Designing Interfaces that Support Temporal Work Across Meetings with Generative AI

Authors: Rishi Vanukuru, Payod Panda, Xinyue Chen, Ava Elizabeth Scott, Lev Tankelevitch, Sean Rintel
Venue: DIS2025
Free from the publisher

AThe study

Exploratory analysis of eight recorded weekly hybrid meetings of two research teams in a technology company over four weeks. The analysis informed a framework and three prototype interfaces.

  1. Eight recorded meetingsTwo research teams met weekly over the same four weeks, 6 hours and 45 minutes of meetings in total.
  2. Tagging temporal workResearchers tagged every moment of looking back or ahead by its form and how far in time it reached.
  3. Mapping linksEach moment was linked to other meetings, recurring topics and milestones.
  4. Framework and designExisting meeting tools were placed on a time framework, and prototypes were designed for the regions left empty.

BThe prototypes

  1. Combine meeting recordsIn a research canvas, researchers connect transcripts, recordings, slides and prompts, and GPT-4o summarises each combination.goalsprogressplansimage highlights
  2. Instant RecapsAs a meeting ends, a summary card appears that people can tick off and add notes to.
  3. Adaptive Meeting HandoffOpened from the calendar, it shows the meeting just finished and the next one, shifting emphasis with the time available.
  4. Project BrowsersDragging a handle in team chat, or selecting a calendar range, sets the span of time to summarise.

The prototypes show real model output on public meeting recordings. They were not evaluated with users or compared with other tools.

CWhat the study found

Time talk in meetingswhat the eight recorded meetings showed

  • In these eight meetings, team members looked back and ahead throughout every meeting.Attention shifted across timescales from specific meetings and events to longer periods. This happened beyond the first meeting after an internal conference, which invited it.
  • In these meetings, team members recalled the past in order to plan next steps.Recollection fed reflection, intentions and plans, and predictions were used to justify plans.
  • Each team's weekly meeting linked repeatedly to other meetings, topics and distant milestones.The authors note these webs of looking back and ahead often go unnoticed and collapse soon after the meeting.

Gaps in today's toolswhat the framework showed and the designs proposed

  • Placed on the framework, existing meeting tools covered only a narrow slice of time.Most recap tools supported looking back and ahead over a few days to weeks, leaving large regions untouched.
  • The paper proposes three interfaces for the untouched regions of time.Instant Recaps appear as a meeting ends, a Handoff sits between meetings, and Project Browsers support reviewing months to years.
  • In informal trials, GPT-4o found many of the links between meetings the analysis had found.These trials used simple prompts and were not formally evaluated for quality or accuracy.

DKey takeaway

Meeting intentionality can span a series of meetings, linked by temporal work as people look back and ahead. Generative AI may help carry goals and context between meetings, through recaps, handoffs and project browsers.

Made for this site with AI assistance from the verified research brief. It is not the paper's own figure.

Abstract

Temporal work is an essential part of the modern knowledge workplace, where multiple threads of meetings and projects are connected across time by the acts of looking back (retrospection) and ahead (prospection). As we develop Generative AI interfaces to support knowledge work, this lens of temporality can help ground design in real workplace needs. Building upon research in routine dynamics and cognitive science, and an exploratory analysis of real recurring meetings, we develop a framework and a tool for the synergistic exploration of temporal work and the capabilities of Generative AI. We then use these to design a series of interface concepts and prototypes to better support work that spans multiple scales of time. Through this approach, we demonstrate how the design of new Generative AI tools can be guided by our understanding of how work really happens across meetings and projects.

Research brief

By Rishi Vanukuru, Payod Panda, Xinyue Chen, Ava Elizabeth Scott, Lev Tankelevitch, and Sean Rintel.

Takeaway: Meeting intentionality spans a series of meetings. People's temporal work (looking back and ahead) links them, yet today's tools cover only a narrow slice of time. So generative AI should carry goals and context between meetings, for example through instant recaps, handoffs to the next meeting and project browsers.

