Why it matters
How much upfront spec-writing an agent needs before it can execute a multi-file change is a direct lever on wasted context and retries -- this talk is squarely about the workflow discipline that determines an agent's effective token efficiency.
The tokenmaxxing angle
The talk breaks spec-driven development into three phases and names the tradeoffs of writing a spec before letting an AI build from it -- precisely the kind of prompt/context discipline that changes how many tokens and retries a coding agent burns per feature.
From the organizers
Speaker Erik Hanchett, a developer and technical content creator, presents at EqualSpace's ES550 Tech & Innovation Campus in Newark, hosted by Raphael Husbands as part of NJ Code & Coffee's regular monthly meetup, doors at 2pm and talk at 3pm.