Voice review

Review this plan by voice

The voice agent explains any task, answers questions grounded in this plan, and records a brain dump. Voice feedback is read-only and cannot approve or execute work. Allow the microphone when your browser asks.

The explicit Sonnet rollback keeps the existing capsule-only protocol. A lost connection retries the same Sonnet engine twice and reloads the complete persisted conversation. It never changes models during a conversation.

Sonnet 5 · no fallback
Using: browser default until voice starts
offline

Live transcript

not started

Voice answers are review evidence. Approval, sending, spending, and production changes stay in the originating chat. Voice trouble in this page? Open the standalone voice page.

0 of 36 decided

Review — August 14 idea inventory

Plan: /Users/agency/Documents/Agty/telegram coach/action-plans/2026-08-15-august-14-idea-session.md

How to review this

T-001 — Set a frictionless idea-capture default

UndecidedNeeds input
Task

Set a frictionless idea-capture default. Smallest useful result: Use the currently available voice capture once without building new infrastructure, if that direction is confirmed.

Problem solved

Debating video versus voice delays capture and distracts from the ideas.

Cost of inaction

Useful ideas may remain uncaptured while attention is spent choosing a medium.

Benefit

Faster capture with less meta-work.

Not expanded — the source explicitly leaves video versus voice open and does not establish that a separate capture system is worth building.

Open questions
  • UNKNOWN. Exact question: Should the default be direct voice-agent capture, video capture, or a rule that uses whichever is available first?
Decision

T-002 — Adopt a cost-of-inaction priority and single-focus execution rule

Undecided
Task

Adopt a cost-of-inaction priority and single-focus execution rule. Smallest useful result: Pick one highest-cost unfinished item, close everything unrelated, and finish it before switching.

Problem solved

Too many concurrent AI and Codex tasks are open while overdue work and revenue-impacting work remain unfinished.

Cost of inaction

People keep waiting, deadlines slip, and directly revenue-losing work remains delayed.

Benefit

A simple daily rule converts plans into completed high-leverage work.

Done

Each active task is ranked by who is waiting, what is lost, and direct money impact. Only the highest-cost item is active until complete or genuinely long-running. 80/20: Pick one highest-cost unfinished item, close everything unrelated, and finish it before switching.

flowchart LR
  A[Too many concurrent AI and Codex tasks are open while overdue work and] --> B[Adopt a cost-of-inaction priority and single-focus execution rule] --> C[A simple daily rule converts plans into completed high-leverage work]
flowchart TD
  S1[Ground current source and evidence] --> S2[Pick one highest-cost unfinished item close everything unrelated and]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • None material. The exact operating cadence can be refined inside T-009.
Decision

T-003 — Run a focused sales-representative lead-generation experiment

Undecided
Task

Run a focused sales-representative lead-generation experiment. Smallest useful result: One uninterrupted work block on one already-known lead source, once the target and success threshold are confirmed.

Problem solved

The sales representative needs more leads, while scattered work prevents a concentrated attempt to create them.

Cost of inaction

The sales pipeline may stay under-supplied and immediate commercial pressure continues.

Benefit

A bounded deep-work test can produce leads and test the no-switching execution rule.

Done

UNKNOWN. Exact question: Which target market, channel, offer, and minimum result should define a successful lead-generation experiment? 80/20: One uninterrupted work block on one already-known lead source, once the target and success threshold are confirmed.

flowchart LR
  A[If the goal is a qualified lead list] --> B[Build one bounded list] --> C[Usable targets exist]
flowchart LR
  A[If the goal is an outreach experiment] --> B[Prepare one focused test] --> C[Response evidence exists]
flowchart LR
  A[If the goal is booked conversations] --> B[Choose one proven route] --> C[Meetings can be counted]
Open questions
  • UNKNOWN. Exact question: Which target market, channel, offer, and minimum result should define a successful lead-generation experiment?
Decision

T-004 — Define a current sustainable financial-balance objective

UndecidedNeeds input
Task

Define a current sustainable financial-balance objective. Smallest useful result: Write one measurable current target without investigating the uncertain Gungli Laci history.