Why this matters

What was known
Meeting support tools, including recent Generative AI features, mostly focus on making individual meetings efficient, overlooking how work connects across meetings over time. Much of current commercial and academic work, however, focuses on understanding how the emergent capabilities of Generative AI may be applied to the context of individual meetings (§1, p. 2)
What this paper adds
An analysis of real recurring meetings, a framework of objective and subjective time, a Generative AI Content Sandbox for prompting over many meetings at once, and three interface concepts with open-source prototypes. (§1, p. 2)
Who it matters to
Designers and researchers of Generative AI meeting and knowledge-work tools, and researchers in Meeting Science and CSCW. Instead of features driven largely by what Generative AI" can" do, we offer a pathway for more considered design that is grounded in the reality of temporal work. (§6.3, p. 17)

Research questions

Method and data

Design
Exploratory analysis of recorded recurring meetings, followed by framework development and a design and prototyping exercise using a purpose-built research tool (see Prototype).
Prototype
Two kinds. The Generative AI Content Sandbox is a research tool built with React Flow. It is a web canvas where researchers drag in and connect nodes for meeting transcripts, recordings, slides and prompts. The system sends each combination to GPT-4o and shows the output, which can feed further nodes (for example, to combine summaries of several meetings). A template node produces a structured summary of goals, progress, plans and image highlights, which fills three interactive web prototypes. Instant Recaps: a summary card shown as a meeting ends, which people can tick off and add notes to. Adaptive Meeting Handoff: opened from the calendar's current-time marker, it shows the meeting just finished and the next one, shifting emphasis with the time available. Project Browsers: dragging a scroll handle in a team chat, or selecting a calendar range, sets the span of time to summarise. The prototypes show real model output on public AMI corpus meetings. They were not evaluated with users or compared with other tools.
Overview in four linked parts: a framework of objective and subjective time, the Generative AI Content Sandbox, interface concepts, and prototypes, with arrows showing how each informs the next.
Figure 1: Designing Interfaces grounded in Temporal Work & Generative AI: (1) We can use the framework of temporal work to define regions in the spaces of objective and subjective time that require support for a given context. (2) The Generative AI Content Sandbox enables us to explore and understand how Generative AI might best support such forms of work. (3) Building on the framework and Sandbox, we can design interface concepts for different times and purposes. (4) These concepts can be operationalized as interactive prototypes, with real AI-generated content coming from the Sandbox. From the paper, Figure 1, p. 1.
Setting
Two research teams (A and B) in a large technology company, in consecutive weekly hybrid meetings over the same four weeks following an internal conference.
Participants / data
Eight recorded meetings, each lasting 38 minutes to an hour, 6 hours and 45 minutes in total (Table 1 gives participants per meeting).
Duration
Four weeks of meetings.
Analysis
Researchers watched each recording and tagged every instance of temporal work by form (recollecting, reflecting, intention, planning, prediction, simulation) and time horizon, then mapped links to other meetings, topics and milestones.

Findings

Questions this paper answers

How much do teams talk about the past and future in meetings?
A lot, and not only when reviewing. This temporal work (looking back and looking ahead) ran through every recurring meeting examined, reaching from the same day to months and beyond. Recalling the past usually served reflecting, setting intentions or planning. (See §3.3, p. 6; §3.3, p. 6.)
Why aren't meeting summaries enough?
Because a summary sits at one point in time, while meeting intentionality (knowing why you meet) stretches across a series of meetings. Most recap tools look back and ahead only days to weeks. The threads woven in meetings tend to collapse soon after unless something carries them forward. (See §4, p. 7; §4, p. 9.)
When is the best moment to reflect on a meeting?
We argue it is right after it ends, when memories are freshest and plans are forming (a moment design usually writes off). It is also often the most time-pressured moment. So support there must be light, such as a short card people can tick and add notes to. (See §5.2, p. 12.)
Can generative AI find the links between a team's meetings by itself?
Early signs are promising. When run with simple prompts on two teams' recorded meetings, a large language model connected many of the links between meetings that human qualitative analysis had found. We have not yet evaluated the quality or accuracy of its output. (See §6.3, p. 17.)

