pith. sign in

arxiv: 2604.19425 · v1 · submitted 2026-04-21 · 💻 cs.HC

seneca: A Personalized Conversational Planner

Simon Bohnen , Gabriel Garbers , Lukas Ellinger , Georg Groh This is my paper

Pith reviewed 2026-05-10 01:37 UTC · model grok-4.3

classification 💻 cs.HC
keywords personalized planningconversational agentsAI-assisted toolsself-regulationgoal trackingproductivity systemshuman-computer interaction
0
0 comments X

The pith

Seneca combines a conversational agent, persistent database, and synchronizing processor to create a planner that better aligns tasks with users' actual needs.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Knowledge workers struggle with self-regulation because current tools either persist tasks without goals, offer strategies without adaptation, or enable reflection without memory. The paper introduces seneca as a framework that links these missing pieces so a user can reflect on priorities while the system remembers patterns and updates plans accordingly. If the integration works, planning would shift from reactive lists to ongoing, personalized support that closes the gap between what people say they want and what actually serves their goals. The authors describe the three-part architecture and sketch an evaluation plan that starts with simulated users before moving to real longitudinal tests of goal progress and realism.

Core claim

The paper claims that seneca, by combining a conversational agent that scaffolds reflection with clarifying questions, a persistent database that tracks goals and behavioral patterns, and a processor that synchronizes information between them, provides a personalized AI-assisted planner capable of addressing the divergence between expressed demands and underlying needs in ways that isolated tools cannot.

What carries the argument

The seneca framework, which uses a processor to keep a conversational agent and a persistent database in sync so that reflective dialogue can draw on historical patterns and update stored goals over time.

If this is right

  • Users would maintain more realistic and adaptive plans that incorporate past behavior rather than starting from scratch each session.
  • Goal tracking would become continuous, allowing the system to surface patterns that help users prioritize value-aligned tasks.
  • Reflection would gain accountability because the database retains context across conversations and prevents repeated drift from stated intentions.
  • Evaluation metrics focused on goal attainment and alignment would provide direct evidence of whether the integrated approach outperforms separate tools.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same integration pattern could be tested in adjacent domains such as health habit formation or skill acquisition where expressed goals often diverge from daily actions.
  • Future versions might add automated pattern detection in the processor to reduce reliance on the user explicitly stating every insight.
  • If the processor layer proves robust, similar hybrid designs could appear in productivity software that currently treats conversation and storage as separate features.

Load-bearing premise

The three components will work together in practice to close the gap between what users state they need and their deeper underlying needs.

What would settle it

A controlled longitudinal study in which participants using seneca show no measurable gains in goal attainment, planning realism, or goal-value alignment compared with control groups using standard to-do apps or standalone conversational interfaces.

Figures

Figures reproduced from arXiv: 2604.19425 by Gabriel Garbers, Georg Groh, Lukas Ellinger, Simon Bohnen.

Figure 1
Figure 1. Figure 1: seneca’s Core Components: The user interface combines a conversational agent with a structured work item view [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
read the original abstract

Knowledge work demands sustained self-regulation, prioritization, and reflection-yet existing planning tools only partially support these needs. Digital to-do list applications feature task persistence but lack goal representation. Paper-based planning frameworks offer effective planning strategies but cannot adapt to individual users. Conversational AI systems enable flexible reflection but lack persistence and accountability. Moreover, none of these tools address a fundamental challenge: users' expressed demands often diverge from their underlying needs. This paper introduces seneca, a conceptual framework for a personalized, AI-assisted planner that integrates the complementary strengths of these three approaches. seneca combines a conversational agent that scaffolds reflection and asks clarifying questions, a persistent database that tracks goals and behavioral patterns, and a processor that synchronizes information between them. We describe this architecture and outline a phased evaluation strategy combining automated testing with simulated users and longitudinal human studies measuring goal attainment, planning realism, and goal-value alignment.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The paper introduces seneca, a conceptual framework for a personalized AI-assisted planner that integrates a conversational agent for scaffolding reflection and asking clarifying questions, a persistent database for tracking goals and behavioral patterns, and a synchronization processor to combine the two. It positions this architecture as addressing limitations in existing tools (task persistence without goals, non-adaptive paper frameworks, and non-persistent conversational systems) and the core problem of divergence between users' expressed demands and underlying needs. The manuscript describes the high-level architecture and outlines a phased evaluation strategy using automated testing with simulated users followed by longitudinal human studies on goal attainment, planning realism, and goal-value alignment.