Problem solved

Commercial pressure is recognized, but the present-day balance target is not defined.

Cost of inaction

Reactive urgency can continue without a measurable outcome.

Benefit

A current target can connect prioritization to sustainable results.

Not expanded — the source explicitly rejects historical reconstruction and gives no current target, metric, or planning horizon.

Open questions
  • UNKNOWN. Exact question: What current financial result and time horizon would count as being sustainably balanced?
Decision

T-005 — Run the simple 11-day reset

Undecided
Task

Run the simple 11-day reset. Smallest useful result: Track the duration plus the smallest deciding behaviors: water ready, shakes prepared, priority work started, and a daily outdoor walk.

Problem solved

Daily basics have become inconsistent and reduce focus.

Cost of inaction

Unstable routines continue to impair execution.

Benefit

A short tracked reset can restore the conditions for focused work.

Done

An eleven-day reset is tracked against its confirmed core behaviors. 80/20: Track the duration plus the smallest deciding behaviors: water ready, shakes prepared, priority work started, and a daily outdoor walk.

flowchart LR
  A[Daily basics have become inconsistent and reduce focus] --> B[Run the simple 11-day reset] --> C[A short tracked reset can restore the conditions for focused work]
flowchart TD
  S1[Ground current source and evidence] --> S2[Track the duration plus the smallest deciding behaviors water ready]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • Does Matt want every listed habit to be a pass/fail condition, or are water, shake preparation, priority work, and the walk the required core?
Decision

T-006 — Stabilize sleep and deliberate recovery

Undecided
Task

Stabilize sleep and deliberate recovery. Smallest useful result: Protect the next long sleep and complete two short recovery sessions.

Problem solved

Insufficient sleep and recovery are reducing capacity.

Cost of inaction

Fatigue increases the risk of poor judgment and weak focus.

Benefit

Protected sleep and deliberate recovery can restore energy.

Done

The selected sleep and recovery commitments are scheduled and tracked. 80/20: Protect the next long sleep and complete two short recovery sessions.

flowchart LR
  A[Insufficient sleep and recovery are reducing capacity] --> B[Stabilize sleep and deliberate recovery] --> C[Protected sleep and deliberate recovery can restore energy]
flowchart TD
  S1[Ground current source and evidence] --> S2[Protect the next long sleep and complete two short recovery sessions]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • Which actual date should receive the sauna or spa block, since "tomorrow or Sunday" is relative to the recording?
Decision

T-007 — Resolve a sensitive wellbeing boundary

UndecidedNeeds input
Task

Resolve a sensitive wellbeing boundary. Smallest useful result: Record the contradiction and seek a deliberate decision.

Problem solved

The private source records a sensitive wellbeing conflict without a settled boundary.

Cost of inaction

Leaving the conflict unresolved may undermine stability and routine adherence.

Benefit

A deliberate supported decision can replace ambiguity.

Not expanded — the source directly conflicts between removing or avoiding cannabis and reserving one puff for a weekly walk. It does not settle a stimulant boundary either.

Open questions
  • Which safe decision process should resolve the private wellbeing conflict?
Decision

T-008 — Use coaching to sustain one chosen system

Undecided
Task

Use coaching to sustain one chosen system. Smallest useful result: Ask one coach to hold one weekly check-in on one already-selected routine.

Problem solved

One chosen routine needs continuity rather than another new system.

Cost of inaction

New routines may be started and abandoned repeatedly.

Benefit

One accountability relationship can reinforce continuity.

Done

UNKNOWN. Exact question: Which one system should the coaching commitment protect, with whom, and how will adherence be reviewed? 80/20: Ask one coach to hold one weekly check-in on one already-selected routine.

flowchart LR
  A[If coaching protects the reset] --> B[Review one reset weekly] --> C[Adherence becomes visible]
flowchart LR
  A[If coaching protects evening planning] --> B[Review daily closeouts] --> C[Planning continuity improves]
flowchart LR
  A[If coaching protects single focus] --> B[Review task switching] --> C[Execution discipline improves]
Open questions
  • UNKNOWN. Exact question: Which one system should the coaching commitment protect, with whom, and how will adherence be reviewed?
Decision

T-009 — Establish the fixed evening planning and accountability call

Undecided
Task

Establish the fixed evening planning and accountability call. Smallest useful result: Hold the call at one temporarily chosen evening time and leave with the handwritten top three and next-day calendar blocks.

