Friends+Family DirectAI Monthly Update (March 2024)
Hey all!
Keeping you up to date with what’s new at DirectAI. To reiterate, we’re enabling anyone to build computer vision models that classify, detect, and track objects in images & videos, without training data. If you missed last month’s update, check it out here!
The Good News
We signed a new recurring revenue customer. They have a live stream product and are planning to use DirectAI to provide vision features to their streamers. We signed a Letter of Intent with a streaming company that’s hoping to use DirectAI for sports analytics, starting with identifying professional cyclists in live broadcasts. We also started free evaluation periods with 3 companies. Two of these are using our brand-new self-hosted product.
Along with releasing a self-hosted version of DirectAI, we built out a Proof-of-Concept pipeline for building tiny and fast edge-deployable models.
The Bad News
I really struggled to get this update out. In March, the customer who’d given us our first recurring revenue churned. This was a pretty demoralizing event, even if it was somewhat expected given their drop in usage over the past few months. What’s more, one of our big goals for March was getting in touch with robotics teams to validate their desire for custom-built tiny models. I didn’t prioritize setting up these meetings given pressing asks from leads further down the funnel, and we are no closer to an answer on whether this is a burning desire.
What We Learned
Be rigorous about qualifying leads upfront.
We spent a couple weeks building a self-hosted version of DirectAI for a specific client. We made handshake agreements on pricing and trial terms but failed to discuss specifics around how they would host and run our software. They ended up being unable to test our deliverable because of an operating system mismatch. The deal stalled as a result. In retrospect, we should have explained trial expectations (and ideally secured upfront payment) from the very beginning.
Learning from a customer churning
There’s a risk of this being an overdetermined analysis. We weren’t driving business value for the customer, so they stopped using our API, and they eventually requested to stop paying for it. In the future, if we see a customer consistently decrease their API usage, we need to quickly evaluate whether we’re still providing them value. If not, we need to determine whether there is a way for DirectAI to do so that fits within our roadmap.
Beware of burnout.
We spent most of the month grinding to ship features on short notice and haven't yet converted that effort into revenue. It's a challenging position to be in.
What’s Up Next
We’re focusing on a couple specific leads in the pipeline with a high willingness to pay. We’re also expanding our feature set for existing customers so we can charge them more money. Concretely, that means making our product easier to self-host and easier to run on live streams.
We’ll have more to say in the April update but we helped one of our customers run a live golf analytics event at the Masters last week!
Asks
We’ve increasingly been getting inbound from teams currently using but unsatisfied with AWS Rekognition/Google Vision. See our response in this thread (that generated a sales call this week). We’d appreciate intros to folks currently using those services!
Kindly,
Ben
P.S. Our investor is doing a promo for DirectAI and asked for a picture of the founding team. We realized we had never taken one of just the two of us!