Significance. If the proposed integration can be implemented and empirically shown to improve alignment between expressed demands and underlying needs, the framework would offer a substantive contribution to HCI research on self-regulation tools for knowledge work. The conceptual synthesis of conversational flexibility, persistent memory, and synchronization is a clear strength, and the outlined evaluation plan provides a concrete path for future validation.

major comments (1)
  1. [phased evaluation strategy] The section outlining the phased evaluation strategy: the plan references measurement of 'goal-value alignment' and 'demand-need divergence' but provides no operational definitions, specific metrics, control conditions, or comparison baselines against existing tools. This detail is load-bearing for the paper's claim that the architecture targets a fundamental challenge not addressed by prior approaches.
minor comments (2)
  1. [architecture description] Architecture description: the synchronization processor is introduced at a high level without even a schematic data-flow diagram or pseudocode example, making it difficult to assess how conflicts between conversational inputs and stored patterns would be resolved.
  2. [introduction] Introduction: the phrase 'demand-need divergence' is used as a central motivation but is not formally defined or linked to specific prior literature on goal-setting or self-regulation in HCI.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their thoughtful review and positive assessment of the seneca framework's potential contribution. We have addressed the concern about the evaluation strategy by expanding the relevant section with the requested details.

read point-by-point responses

  1. Referee: [phased evaluation strategy] The section outlining the phased evaluation strategy: the plan references measurement of 'goal-value alignment' and 'demand-need divergence' but provides no operational definitions, specific metrics, control conditions, or comparison baselines against existing tools. This detail is load-bearing for the paper's claim that the architecture targets a fundamental challenge not addressed by prior approaches.

    Authors: We agree that the original high-level outline of the phased evaluation strategy required greater specificity to substantiate the core claims. In the revised manuscript, we have added a dedicated subsection that operationalizes the key constructs. 'Goal-value alignment' is now defined as the Pearson correlation between users' self-reported core values (elicited via an onboarding survey using a validated values inventory) and the goals selected or prioritized during planning sessions, scored on a 0-1 normalized scale. 'Demand-need divergence' is operationalized as the cosine distance between vector embeddings of user-stated demands (extracted from conversational logs) and inferred needs (derived from longitudinal behavioral patterns stored in the database). Specific metrics include these quantitative distances, supplemented by goal attainment rates (percentage of goals completed within planned timelines) and planning realism scores (expert-rated feasibility on a 5-point scale). Control conditions compare seneca against two baselines: (1) a standard persistent task manager without conversational scaffolding or value tracking, and (2) a non-persistent conversational agent without database synchronization. These baselines are drawn from representative prior HCI studies on planning tools. The additions preserve the conceptual focus of the paper while providing a concrete, replicable evaluation path. revision: yes

Circularity Check

0 steps flagged

No significant circularity; purely conceptual design proposal

full rationale

The paper introduces seneca as a high-level conceptual framework combining a conversational agent, persistent database, and synchronization processor. It describes the architecture and outlines a phased evaluation strategy but contains no equations, derivations, fitted parameters, predictions, or self-citation chains that reduce any claim to its own inputs by construction. The contribution is the proposal itself rather than any result derived from internal data or assumptions.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The proposal rests on the domain assumption that expressed demands diverge from underlying needs and that the three-component architecture can resolve this mismatch; no free parameters or invented entities are specified.

axioms (1)
  • domain assumption Users' expressed demands often diverge from their underlying needs.
    Presented in the abstract as the fundamental challenge that existing tools fail to address.

pith-pipeline@v0.9.0 · 5450 in / 1179 out tokens · 49805 ms · 2026-05-10T01:37:04.357134+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

20 extracted references · 20 canonical work pages · 1 internal anchor

  1. [1]

    doi:10.1037/0022- 3514.50.2.229

    Roger Buehler, Dale Griffin, and Michael Ross. 1994. Exploring the "planning fallacy": Why people underestimate their task completion times.Journal of Personality and Social Psychology67, 3 (Sept. 1994), 366–381. doi:10.1037/0022- 3514.67.3.366

    work page doi:10.1037/0022- 1994

  2. [2]

    Manuel Cherep, Chengtian Ma, Abigail Xu, Maya Shaked, Pattie Maes, and Nikhil Singh. 2025. A Framework for Studying AI Agent Behavior: Evidence from Consumer Choice Experiments. doi:10.48550/ARXIV.2509.25609 Version Number: 1

    work page doi:10.48550/arxiv.2509.25609 2025

  3. [3]

    Andy Clark and David Chalmers. 1998. The Extended Mind.Analysis58, 1 (1998), 7–19. http://www.jstor.org/stable/3328150

    work page arXiv 1998

  4. [4]

    Davide Consoli, Giovanni Marin, Francesco Rentocchini, and Francesco Vona

    work page

  5. [5]

    2023), 104658

    Routinization, within-occupation task changes and long-run employment dynamics.Research Policy52, 1 (Jan. 2023), 104658. doi:10.1016/j.respol.2022. 104658

    work page doi:10.1016/j.respol.2022 2023

  6. [6]

    Brown, Christopher J

    Susan Gibson, Tracy Epton, Katie Newby, Katherine E. Brown, Christopher J. Armitage, and Neil Howlett. 2026. The effects of graded tasks on physical activity: a systematic review and meta-analysis.Health Psychology Review(Feb. 2026), 1–19. doi:10.1080/17437199.2026.2618195 seneca: A Personalized Conversational Planner CHI ’26 Workshop on Tools for Thought...

