Friends+Family DirectAI Monthly Update (November 2023)
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!
Last Month’s Rundown
As mentioned last month, the customer we were helping to count carbon forestry stock officially churned at the end of November.
We agreed on usage-based pricing to do automated video editing for a team that produces customized shot analysis videos for golfers.
We’re seeing validation that customers need on-prem deployments of custom computer vision models. We agreed on pricing for a second project with the industrial camera company we mentioned last month. This will involve anomaly detection for an industrial process. We’re also moving a robotics company down the pipeline and are hoping to build them an edge-deployed segmentation model. Separately, we released a module on Viam’s registry that enables users to deploy detection/classification on their robots with ease.
We did a demo installation of DirectAI’s object tracking plugin with an Ant Media customer and built a demo for Livekit to showcase DirectAI’s ability to parse their livestream video.
What We Learned
Building Bespoke Solutions
We’ve come to terms with the fact that computer vision models, in and of themselves, are rarely valuable to customers. The solutions built on top of computer vision models that address specific needs (e.g. analytics systems, navigation instructions, assisted video editing, classification recommendations, etc.) are what people are willing to pay for. We’ve spent the past couple months building bespoke solutions to meet expressed needs but underestimated the 1) engineering effort required and 2) the challenge of re-selling those solutions in the same market. We’re using these learnings to inform how we build the next version(s) of DirectAI’s product.
On Sales and Revenue Growth
We haven't been able to maintain our revenue growth rate from the summer. We have some irons in the fire but the large enterprise leads that would have kept us "on trajectory" have failed to materialize. We don’t have a high bar for contract size AND we’re running a time-intensive sales process. We’re sampling the worst components from bottom-up and top-down sales.
On Prioritization
We realized in our monthly retro that we’re getting frustrated with the things we’re putting off. (e.g. reducing AWS spend, getting our CICD more efficient, etc.) This might actually be an example of engaging with our priorities correctly. In some ways we’re building the discipline to let small fires (internal eng tasks) burn while focusing on the infernos (moving fast on customer needs).
What’s Up Next
Small but frequent product releases to experiment and get more exposure to high variance opportunities. This will have more of a ramp up in January.
Get as close as we can to end-to-end solutions for the customers that we’ve agreed on pricing with. Namely, we’re going to finish deploying a 10x accelerated version of our object detection model.
Asks
Do you know any moderators of large Discord/Slack communities? Or Hive/AWS Rekognition/Google Vision users? We’d love to be introduced to them! I’ll pass along a forwardable blurb if so.
We’d appreciate feedback on our updates as well as frameworks we should be thinking through as we iterate towards Product-Market Fit.
Happy Holidays from the DirectAI team!
Ben