Skip to main content
Question

Builders working with SMBs: how are you finding clients, and what are they actually asking for?

  • Nov 20, 2026 11:15 PM UTC
  • 1 reply
  • 9 views

c_rohrmann
Forum|alt.badge.img+2

Hi everyone, Charlie here. Most of my work is helping small and mid-sized businesses get more out of Airtable, and lately I’ve been thinking a lot about how this market is changing.

A couple of years ago, most SMB conversations started with “we need a better system than spreadsheets.” Now almost every one starts with AI: “Can Airtable + AI take this off our plate?”

Here’s what I’m hearing most often from SMB owners:

  • “We don’t have anyone who can build it.” They know what they want but don’t have the time or the technical person to make it happen.
  • “Our data is everywhere.” Spreadsheets, email, a CRM nobody updates, and answers living in one person’s head.
  • “We tried before and it didn’t stick.” Either a tool nobody adopted or a contractor who disappeared.
  • “Just show me where to start.” They don’t want a big project. They want one win that saves real time.

I’m curious how others here approach this:

  1. Where do your best SMB clients come from? Referrals, communities, LinkedIn, partners, content?
  2. What’s the #1 thing they ask for right now? Automation, AI, reporting, client portals, something else?
  3. What helps them say yes? A small paid starter project, a free audit, a demo with their own data?
  4. What’s changed in the last year? Are SMBs more open to AI, or more cautious?

I’ll share what’s working for us in the replies, and I’d love to learn from this group too. 🙌

1 reply

WorkflowEngineer
Forum|alt.badge.img+2

For the “one win,” I’d separate the workflow problem from the AI problem. Start with a recurring task that has a clear input, a named owner, and a measurable outcome. Check whether existing configuration or integrations can handle it before building anything custom.

If missing data or an unclear handoff is the bottleneck, fix that first. If the bottleneck is interpreting unstructured text, test that step with human review before letting it trigger downstream actions.

I’d compare total handling time, including review and corrections, with the current process. Faster generation doesn’t necessarily mean less work.

For a starter deliverable, I’d aim for one working path, somewhere to put exceptions with someone responsible for them, and a short handover. A successful demo alone isn’t enough.

-Roe