what's missing from your website and FAQ.
None of this is exciting. It's the part that decides whether a call at 11 PM turns into a customer or a missed call you find the next morning.
If you run a small business and want to see Sitee Agent when it goes live next week, follow along here.
Saying AI made custom software easy is like saying a good camera made wedding photography easy.
The tool got better.
Knowing which shot actually matters didn't get any cheaper.
30+ projects in, my observations on what makes a build go well:
1. One person on the client side can say yes
2. Version one does 3 things, not 30
3. Real users touch it early
4. Someone owns it after launch
The tech is rarely the problem.
@KevinNaughtonJr The upside is that focus is now rare enough to be a real edge. Anyone who can sit with one hard problem for two hours is competing with very few people.
@jasonlk Same test works for agencies and dev leads. Ask who from their last team would follow them. If the answer is nobody, the references are telling you something the resume won't.
@tibo_maker The tooling is the easy part now. What separates the ones that convert is the script: a real customer problem in the first two seconds. AI makes 50 bad hooks as fast as 50 good ones.
Your pitch shouldn't read like a menu.
Websites, SEO, branding, funnels, all at once? Prospects get confused.
Pitch one clear problem, why it matters, and the outcome you deliver. Clear beats complicated.
We almost built the wrong AI receptionist.
The first idea was simple: an AI that picks up the phone and talks.
Then we asked a better question.
What does a business owner actually need after the call ends?
That changed everything about Sitee Agent.
Here's what we built instead:
1. It answers using your company's own info.
Hours, services, prices, FAQs. No made-up answers.
2. It knows when to step aside.
Urgent call? It passes it to whoever is on call. Not urgent? It takes a clear message.
3. Every call becomes something you can use.
A recording, a short summary and a full transcript. No more "who called and what did they want?"
4. Calls turn into jobs.
A caller asking for a quote shouldn't live in a voicemail. It lands as a job your team can act on.
5. The right people see the right things.
Four roles: Admin, Owner, Staff and SDR. Your sales person sees leads. Your owner sees the big picture.
6. A dashboard that tells the story.
How many calls, what people asked, what needs follow-up.
The lesson for me as a builder:
The talking part is the easy part.
The real value is what happens after you hang up.
Sitee Agent launches next week.
If missed or messy calls are costing you customers, reply "phone" and I'll show you how it works.
Your competitor's 4 person team replies to new leads in 2 minutes because one simple automation pings them.
Meanwhile your leads sit in a shared inbox until someone remembers to check it.
@kalashvasaniya Same. The part people miss is how early most businesses still are. Plenty still answer every call and email by hand, so the biggest upside is in the boring stuff.
@syedbalkhi@AwesomeMotive Saying it publicly matters. Most customers never read the privacy policy, but they remember a founder who said no to selling their data.
@aaditsh It works because a showcase answers the buyer's real question: what would I actually do with this? A feature list never answers that as well as five real examples.
@chddaniel The markup is what you pay to skip managing keys, limits and billing per model. Fine while prototyping. Once usage is steady, going direct to the provider is often worth the setup.
@bindureddy There's a lot of value left in boring digital work though. Plenty of small businesses still run on spreadsheets and missed calls. Fixing that isn't a loop, it's revenue.