Contributions and implications

Limitations

›

Concepts

Temporal work
Looking back (retrospection) and ahead (prospection) to make sense of and act in the present, individually and as a team. Temporal work involves the acts of prospection (future-oriented cognition) and retrospection (past-oriented cognition). (§1, p. 2)
Objective and subjective time
Shipp and Jansen's distinction, which gives our framework its two axes: "objective" time (the actual passage of time based on the clock and calendar) and "subjective" time (experiencing the past, present, and future within the current moment) (§2.2, p. 4)
Prospection and retrospection forms
Four forms of looking ahead (intention, planning, prediction, simulation) and two of looking back (recollecting, reflecting), taken from prior taxonomies and used to tag the meetings. We therefore use Recollecting to refer to thinking about the past as it happened, and Reflecting to refer to the past being reframed in the present. (§3.2, p. 5)
Chain of intentionality
Linked interfaces that carry reflection and intentions from one meeting to the next and across projects. their true potential emerges when they are integrated to chain together, supporting temporal work across different timescales and contexts (§6.1, p. 14)

Figures

Overview in four linked parts: a framework of objective and subjective time, the Generative AI Content Sandbox, interface concepts, and prototypes, with arrows showing how each informs the next.
Figure 1: Designing Interfaces grounded in Temporal Work & Generative AI: (1) We can use the framework of temporal work to define regions in the spaces of objective and subjective time that require support for a given context. (2) The Generative AI Content Sandbox enables us to explore and understand how Generative AI might best support such forms of work. (3) Building on the framework and Sandbox, we can design interface concepts for different times and purposes. (4) These concepts can be operationalized as interactive prototypes, with real AI-generated content coming from the Sandbox. From the paper, Figure 1, p. 1.
Timelines of eight team meetings with dense colour-coded marks for intention, planning, prediction, simulation, recollecting and reflecting. Lines above and below each timeline show how far ahead or back each remark reaches.
Figure 2: Temporal Work in Recurring Meetings. The timeline visualizations for Meeting A1 and B1 span the duration of the meeting. Each instance of prospection and retrospection is marked on this timeline, with vertical lines emerging upward and downward respectively to indicate the time horizon of thought (ranging from the present day, to weeks, months, and beyond). Smaller visualizations for the other meetings in the series are included to the right. These figures showcase the extent and diversity of temporal work in team meetings. From the paper, Fig. 2, p. 6.
Two diagrams. Objective time: time since last meeting against time until next, with in-meeting supports and meeting recaps in small boxes. Subjective time: past against future, minutes to years, with large unexplored regions.
Figure 6: Spaces of the framework. (1) The space of objective time, defined by one's temporal position between meetings. Across both spaces, boxes indicate temporal regions currently occupied by meeting support tools, and lighter boxes indicate as-yet unexplored spaces. (2) The space of subjective time, formed by bending the axis about the origin. The figure shows different points relating to different scales of looking back and ahead. The colors correspond to the various forms of retrospection and prospection in Figure 2. From the paper, Fig. 6, p. 10.
A timeline with linked retrospection and prospection points, and screenshots of three interfaces attached along it: an Instant Recap card, an Adaptive Handoff in a calendar, and a Project Browser in a team chat.
Figure 11: The Chain of Intentionality. The three interfaces shown here-Instant Recaps (§5.2.1), Adaptive Meeting Handoff (§5.2.2), and Project Browsers (§5.2.3)-are examples of supports for retrospection and prospection in knowledge work, designed using our framework for exploring the intersection of objective and subjective time (§4). From the paper, Fig. 11, p. 17.

How to cite

Rishi Vanukuru, Payod Panda, Xinyue Chen, Ava Elizabeth Scott, Lev Tankelevitch, and Sean Rintel. 2025. Designing Interfaces that Support Temporal Work Across Meetings with Generative AI. DIS2025. https://doi.org/10.1145/3715336.3735833

This brief was prepared with AI assistance and published by Sean Rintel. It is not the authors' published text. Page numbers refer to this PDF.

BibTeX

@inproceedings{vanukuru2025designing,
author = {Vanukuru, Rishi and Panda, Payod and Chen, Xinyue and Scott, Ava Elizabeth and Tankelevitch, Lev and Rintel, Sean},
title = {Designing Interfaces that Support Temporal Work Across Meetings with Generative AI},
booktitle = {DIS2025},
year = {2025},
month = {July},
publisher = {ACM},
url = {https://doi.org/10.1145/3715336.3735833},
doi = {10.1145/3715336.3735833},
}