Problem solved

Work continues without a consistent comparison between the plan and actual completion.

Cost of inaction

Long workdays can displace reflection, priorities drift, and the next day starts without a grounded plan.

Benefit

A daily closeout creates direct accountability and a prepared next day.

Done

A consistent daily call asks "what was the plan versus what I did?", course-corrects and closes the day, creates next-day calendar blocks, a handwritten top three, and handwritten steps, and includes the prompt to prepare shakes. The preferred rule is no work after the call. 80/20: Hold the call at one temporarily chosen evening time and leave with the handwritten top three and next-day calendar blocks.

flowchart LR
  A[Work continues without a consistent comparison between the plan and] --> B[Establish the fixed evening planning and accountability call] --> C[A daily closeout creates direct accountability and a prepared next day]
flowchart TD
  S1[Ground current source and evidence] --> S2[Hold the call at one temporarily chosen evening time and leave with the]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • What fixed time should be tested first within the unresolved 16:00 to 20:00 range?
Decision

T-010 — Feed the evening call a same-day session audit

Undecided
Task

Feed the evening call a same-day session audit. Smallest useful result: Give the prepared session comparison directly to the voice agent with no audio generation.

Problem solved

The planning call needs evidence of what actually happened across Codex and Claude sessions.

Cost of inaction

The plan-versus-done comparison may rely on memory and miss scattered or unfinished work.

Benefit

Prepared evidence makes the daily review short and accurate.

Done

Before the call, inspect the day's Codex and Claude sessions against the previous plan and provide the resulting context to the live voice agent. A six-minute audio digest remains optional. 80/20: Give the prepared session comparison directly to the voice agent with no audio generation.

flowchart LR
  A[The planning call needs evidence of what actually happened across Codex] --> B[Feed the evening call a same-day session audit] --> C[Prepared evidence makes the daily review short and accurate]
flowchart TD
  S1[Ground current source and evidence] --> S2[Give the prepared session comparison directly to the voice agent with]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • None material for the first version.
Decision

T-011 — Create a difficult-to-miss evening reminder chain

Undecided
Task

Create a difficult-to-miss evening reminder chain. Smallest useful result: One advance alarm plus one start-time alarm, followed by manually opening the planning call.

Problem solved

A single ordinary reminder may be ignored while Matt is absorbed in work.

Cost of inaction

The daily accountability loop fails at the point of initiation.

Benefit

A simple reliable signal makes stopping, reviewing, and planning more likely.

Done

UNKNOWN. Exact question: Which minimal combination should be tested first: alarms, advance warning, phone call, desktop popup, music, an already-open messaging channel, or automatic voice-page opening? 80/20: One advance alarm plus one start-time alarm, followed by manually opening the planning call.

flowchart LR
  A[If alarms are enough] --> B[Use advance and start alarms] --> C[The call begins on time]
flowchart LR
  A[If direct contact is needed] --> B[Add one call or open channel] --> C[The reminder is harder to miss]
flowchart LR
  A[If automation is needed] --> B[Open the voice page automatically] --> C[Planning starts with less friction]
Open questions
  • UNKNOWN. Exact question: Which minimal combination should be tested first: alarms, advance warning, phone call, desktop popup, music, an already-open messaging channel, or automatic voice-page opening?
Decision

T-012 — Make the personal voice agent accessible from a browser shortcut

Undecided
Task

Make the personal voice agent accessible from a browser shortcut. Smallest useful result: Bookmark the existing live voice page before building a custom backend.

Problem solved

The planning agent is harder to use consistently if access requires technical setup each time.

Cost of inaction

Friction can weaken adoption of the evening and weekly planning routines.

Benefit

A personal page and shortcut create a repeatable entry point.

