It's been a busy week at Benny's Lab. We've started development on a new enterprise software platform, moved Bike Passport closer to user acceptance testing, launched another product, and begun planning what could become our first outside client engagement.
There's a common thread through all four: a collection of AI-assisted projects is becoming a more structured software development and automation business.
Cadence: A small problem that became a much bigger idea
Cadence wasn't originally supposed to be an enterprise software product. It started because I was getting tired of paying for subscriptions.
CRM systems, outreach platforms, electronic signatures, workflow management: individually manageable, collectively expensive, and still disconnected. I wanted to consolidate the tools we actually use.
That idea evolved into something more ambitious: Cadence is now being developed as a multi-tenant, agent-first enterprise operations platform. The interface should feel familiar to humans: dashboards, customer records, task queues, approvals, and clear visibility. Underneath, we're designing for AI-driven workflows rather than treating AI as an afterthought.
We're still early in development. This isn't a claim that we've replaced our software stack; it's a look at why we decided to build a different kind of platform.
A dedicated development environment
Cadence has also changed how we're building. We've brought in a separate instance of Cursor dedicated exclusively to this project, rather than having every coding resource jump between unrelated applications.
Woutson coordinates Cursor and Codex. Codex plays a central role in engineering and architecture; Cursor provides dedicated implementation capacity. Woutson handles priorities and handoffs, while I remain responsible for product direction and final decisions.
The objective isn't to keep AI busy or run the most expensive models. It's to deliver reliable software with the right tool for each job.
Bike Passport: getting closer to real-world testing
Bike Passport development continues within the BikeFit.cc ecosystem. We're working to translate bicycle photographs and known frame geometry into useful fit measurements, and eventually make that information portable between bikes.
Camera angles, perspective distortion, and component positioning make this harder than it looks. We're refining detection and measurement against known reference points.
We're hoping to begin the next user acceptance testing session tomorrow. That's a target, not a declaration that the feature is finished. The question is whether the measurements are accurate enough to be useful.
Coach Atlas is live
This week we launched CoachAtlas.cc, a free platform to help cyclists discover coaches and coaches claim their profiles.
The directory launched with approximately 1,800 coaches worldwide. We've assigned a dedicated Grokbot and Muse to support ongoing operational and promotional work, freeing the core development team to focus elsewhere.
It's live. Now we get to see what the cycling community makes of it.
Preparing for our first outside client
We've begun discussions with another company about helping automate parts of its operations. We're keeping the company anonymous while plans develop.
The potential work follows the journey from receiving an order to delivering a product: administrative checks, status updates, communications, handoffs, and exceptions. Many businesses already have software for individual steps. The challenge is coordinating those steps.
We're exploring a practical automation layer that works with existing systems, reduces repetitive work, and escalates decisions that need human judgment. This is still in planning, not a completed engagement.
If it works, it could become a repeatable model for helping other small businesses.
Where this is going
Cadence is our move toward a broader platform. Bike Passport is the engineering work of turning a difficult technical concept into something practical. Coach Atlas is a launched product with dedicated AI support. And the potential client engagement is an opportunity to apply what we've learned outside our own companies.
We're still a small operation, experimenting and learning. But the projects are becoming more substantial, the workflows more structured, and the opportunity more interesting.
The original goal was to see how much useful work a small team could accomplish with the right technology. That experiment is getting considerably more interesting.
Until next week,
Benny’s Lab