Claude Was Fast. It Didn’t Feel Fast.
A developer discusses how Claude's actual speed didn't translate to perceived speed during long conversations because the model remained silent without vocalizing its actions, creating user uncertainty. The speaker emphasizes that real-world model performance, perceived latency, response verbosity, and response format all significantly impact end-user experience in agent development.
Summary
The speaker describes a paradoxical experience with Claude where the model's actual computational speed didn't align with the user's perception of that speed. Instead of vocalizing or indicating its own actions in the typical manner associated with Claude, the model remained completely silent for 8-9 minutes during operation. This silence created a poor user experience where the developer repeatedly questioned whether the model was still working, present, or doing anything at all, despite the model actually performing efficiently in the background. The speaker reflects on this disconnect to highlight a broader principle in agent development: that multiple factors beyond raw performance speed determine the quality of user experience. These factors include the actual real-world performance metrics of the model, the user's perception of latency (which may differ from actual latency), the verbosity level of the model's responses, and the format in which those responses are presented. The speaker emphasizes that all of these elements matter critically to end-user experience and should be carefully balanced during development.
Key Insights
- Claude remained silent for 8-9 minutes without vocalizing its actions, creating uncertainty about whether the model was actually working despite being performant
- The speaker repeatedly questioned 'Are you working? Are you there? What is happening?' during the silent processing period, demonstrating how lack of feedback degrades user confidence
- Actual model speed and user-perceived latency are decoupled—fast performance doesn't guarantee users will perceive responsiveness during long conversations
- Model verbosity directly affects perceived latency; insufficient vocalization of intermediate steps or status makes long processing periods feel uncertain to users
- The balance between real-world performance, perceived latency, response verbosity, and response format are all critical and interdependent factors for end-user experience in agent development
Topics
Transcript
[0:00] So, instead of vocalizing its own actions in that very annoying Claude manner, it just remained silent for eight or nine minutes. So, although the model was fast, it didn't seem that way during long conversations because it was too quiet, and I kept asking myself, "Are you working?" Are you there? What is happening? I think you know, if any of you are also developing agents, that this balance between the real-world performance of the model, its perceived latency to the user, the verbosity of [0:30] the responses, and the format of those responses—all of these, to me, really matter for the end user experience.
Full transcript available for MurmurCast members
Sign Up to AccessMore from How I AI
Warp agents open PRs to fix the factory itself
Programming agents can autonomously improve factory systems by analyzing failed launches and proposing specific updates to agent definitions. A self-improvement loop enables observer agents to detect failures and generate evidence-based modifications that prevent recurring issues, such as changing specific steps in factory agent procedures.
Humans are still the bottleneck in Warp’s AI factory
Warp discusses how human code review has become the main bottleneck in their AI-assisted software development process, with a 3.5-hour delay from PR to first human review compared to 35 minutes from launch to PR. They're evolving their workflow to reduce human dependency by allowing requesters to review agent-generated code themselves, and plan to eventually skip review entirely for low-risk tasks by treating code review as a risk management exercise.
I Quit Claude Because It Was Annoying
The speaker explains why they stopped using Claude, citing frustrations with its tendency to produce nonsensical output and communicate in an unnatural, non-human manner. They mention considering a switch to Opus 5.5 but remain uncertain about fully migrating their work tasks.
Claude Is Not a Party Boy
A humorous character description of Claude as someone with traditional values who prioritizes work over social indulgence. The transcript portrays Claude as principled, occasionally frustrating, and willing to push back on tasks he finds objectionable.
Claude Is Back. I Still Reach for Codex.
The speaker explains their preference for using Codex over Claude, citing superior tooling, a better desktop application, and specific strengths in front-end design and SVG work. Despite Claude's return, they continue to reach for Codex for their development needs.