Done

UNKNOWN. Exact question: Which existing voice-agent service and host should the personal browser page use, and who is allowed to access it? 80/20: Bookmark the existing live voice page before building a custom backend.

flowchart LR
  A[If the existing page is sufficient] --> B[Create a browser bookmark] --> C[Access takes one click]
flowchart LR
  A[If a personal landing page is needed] --> B[Wrap the current voice route] --> C[Entry becomes consistent]
flowchart LR
  A[If a new backend is necessary] --> B[Choose host and access boundary] --> C[A dedicated service can be tested]
Open questions
  • UNKNOWN. Exact question: Which existing voice-agent service and host should the personal browser page use, and who is allowed to access it?
Decision

T-013 — Write and read a one-page morning manifesto

Undecided
Task

Write and read a one-page morning manifesto. Smallest useful result: Handwrite one page from already-confirmed priorities and read it the next morning.

Problem solved

Goals, priorities, decisions, and principles are easy to forget at the start of the day.

Cost of inaction

Daily choices can drift away from settled priorities and lessons.

Benefit

A short handwritten page makes the most important ideas visible each morning.

Done

One short handwritten page covering goals, priorities, important decisions, and personal principles is read each morning. AI may help brainstorm, but the final page remains handwritten. 80/20: Handwrite one page from already-confirmed priorities and read it the next morning.

flowchart LR
  A[Goals priorities decisions and principles are easy to forget at the] --> B[Write and read a one-page morning manifesto] --> C[A short handwritten page makes the most important ideas visible each]
flowchart TD
  S1[Ground current source and evidence] --> S2[Handwrite one page from already-confirmed priorities and read it the]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • None. Not drinking alcohol remains only an example, not a settled manifesto rule.
Decision

T-014 — Build the deferred morning Anki principles deck

Undecided
Task

Build the deferred morning Anki principles deck. Smallest useful result: Create a small reviewed deck from the final morning manifesto, not from unfiltered AI suggestions.

Problem solved

Important concepts, rules, lessons, hard truths, and gratitude can fade from memory.

Cost of inaction

Useful principles may need to be rediscovered repeatedly.

Benefit

Spaced repetition can surface each idea when it is due in a short morning review.

Done

An Anki deck contains reviewed concepts, rules, principles, lessons, frequently forgotten truths, hard-to-swallow pills, and gratitude prompts, and supports an approximately 10-minute morning review. 80/20: Create a small reviewed deck from the final morning manifesto, not from unfiltered AI suggestions.

flowchart LR
  A[Important concepts rules lessons hard truths and gratitude can fade] --> B[Build the deferred morning Anki principles deck] --> C[Spaced repetition can surface each idea when it is due in a short]
flowchart TD
  S1[Ground current source and evidence] --> S2[Create a small reviewed deck from the final morning manifesto not from]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • What event should move this explicitly deferred idea into active work?
Decision

T-015 — Hold a weekly voice-agent planning walk

Undecided
Task

Hold a weekly voice-agent planning walk. Smallest useful result: Run one walk with the existing voice agent and no new interface.

Problem solved

Daily planning does not provide enough space for a broad brain dump and weekly course correction.

Cost of inaction

Goals, habits, and system updates may accumulate without deliberate review.

Benefit

A regular long walk combines outdoor time, reflection, and light coaching.

Done

Once a week, preferably Friday evening or Saturday, hold a planning walk that is roughly 90 percent Matt brain dump and 10 percent a small number of useful agent questions about goals, habits, and updates. 80/20: Run one walk with the existing voice agent and no new interface.

flowchart LR
  A[Daily planning does not provide enough space for a broad brain dump and] --> B[Hold a weekly voice-agent planning walk] --> C[A regular long walk combines outdoor time reflection and light coaching]
flowchart TD
  S1[Ground current source and evidence] --> S2[Run one walk with the existing voice agent and no new interface]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • None material for a first trial.
Decision

T-016 — Explore live iPad mind-mapping with a voice agent

UndecidedNeeds input
Task

Explore live iPad mind-mapping with a voice agent. Smallest useful result: Draw the mind map manually during or after one weekly voice walk.