    work page doi:10.1080/17437199.2026.2618195 2026

  7. [7]

    Gray, Daniel J

    Jacob S. Gray, Daniel J. Ozer, and Robert Rosenthal. 2017. Goal conflict and psychological well-being: A meta-analysis.Journal of Research in Personality66 (Feb. 2017), 27–37. doi:10.1016/j.jrp.2016.12.003

    work page doi:10.1016/j.jrp.2016.12.003 2017

  8. [9]

    Mona Haraty, Joanna McGrenere, and Charlotte Tang. 2016. How personal task management differs across individuals.International Journal of Human-Computer Studies88 (April 2016), 13–37. doi:10.1016/j.ijhcs.2015.11.006

    work page doi:10.1016/j.ijhcs.2015.11.006 2016

  9. [10]

    Job Hudig, Ad W. A. Scheepers, Michaéla C. Schippers, and Guus Smeets. 2025. Goalsetting is Mindsetting: Guided Reflection on Life Goals Taps Into the Plastic- ity of Motivational Mindsets.Psychological Reports128, 4 (Aug. 2025), 2710–2731. doi:10.1177/00332941231180813

    work page doi:10.1177/00332941231180813 2025

  10. [11]

    Markus Langer, Sara Mann, and Eva Schmidt. 2026. Why Should we Invest Epistemic Labor in a World of Generative AI? doi:10.31234/osf.io/9hbaf_v1

    work page doi:10.31234/osf.io/9hbaf_v1 2026

  11. [12]

    Liu, Kevin Lin, John Hewitt, Ashwin Paranjape, Michele Bevilacqua, Fabio Petroni, and Percy Liang

    Nelson F. Liu, Kevin Lin, John Hewitt, Ashwin Paranjape, Michele Bevilacqua, Fabio Petroni, and Percy Liang. 2024. Lost in the Middle: How Language Models Use Long Contexts.Transactions of the Association for Computational Linguistics 12 (Feb. 2024), 157–173. doi:10.1162/tacl_a_00638

    work page doi:10.1162/tacl_a_00638 2024

  12. [13]

    E. J. Masicampo and Roy F. Baumeister. 2011. Consider it done! Plan making can eliminate the cognitive effects of unfulfilled goals.Journal of Personality and Social Psychology101, 4 (2011), 667–683. doi:10.1037/a0024192

    work page doi:10.1037/a0024192 2011

  13. [14]

    2014.Essentialism: The Disciplined Pursuit of Less(1

    Greg McKeown. 2014.Essentialism: The Disciplined Pursuit of Less(1. ed ed.). Crown Business, New York

    work page 2014

  14. [15]

    Tarek Naous, Philippe Laban, Wei Xu, and Jennifer Neville. 2025. Flipping the Dialogue: Training and Evaluating User Language Models. doi:10.48550/ARXIV. 2510.06552 Version Number: 1

    work page internal anchor Pith review doi:10.48550/arxiv 2025

  15. [16]

    James Pierce. 2014. Undesigning interaction.Interactions21, 4 (July 2014), 36–39. doi:10.1145/2626373

    work page doi:10.1145/2626373 2014

  16. [17]

    Sina Rismanchian, Peter Liu, Gabe Avakian Orona, Duncan Pritchard, and Shayan Doroudi

    Evan F. Risko and Sam J. Gilbert. 2016. Cognitive Offloading.Trends in Cognitive Sciences20, 9 (Sept. 2016), 676–688. doi:10.1016/j.tics.2016.07.002

    work page doi:10.1016/j.tics.2016.07.002 2016

  17. [18]

    Alexandar Schkolski. 2025. The influence of goal setting on the personal pro- ductivity of knowledge workers: a systematic literature review.International Journal of Productivity and Performance Management74, 11 (Dec. 2025), 93–118. doi:10.1108/IJPPM-10-2024-0727

    work page doi:10.1108/ijppm-10-2024-0727 2025

  18. [19]

    Freda Shi, Xinyun Chen, Kanishka Misra, Nathan Scales, David Dohan, Ed Chi, Nathanael Schärli, and Denny Zhou. 2023. Large language models can be easily distracted by irrelevant context. InProceedings of the 40th International Conference on Machine Learning (ICML’23). JMLR.org

    work page 2023

  19. [20]

    Woods, Uwe Napiersky, and Wladislaw Rivkin

    Stephen A. Woods, Uwe Napiersky, and Wladislaw Rivkin. 2023. Learning to self-lead: Examining self-leadership strategies, personality traits and learning attainment.Applied Psychology72, 3 (July 2023), 1324–1338. doi:10.1111/apps. 12422

    work page doi:10.1111/apps 2023

  20. [21]

    Zimmerman

    Barry J. Zimmerman. 2002. Becoming a Self-Regulated Learner: An Overview. Theory Into Practice41, 2 (May 2002), 64–70. doi:10.1207/s15430421tip4102_2 Received 12 February 2026

    work page doi:10.1207/s15430421tip4102_2 2002