Your Next SaaS: Camera Feed Anomaly Detection

I sent out the latest issue of Unvalidated Ideas, this morning (JST) -- here's a nice SaaS-y idea I included:

Camera feed anomaly detection (Tier: A)

More and more of the world is using more and more cameras. Homes, schools, businesses, and hobbyists use cameras to record what's happening in their lives. We won't get into the repercussions of such a trend, but what I can say is that the amount of data this is generating is enormous, and people don't want the data, they want the insight.

Generating data is all well and good, but quite obviously, generating data and generating insight are two very different things. This is why the data analyst/data engineer/data science roles exist and are growing rapidly -- just having lots of data does not make you smart, nor does it make you or your business more effective (unless you're a data repository company).

Let me get to the point(s) a bit quicker:

• More and more people are pointing cameras at things
• What people actually want to know is when something moves or when recognizable objects enter view
• People will pay for solutions that turn video feeds into notifications or actionable insight with no muss or fuss
• Integrate with Zapier and IFTTT, and you don't have to build the alerting/action runtime!

Great integrations and good UX for use cases is key. It's fairly easy to hack together a version of this idea with the heavy lifting outsourced to a cloud vision/image recognition API. The hard part is making a truly easy experience for setting it all up.

It probably won't be economical to outsource the detection (vs. building/running anomaly detection yourself), but worrying only about great UX and drop-dead easy integration will absolutely be more valuable versus building the backend yourself as well. A few points:

• If you want to sell to consumers first, you'll have to make your solution take nearly zero work to set up
  • A new user should be able to buy some random IP camera online, sign up for your service, get tailored instructions for them (or even better, download a program that configures everything for them as long as the camera is on the network/hooked up)
• If you want to sell to Enterprise first, run a business model like Snowflake, control over the infrastructure might be more important than zero work.
  • They probably won't want to push their data outside their cloud (security/cost)

Just a bit more on the consumer side -- Ring has features like motion detection, and your job will be to make something better and maybe even white-label your solution and sell that.

One of the great things about this solution is that the enterprise solution and the customer solution are extremely similar. Two very different markets, a largely similar product that has very little active interaction and UX design/churn needs.

Liked looking at this wall of text?

You might like the newsletter, check it out.