Problem solved

Spoken planning alone may not capture spatial relationships in a developing mind map.

Cost of inaction

Potential visual collaboration remains unavailable, though no current priority depends on it.

Benefit

Matt could draw while the agent observes, edits, and contributes a requested number of ideas.

Not expanded — the source itself calls the concept possible overcomplication and provides no chosen devices, software, or interaction model.

Open questions
  • UNKNOWN. Exact question: What planning problem must the live iPad interface solve that a manual drawing plus voice conversation does not?
Decision

T-017 — Apply a manual-first rule to systems, workflows, and generated skills

Undecided
Task

Apply a manual-first rule to systems, workflows, and generated skills. Smallest useful result: Require one manual sketch and one full human read before accepting an AI-generated workflow or skill.

Problem solved

Delegating the first version entirely to AI can create systems Matt has not thought through or read carefully.

Cost of inaction

Workflows may become complex, poorly understood, and difficult to trust.

Benefit

Human reasoning stays visible and AI becomes a reviewer and accelerator instead of the sole designer.

Done

First versions of systems, loops, workflows, skills, personas, and related thinking are manually drawn or written. Generated skills are read carefully before use. 80/20: Require one manual sketch and one full human read before accepting an AI-generated workflow or skill.

flowchart LR
  A[Delegating the first version entirely to AI can create systems Matt has] --> B[Apply a manual-first rule to systems workflows and generated skills] --> C[Human reasoning stays visible and AI becomes a reviewer and accelerator]
flowchart TD
  S1[Ground current source and evidence] --> S2[Require one manual sketch and one full human read before accepting an]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • None.
Decision

T-018 — Revisit visual n8n workflow documentation with Dani

UndecidedNeeds input
Task

Revisit visual n8n workflow documentation with Dani. Smallest useful result: Ask Dani to walk through one existing workflow while Matt sketches it manually.

Problem solved

Existing n8n systems may lack a robust, shared visual explanation.

Cost of inaction

The systems may remain harder to understand and maintain, but the source identifies no immediate consequence.

Benefit

Dani's explanations could become robust, reusable workflow documentation.

Not expanded — the source explicitly says to skip this complexity for now and does not name the workflows or desired output.

Open questions
  • UNKNOWN. Exact question: Which existing n8n workflows should Dani explain, and what reviewable artifact should result?
Decision

T-019 — Clarify the future agent-club or thinking-framework course idea

UndecidedNeeds input
Task

Clarify the future agent-club or thinking-framework course idea. Smallest useful result: Save the idea with its uncertainty and revisit only during a designated future-ideas review.

Problem solved

A possible course or community concept exists only as a partly unclear reference.

Cost of inaction

A hobby or future-product idea may be forgotten, but no current priority is blocked.

Benefit

Clarification could preserve a potential product without allowing it to distract from urgent work.

Not expanded — the source cannot safely identify the exact course reference, audience, format, or offer.

Open questions
  • UNKNOWN. Exact question: Is the intended idea an agent club, a course on thinking frameworks and project management, Anki material, or some combination, and for whom?
Decision

T-020 — Organize active work inside Codex

Undecided
Task

Organize active work inside Codex. Smallest useful result: Rename and categorize the current sessions, then reduce the active list to no more than five.

Problem solved

Current priorities are not visibly represented where most work already happens.

Cost of inaction

Too many sessions stay active, waiting work is unclear, and task switching continues.

Benefit

A low-friction view in the existing tool makes focus and status visible.

Done

Each Codex session represents one task and has a status, project or client identity in the title with a custom emoji, categories for current work, waiting, and unprocessed ideas, and an active list limited to four or five items. 80/20: Rename and categorize the current sessions, then reduce the active list to no more than five.

flowchart LR
  A[Current priorities are not visibly represented where most work already] --> B[Organize active work inside Codex] --> C[A low-friction view in the existing tool makes focus and status visible]
flowchart TD
  S1[Ground current source and evidence] --> S2[Rename and categorize the current sessions then reduce the active list]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • What exact status separates a waiting item from a waiting item that Matt has already processed?
Decision

T-021 — Decide whether long-lived automation work needs Linear or another ticket system

Undecided
Task

Decide whether long-lived automation work needs Linear or another ticket system. Smallest useful result: Use a technical specification and small tickets inside the current project before adopting a new tool.

