The Signal Layer: What to Build When Anything Can Be Built — Lena Hall, Akamai

Lena Hall resolved a production incident from a hiking trail near a waterfall. A friend of hers ran 18 agents while riding his bike. That abundance is the setup for her actual argument, which is that the same leverage reached your competitors on the same morning, so the cost of average work fell to zero and took its value with it. Ask a model what users want or what to build next and it answers from common knowledge, competently and confidently and identically to whoever asked it before you. She calls it a convergence machine, and says the single decision it cannot make is where to point it. Her uncomfortable move is refusing the easy consolation that taste will save you. Taste is preference under feedback, and preference under feedback is precisely what these systems learn. What survives is narrower: judgment about things that have not happened yet, since no data exists for them, and judgment inside a relationship the model cannot observe, because it has read everything written about your customer and never met them. Hamming said a problem is important only when you have an attack on it, and her point is that agents just handed everyone an attack on everything, so the scarce skill became choosing which problem deserves one. The rest of the talk is about signal surviving the trip to a customer, through founders who compress past legibility and org charts that round toward the mean. Speaker info: - https://x.com/lenadroid - https://www.linkedin.com/in/lena-hall Timestamps: 0:00 - Drowning in abundance 1:37 - Why everyone gets the same answer 2:58 - An expo hall where everything sounds alike 4:21 - Benchmarks with graders, and shipping without one 5:47 - The most buildable thing is rarely the most valuable 7:13 - Why taste is trainable and judgment is not 8:34 - Hamming, and having an attack on a problem 9:56 - What the convergence machine did to content 11:16 - Two ways to use it that look identical 12:38 - Source distortion, and the deleted customer pain 14:00 - Organization distortion as an investment problem 15:21 - When a narrow eval becomes a promise 16:42 - Welding the limit to the claim 18:04 - Trust as the thing with no grader