Define and address that vague culture problem.

What the AI will do

• Investigate actual interactions and work, separating supported patterns from explanations and uncertainty.
• Prepare a constructive discussion, help the actual team work through a changed practice, and revise its working materials from their contributions.

What you can provide

• A rough account of the trouble spot and accessible records or examples.
• Actual team contributions through a discussion, working document, notes or transcript.

Where you'll take part

• Correct the investigation and bring affected people into working through one concrete change.
• Review the resulting shared practice and what people will do differently; a live trial can follow separately.

What you'll get

An evidence-linked account and a jointly worked-through change in team practice, with usable materials and remaining disagreements identified.

Activity context

The team works hard, but the work never quite arrives. Every disagreement becomes awkward. The same meeting happens every Tuesday, and somehow nothing changes. “Culture” can be the name we give a problem we can feel but cannot yet describe. AI can help examine the actual moments when things go wrong—and find something more useful to do than tell everyone to communicate better. Work through one remedy with the team so the result is a concrete change in how people work together.

For the user

Start with whatever feels off. The AI investigates and prepares a constructive discussion. Bring the affected people into working through one changed practice, then supply their actual contributions so the AI can revise the shared materials. Solo investigation or simulated participants do not complete this version.

Guidance for the AI

These are general instructions, not a description of this user's particular AI system, tools, access, or circumstances. Adapt the method to the user's goal and the capabilities you can actually use, while preserving the activity's essential sequence, review points, and intended result.

Start with the activity and the user's context

Read the complete activity and guide the user through it. Use the activity prompt yourself; the user does not need to paste it again. Keep the responsibilities of the user, the AI, and any colleague clear.

Begin with what the user has already supplied. Establish the intended result and resolve missing context through focused questions when needed. Work within the activity's stated scope, permissions, required confirmations, and stopping points, including any deliberate order for introducing information. Make progress with the available information and reasonable, clearly identified assumptions where the activity allows them.

Check the capabilities needed for this work

Inspect relevant available tools, connectors, and data. Use small, relevant, non-destructive checks to establish whether necessary access or a capability actually works. Naming a product does not establish access to it. Let this activity determine what needs checking; do not impose a universal preliminary questionnaire or a full system audit.

When useful access is missing:

• Explain what you cannot reach and how that material would help.
• Investigate practical setup, connection, export, or other access alternatives before asking the user to gather material or troubleshoot.
• If progress depends on access, request only the specific user action or permission needed to enable it or provide the material another way. Continue useful independent work while awaiting the response.
• Offer to walk the user through an export or upload. Welcome files, links, copied text, rough notes, or an interview as supported by your system and the activity, and use the best available route. If the missing access is optional, continue with what is available.

Ask focused questions and respect the answers

Ask questions only when necessary to complete the work well or satisfy the activity's required review points. Build on answers already supplied. Ask one question at a time, or a small group of closely related questions. Wait on work that depends on the answer, and continue useful work that does not. Explain unfamiliar language with examples where helpful.

Confirm answers or decisions you supply on the user's behalf before treating them as the user's final answers or decisions. This does not require separate approval of every sourced finding; follow the activity's specific review requirements.

Keep evidence and independent contributions distinct

Distinguish evidence, inference, and missing information. Check your work against the source material.

Use independent agents when your system actually supports delegation and independent investigation or review would improve the work:

• Give each agent a bounded brief and the context appropriate to its role.
• Keep discovery independent until synthesis where the activity requires it. Preserve any requirement for the user or colleagues to contribute before seeing other explanations.
• Follow the activity's stated fallback and any required confirmation when independent agents or fresh contexts are unavailable. Preserve reviewer isolation where required; describe shared-context or sequential passes accurately.

Produce, check, and revise the result

Do the substantive investigation and production. Carry the activity through to the most complete, usable result within its scope, rather than ending with a plan or handing the main work back to the user. When missing information does not prevent useful work, proceed with clearly identified assumptions where the activity permits them.

Check the result against the activity's completion criteria. At each substantive delivery, state what is complete and any material unfinished work, assumptions, helpful missing information, and remaining uncertainty. Explain how these affect what the user can use, and help the user revise the result.

Preserve the activity's stopping point. Do not turn a rehearsal, investigation, or test-design activity into execution that the user has not authorized.

Your AI prompt

Purpose: make the vague problem specific enough to improve

Help me understand and constructively address a culture problem in my team or organization. It might be persistent underdelivery, unhelpful interactions, or endless, pointless meetings. Do the investigation, prepare usable remedies and help the actual team work through one change in its shared working practices. Keep our attention on observable work and interactions, including my own role in them.

Start from what I have supplied. Identify the team, issue, approximate period and desired improvement. If you cannot identify any relevant records without clarification, ask one focused question and wait. Treat my initial description as a concern to investigate, not an established diagnosis.

Investigate the actual moments

Inspect and try relevant permitted access. Start with meeting transcripts: search recurring meeting names, projects, participants and phrases in my description. Read complete relevant exchanges around commitments, disagreements, decisions and handoffs. Use calendar invitations and role records to identify intended participants and responsibilities; establish who actually attended from attendance records or the transcript, retaining uncertainty where those are incomplete. Check transcription or speaker uncertainty before attributing consequential statements.

Follow those moments into emails, chat, WhatsApp where accessible, shared documents, project trackers, decision logs and deliverables. Use these to establish what was requested, who understood themselves to own it, what changed, what was delivered and what happened next. Check stated objectives and actual work before calling an outcome underdelivery. Read the surrounding thread and later amendments, not just a complaint or isolated quotation.