Problem solved

Large automation systems may need vision-to-spec-to-ticket decomposition, while another tool could create overhead.

Cost of inaction

Complex builds may be handed to agents as oversized task lists, or a new product-management system may be overbuilt unnecessarily.

Benefit

A proportional rule can preserve rigor for long-lived work without complicating daily personal task handling.

Done

UNKNOWN. Exact question: Which concrete automation project cannot be managed adequately in Codex alone, and what capability would Linear, Notion, Kanban, or paper add? 80/20: Use a technical specification and small tickets inside the current project before adopting a new tool.

flowchart LR
  A[If Codex is sufficient] --> B[Use specs and task sessions] --> C[No new tool is added]
flowchart LR
  A[If a simple board is needed] --> B[Use Notion or Kanban] --> C[Long work becomes visible]
flowchart LR
  A[If engineering tickets are needed] --> B[Pilot Linear on one project] --> C[Tool value can be measured]
Open questions
  • UNKNOWN. Exact question: Which concrete automation project cannot be managed adequately in Codex alone, and what capability would Linear, Notion, Kanban, or paper add?
Decision

T-022 — Finish the overdue paid-client design and proposal backlog

Undecided
Task

Finish the overdue paid-client design and proposal backlog. Smallest useful result: Ground in the existing HHÉP and proposal assets, identify the smallest promised deliverable, and make it review-ready before any send.

Problem solved

A paid-client design and earlier proposals remain pending.

Cost of inaction

Clients wait longer and commercial trust declines.

Benefit

Finishing the highest-cost obligation removes an urgent liability.

Done

The exact promised package is review-ready. External sending remains separately approval-gated. 80/20: Ground in the existing HHÉP and proposal assets, identify the smallest promised deliverable, and make it review-ready before any send.

flowchart LR
  A[A paid-client design and earlier proposals remain pending] --> B[Finish the overdue paid-client design and proposal backlog] --> C[Finishing the highest-cost obligation removes an urgent liability]
flowchart TD
  S1[Ground current source and evidence] --> S2[Ground in the existing HHÉP and proposal assets identify the smallest]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • Which exact assets define the backlog, and what approval is required before delivery?
Decision

T-023 — Productize the fast proposal workflow for the sales representative

Undecided
Task

Productize the fast proposal workflow for the sales representative. Smallest useful result: Test the current mechanism on one real proposal and add only the smallest frontend improvement that removes a demonstrated blocker.

Problem solved

Proposal creation after sales calls is a bottleneck that keeps Matt involved and delays follow-up.

Cost of inaction

Proposals continue to arrive late and the sales representative cannot independently move calls toward a ready offer.

Benefit

A simple repeatable tool enables rapid, evidence-informed proposals and clearer delegation.

Done

The existing mechanism is tested through real proposal creation. It can generate three or four short two-page variants, use relevant cold-call or cold-email transcripts, let the representative select and optionally edit one, and finalize it after the sales meeting from the transcript plus selected HTML and Figma assets. The target is about five minutes to fill and about 15 minutes after the call to be ready to send. Promised design, screenshots, and annotations are included when relevant. 80/20: Test the current mechanism on one real proposal and add only the smallest frontend improvement that removes a demonstrated blocker.

flowchart LR
  A[Proposal creation after sales calls is a bottleneck that keeps Matt] --> B[Productize the fast proposal workflow for the sales representative] --> C[A simple repeatable tool enables rapid evidence-informed proposals and]
flowchart TD
  S1[Ground current source and evidence] --> S2[Test the current mechanism on one real proposal and add only the]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • Which current proposal mechanism is authoritative, and what edits may the sales representative make without Matt's review?
Decision

T-024 — Correct PPC measurement across the four ongoing projects

Undecided
Task

Correct PPC measurement across the four ongoing projects. Smallest useful result: Inventory the four projects and verify one highest-value conversion path per project before designing a universal system.

