AI Goal Coaching: Track What Changes After the Chat

Imagine finishing an AI coaching conversation with a precise goal: update two portfolio case studies by Friday. The deadline is clear, the steps look manageable and the plan reads well. Friday’s review still has a separate job: establish which case studies have actually changed. The conversation has produced a specification that the creator can now compare with the work.

For independent creators and business owners, the useful question is what changes after the chat. A planning tool may help clarify an ambition, identify missing details or create a reason to return. Assess each contribution on its own terms, then look for evidence of progress on the work you chose.

A better goal statement is a meaningful first result

Michel Schimpf and colleagues’ CHI 2026 study, available in an open version posted on 9 February 2026, analysed 543 UK participants across five chatbot designs. Feedback added to structured guidance improved rated goal specificity and if–then plan quality; adding personalised examples to guidance did not show an additional benefit.

These were immediate goal-formulation outcomes, alongside self-report measures. The study did not track subsequent goal attainment. Its results support the tested feedback design as a way to improve planning, while leaving the later behavioural question open.

This distinction gives a sensible place to begin a personal review. Read the resulting goal as a working agreement: what is supposed to change, what would count as completion and which decisions still belong to you? A detailed paragraph is useful when it resolves those questions. Its elegance is a separate matter.

Follow-up evidence needs a useful comparison

A separate preprint by Schimpf, Julian Voigt and Thomas Bohné, posted on 18 March 2026 and not peer-reviewed, examined progress after two weeks. It randomised 517 employed adults from the UK and US; 323 passed the required checks and completed both sessions for the main analysis. Participants received an AI coaching conversation, structured written reflection or no support.

The AI group reported greater progress than the no-support group. The study did not detect a significant overall progress difference between AI and written reflection. The AI group reported greater accountability than both comparison groups. Statistical mediation linked accountability with progress, but the feeling of accountability was not independently randomised.

The outcome was self-reported progress on participants’ selected goals, rather than independently verified career achievements. The short follow-up and reduced analytic sample limit what can be concluded about lasting benefits.

For someone considering AI coaching, the comparison suggests a practical evaluation question: what does conversation add to the reflective work you could already do? Look for a contribution you can describe, such as a clarifying question you had overlooked or a review you find useful enough to revisit. Keep your own observation separate from a claim about everyone else.

Keep clarity, willingness and action visible

Return to the fictional portfolio project. A useful review can distinguish three things:

• Clarity: the two case studies are named, their required changes are defined and the deadline is recorded.
• Willingness: the creator still wants this project to take priority and can explain why.
• Action: the selected pages have actually been revised, with the agreed changes visible.

These are proposed review questions, not a validated psychological scale. Each answers a different practical concern. A clearer plan with no action may need a change in resources, timing or scope. A completed task that no longer serves the creator’s purpose calls for reconsidering the goal itself.

Suppose Friday arrives with one case study revised and the second waiting for an image the creator has yet to choose. Record those facts. “One revised; one blocked by an image decision” gives the next conversation something specific to work with. A general statement that the project feels more organised would leave that obstacle unresolved.

Avoid inventing a retrospective explanation. If the creator does not know why the image decision stalled, keep it as a question for review. An assistant’s plausible account of perfectionism, distraction or lost motivation should not become a fact about the person merely because it sounds insightful.

Make the return visit useful

Benjamin Harkin and colleagues’ 2016 meta-analysis examined 138 experimental studies of progress monitoring. Interventions encouraging monitoring improved goal attainment on average. This broader evidence concerns monitoring across varied settings; it does not establish that a particular AI coach or review routine will work for everyone.

For the portfolio example, keep a short record of the starting pages, the agreed edits and the next review date. At the review, compare the pages with that record. Then choose whether to continue, reduce the scope, obtain something missing or replace the goal. The review date is a practical choice for this example, not a scientifically established optimum.

Ask the assistant to help organise what you report. It can be given an instruction such as: “Separate completed actions from intentions and unresolved decisions. Ask about missing information. Keep possible explanations labelled as possibilities. Help me choose the next step using the time and materials I actually have.” This is a suggested instruction, not a tested intervention.

Keep room for an honest update. Reporting that a goal has become unsuitable can be useful information. A return conversation should allow a revised decision, rather than treating agreement with the earlier plan as the only acceptable outcome.

Use the coaching conversation as part of the work

The practical value of an AI coach can be assessed through what it helps you clarify, what you choose to act on and what the later review reveals. Preserve those observations in your own record. They give you a basis for deciding whether to continue the same approach, simplify it or seek a different kind of support.

If the missing step is getting started, MaryChuks’s existing guide to building an if–then plan explains how to connect a chosen action with a recognisable cue. Keep that starting plan alongside a clear description of the result you will review.

A productive goal-setting conversation leaves you able to say what you intend to do. A useful follow-up lets you say what happened, what remains uncertain and what you have decided next. Give both conversations a place in the process.

Sources

• Schimpf and colleagues, Supporting Effective Goal Setting with LLM-Based Chatbots — CHI 2026; open version posted 9 February 2026.
• Schimpf, Voigt and Bohné, AI-Assisted Goal Setting Improves Goal Progress Through Social Accountability — unreviewed preprint posted 18 March 2026.
• Harkin and colleagues, Does Monitoring Goal Progress Promote Goal Attainment? A Meta-Analysis of the Experimental Evidence — Psychological Bulletin, February 2016.