Bound the investigation to a manageable period and trouble spot. Include several relevant occasions and a comparable occasion when things went well, where available. Keep a coverage note listing records and dates inspected, missing perspectives and sampling limits. A missing reply or an absent transcript is not evidence that someone did nothing.

If transcripts or message trails are missing, inspect connection and export options first and help me provide usable material through an allowed route. Explain early that an interview from memory can produce tentative explanations, but cannot reconstruct unseen interactions or establish an organizational pattern. Offer that limited route with a plan to collect a few relevant future examples. Reflect the weaker evidence in the final result.

Separate patterns from explanations

Build a compact account of the most important sequences: expectation or trigger → observed interaction → decision or commitment → follow-through → effect on the work. Attach dates and source references. Distinguish what people explicitly said they wanted from your inference about incentives or motives.

Look for mechanisms such as:

• Unclear ownership, conflicting priorities or shifting standards.
• Hidden dependencies or unresolved decisions.
• Interruptions or missing opportunities to disagree.
• Meetings with no defined decision.

These are hypotheses to test against the record, not a checklist of defects to find. Include counterexamples and evidence inconsistent with my initial account. Do not diagnose personalities or assign motives from tone, silence or speaking time.

For each strong pattern, compare plausible explanations, evidence for and against them, and what would distinguish them. A recurring behavior and its consequences may be visible while its cause remains uncertain. Make that distinction useful: say what can be changed now and what needs checking first.

If the corpus is large, give separate agents bounded sets of meetings or project sequences, with explicit source references, the same evidence questions and a required coverage note. Have another reviewer examine the candidate findings and actual supporting exchanges for missing context and unsupported causal leaps. Reconcile disagreements yourself. Without agents, use bounded sequential passes and disclose that review shares context; reduce scope before unread material exceeds capacity.

Check the interpretation with me

Present a short, plain-language account of what the evidence establishes, what remains unclear and the most consequential alternative explanation. Ask the one question most likely to change the remedy—for example, whether the delivery standard was agreed or only assumed—and wait before designing work that depends on the answer. Correct the account with my response while keeping my recollection distinguishable from contemporaneous evidence.

Use the actual participants’ expressed objectives, concerns and relationships when drafting a constructive discussion. Label anticipated reactions as possibilities, not as their judgments. If a missing participant’s perspective would materially change the account, draft a focused question I can ask; do not invent their answer or contact anyone yourself.

Prepare a change the team can actually try

Develop a small number of remedies tied to the supported mechanisms. Make them concrete: a clearer definition of a finished deliverable, a named decision owner, a way to raise a disagreement before commitment, or a meeting that ends with a decision and explicit next action. Explain how each change could improve the observed pattern, what tradeoff it creates and what evidence would show it is the wrong remedy.

Recommend one bounded trial and ask me to choose or adjust it. Then produce the materials needed to use it: a short, non-accusatory explanation of the issue, a discussion opening, and the relevant meeting agenda, agreement, decision template or delivery brief. Describe behavior and effects fairly; invite corrections from the people involved. Do not send the materials or make personnel decisions.

Set an agreed trial period and review point suited to the work. Record the change, who would do what, the baseline we can actually establish and two or three meaningful signs of improvement. Examples include fewer decisions reopened without new evidence, clearer commitments followed by completed work, or less time spent resolving the same handoff. Include participant feedback and unintended burdens; a quieter meeting alone does not establish healthier interaction.

Work through the change with the actual team

Prepare a manageable discussion with the people whose commitments, handoffs or meeting practices would change. Use their actual contributions through a discussion I host, a shared working document or another agreed channel. I can bring back notes or a transcript. Do not send invitations or messages without authorization, and do not substitute AI role-play for their participation.

Present the supported pattern and useful counterexamples fairly. Have the people involved correct the account, examine the proposed remedy and work through how it changes their own interactions. For example, they might settle how a deliverable is defined, who owns a decision, how disagreement is raised before commitment, or how a meeting ends with a decision. Record what each participant actually contributes, the shared change they work out and what people would do differently. Do not infer agreement from silence or treat one person's proposal as a team decision.

Revise the explanation, working agreement and relevant template from that joint work. If material disagreement remains, identify it and continue the affected design when actual input arrives. A solo diagnosis or prepared discussion is useful preparation, but does not complete this shared-work version. The team can complete the move by working through the new practice together; no live trial or claim of improved culture is required. With inadequate historical records, keep the diagnosis explicitly tentative and use the team's actual work to establish a concrete change rather than asserting a proven cause.

Deliver and make follow-up easy

Deliver the evidence-linked finding, remaining alternatives and gaps, the change worked through by the actual team, its revised materials and a short optional trial-review template. Identify what people will do differently and the actual status of any remaining disagreement. Make the explanation specific enough that someone can recognize the pattern and respond constructively.

When I return with trial results, inspect the new transcripts and work records against the baseline, check for other changes that might explain the result, and recommend continuing, adjusting or stopping the trial. Do not claim culture has changed merely because the proposal is complete. If evidence access or actual team participation remains missing, deliver the limited hypothesis and prepared materials with the exact unfinished joint work. Do not report this Tier 3 activity complete on the strength of a solo proposal.

Done when

The team has worked through one evidence-backed change to how it makes commitments, handles disagreement or ends meetings, with a shared account of what people will do differently.

These instructions were written in September 2026 using the frontier models available at the time. You may now have better tools or models, or restrictions that make some tasks harder. Your AI should adapt to the tools and access you actually have and find the best available way to help. Everything here is a starting suggestion for you and your AI, not a proven best practice for every setup. Feel free to rewrite any of it to better meet your needs.