Problem solved

Measurement across four active PPC projects is not sufficiently correct or consistent.

Cost of inaction

Client results and optimization decisions remain unreliable, weakening both delivery and reporting.

Benefit

A good-enough shared measurement structure creates trustworthy inputs for each client portal.

Done

UNKNOWN. Exact question: Which four PPC projects are in scope, what measurement is wrong in each, and what verified event or business outcome should each project count? 80/20: Inventory the four projects and verify one highest-value conversion path per project before designing a universal system.

flowchart LR
  A[If the gap is tracking configuration] --> B[Verify one conversion path per project] --> C[Events become trustworthy]
flowchart LR
  A[If the gap is business reconciliation] --> B[Link events to real outcomes] --> C[Reported results gain meaning]
flowchart LR
  A[If the gap is cross-client consistency] --> B[Define one shared measurement contract] --> C[Projects become comparable]
Open questions
  • UNKNOWN. Exact question: Which four PPC projects are in scope, what measurement is wrong in each, and what verified event or business outcome should each project count?
Decision

T-025 — Build version one of the shared PPC client intelligence portal

Undecided
Task

Build version one of the shared PPC client intelligence portal. Smallest useful result: Choose one client and publish a read-only internal prototype with real verified measurement plus one visual each for persona, offer, funnel, and.

Problem solved

Research, strategy, measurement, ads, CRM information, and design are scattered instead of visible in one consistent client surface.

Cost of inaction

Clients and sales conversations lack a coherent view of agency thinking and results, while each project risks being rebuilt differently.

Benefit

A reusable portal combines real data with clear strategy and research visuals.

Done

A reusable V1 for one selected client shows a deliberately bounded subset of personas and objections, problems and pain points, offers, money models, funnels, verified measurement and results, competitors, ads and active tests, CRM information, and website design. 80/20: Choose one client and publish a read-only internal prototype with real verified measurement plus one visual each for persona, offer, funnel, and current tests.

flowchart LR
  A[If V1 is measurement first] --> B[Show verified results and tests] --> C[Delivery becomes reviewable]
flowchart LR
  A[If V1 is sales enablement first] --> B[Show persona offer and funnel] --> C[Expertise becomes visible]
flowchart LR
  A[If V1 is an internal intelligence hub] --> B[Combine research data and design] --> C[One operating surface exists]
Open questions
  • Which client is the V1 pilot, which sections are required at launch, and which data sources are authoritative?
Decision

T-026 — Create reusable research visuals and Figma-to-HTML templates

Undecided
Task

Create reusable research visuals and Figma-to-HTML templates. Smallest useful result: Manually draft one persona, problem, offer, and funnel set, render them as simple Mermaid or one Figma template, then test HTML comment review.

Problem solved

Persona, problem, offer, money-model, and solution thinking lacks a consistent reusable presentation.

Cost of inaction

Every client analysis remains custom and harder to compare, discuss, or publish into the portal.

Benefit

Human-authored research can be reviewed consistently and converted into reusable portal views.

Done

Matt and the client form initial persona ideas. Pen to Profit informs persona and pain-point work. Hormozi's framework, including delivery economics, informs offers. Money models connect offers to funnels. Reusable Figma templates cover personas, problems, offers, and step-by-step solutions, convert to HTML, and accept AI comments through the existing comment-HTML workflow. Simple Mermaid can serve as the first visual version. 80/20: Manually draft one persona, problem, offer, and funnel set, render them as simple Mermaid or one Figma template, then test HTML comment review.

flowchart LR
  A[Persona problem offer money-model and solution thinking lacks a] --> B[Create reusable research visuals and Figma-to-HTML templates] --> C[Human-authored research can be reviewed consistently and converted into]
flowchart TD
  S1[Ground current source and evidence] --> S2[Manually draft one persona problem offer and funnel set render them as]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • Which Pen to Profit and Hormozi source materials are approved inputs for the first template?
Decision

T-027 — Decide portal authentication and information-visibility policy

UndecidedNeeds input
Task

Decide portal authentication and information-visibility policy. Smallest useful result: Keep the first portal internal and read-only until the visibility matrix and authentication reference are confirmed.

