30초 요약
Claire Vo gives preliminary impressions of OpenAI’s personal Dot, Spaces, connected Sites and developer model announcements. Her demos include thumbnail selection and an interactive game whose inference cost was substantial.
- 1DotVo likes some early agent results but reserves a full verdict.
- 2SpaceShared artifacts support collaboration between people and agents.
- 3SitesConnected data access depends on the viewer’s permissions.
- 4ModelsHer model preference reflects her own workload.
- 5Decisions APIAn early vision demo selects thumbnail frames.
- 6Interactive demosFast inference enables interactive experiments but still has latency.
- 7CostHer game test incurred a substantial workload-specific cost.
핵심 정리
- Vo likes some early agent results but reserves a full verdict.
- Shared artifacts support collaboration between people and agents.
- Connected data access depends on the viewer’s permissions.
- Her model preference reflects her own workload.
- An early vision demo selects thumbnail frames.
- Fast inference enables interactive experiments but still has latency.
- Her game test incurred a substantial workload-specific cost.
챕터별로 다듬은 전문 읽기
A personal Dot
Early results and rough edges.
Claire Vo shares early impressions from OpenAI DevDay 2026, including a personal agent, collaborative Spaces, connected Sites and developer tools. These are her initial tests and opinions, and product availability and pricing can change.
Vo describes testing a Dot that can work through several interfaces and delegate tasks. She likes some of the results and proactive suggestions but says the relationship between Dot, chat and Codex interfaces remains confusing, so she withholds a full review.
Space and Sites
Collaboration, data and permissions.
She sees Space as a shared place for people and agents to work on documents and other artifacts. Her example is a collaborative scratchpad for tasks, notes and decisions; whether organizations adopt it remains to be seen.
Vo describes Sites as a way to share a deployed app and use connected plugins as data sources. She emphasizes that a viewer’s access depends on their own connection and permissions, and sees this as relevant to internal tools.
Models and Decisions API
Speed, cost and vision.
She discusses GPT-6.1 Sol in terms of her preferred balance of capability, speed and cost. Her model preference and the event’s quoted prices are specific to that time and workload.
Vo demonstrates early access to a Decisions API with vision, using it to select video frames for a thumbnail and classify a hot dog image. She says the API was not yet generally released when she recorded and plans a more formal comparison.
Interactive experiments
Drawing, games and cost.
She experiments with a faster, more expensive model setting in a collaborative drawing app and a small 3D world that changes in response to prompts. The examples show possibilities and some latency, not a general performance benchmark.
Vo reports spending about $97 on roughly half an hour of the interactive game experiment. The example illustrates how fast inference can create new experiences while making cost measurement important; it is not a universal price estimate.
She closes by recommending that teams explore the collaboration tools and that developers evaluate the new model and API options for their own tasks. Her review remains preliminary.
같은 채널의 다른 글
35:10EN
How I AI· 2026년 10월 28일팟캐스트Four Personal AI Agents on Nine Tasks: A User’s Test진행자 Claire VoClaire Vo compares work and family workflows while acknowledging uneven access and subjective scoring.
29:40ENHow I AI· 2026년 10월 26일팟캐스트 · 2명How Reddit Proto Connects Design, Live Data and Production Code게스트 Blair Metcalf · 진행자 Claire VoBlair Metcalf demonstrates Reddit’s internal prototyping harness and its design-system mapping for Figma and Cursor.
16:22ENHow I AI· 2026년 10월 21일팟캐스트Claire Vo: Claude Duet Is Beautiful, but It Is Not My Number One Agent진행자 Claire VoAnthropic’s phone-based personal agent has the most polished launch Claire Vo has seen: thoughtful onboarding, task lists, first-class artifacts, and Slack-style channels. Polish is not enough when every other agent is competing for the same attention.
36:33ENHow I AI· 2026년 10월 19일팟캐스트 · 2명How Brooke Lacey Uses Claude Code to Edit Short Videos게스트 Brooke Lacey · 진행자 Claire VoA creator combines local media tools, timed cut lists and human feedback.
24:05ENHow I AI· 2026년 10월 14일팟캐스트A First Look at Meta Muse Wonder on Agent and Design Tasks진행자 Claire VoClaire Vo reports strengths and weaknesses from her own small evaluation.
37:47ENHow I AI· 2026년 10월 12일팟캐스트 · 2명Designing Meta Muse: Status, Personalization and Everyday Agents게스트 Alex Cornell · 진행자 Claire VoAlex Cornell discusses consumer agent design and his own experiments with Muse.