Problem solved

The portal needs simple authentication, and the boundary between client-visible, internal, locked, and post-purchase information is unresolved.

Cost of inaction

Sensitive internal work could be exposed, or excessive hiding could reduce the portal's usefulness and credibility.

Benefit

A clear access and visibility model makes portal building and client use safer and more consistent.

Not expanded — the reference client's name is unclear and the source does not settle the visibility tradeoff.

Open questions
  • UNKNOWN. Exact question: What must be public, prospect-visible, client-visible, post-purchase locked, or internal, and which known Cloudflare login implementation should be reused?
Decision

T-028 — Build screenshot-based funnel journey maps

Undecided
Task

Build screenshot-based funnel journey maps. Smallest useful result: Map one ad-to-landing-page journey for one persona with screenshots and thoughts or feelings at each step.

Problem solved

Funnel analysis does not show the full user experience or the target persona's response at each stimulus.

Cost of inaction

Friction and weak messaging across ads, pages, and later steps may remain invisible.

Benefit

A left-to-right journey makes interfaces, reactions, and cross-step patterns reviewable.

Done

For one funnel, a left-to-right table includes a screenshot of every interface the user sees and records what the target persona sees, hears, wants, thinks, and feels from awareness and ad exposure onward. Patterns across observations are summarized. 80/20: Map one ad-to-landing-page journey for one persona with screenshots and thoughts or feelings at each step.

flowchart LR
  A[Funnel analysis does not show the full user experience or the target] --> B[Build screenshot-based funnel journey maps] --> C[A left-to-right journey makes interfaces reactions and cross-step]
flowchart TD
  S1[Ground current source and evidence] --> S2[Map one ad-to-landing-page journey for one persona with screenshots and]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • Is identifying the forgotten formal framework name important, or can the described fields be used directly?
Decision

T-029 — Pilot AI persona interviews, then evaluate scaled simulation

Undecided
Task

Pilot AI persona interviews, then evaluate scaled simulation. Smallest useful result: Run a small labeled pilot in which one AI persona reacts to one ad and landing-page journey, then compare its observations with human judgment.

Problem solved

Real-user research may be unaffordable initially, leaving persona and funnel assumptions weakly tested.

Cost of inaction

The agency relies only on unchallenged educated guesses. Conversely, over-trusting simulation could create false confidence.

Benefit

A clearly labeled simulation pilot can critique early assumptions before real-world validation.

Done

UNKNOWN. Exact question: Which persona and funnel should be piloted, how many simulated interviews are enough for the pilot, and what result would justify a later hundreds-of-interviews phase? 80/20: Run a small labeled pilot in which one AI persona reacts to one ad and landing-page journey, then compare its observations with human judgment.

flowchart LR
  A[Real-user research may be unaffordable initially leaving persona and] --> B[Pilot AI persona interviews then evaluate scaled simulation] --> C[A clearly labeled simulation pilot can critique early assumptions]
flowchart TD
  S1[Ground current source and evidence] --> S2[Run a small labeled pilot in which one AI persona reacts to one ad and]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • UNKNOWN. Exact question: Which persona and funnel should be piloted, how many simulated interviews are enough for the pilot, and what result would justify a later hundreds-of-interviews phase?
Decision

T-030 — Produce lightweight competitor analyses

Undecided
Task

Produce lightweight competitor analyses. Smallest useful result: Analyze one verified competitor's business, offer, funnel, and strongest observable ads using source-linked evidence.

Problem solved

Persona and offer work lacks a consistent view of real competitors' businesses, offers, funnels, and ads.

Cost of inaction

Strategy may ignore proven market patterns and meaningful differentiation.

Benefit

A lighter version of the agency analysis creates usable competitive context before or alongside persona and offer work.

Done

UNKNOWN. Exact question: For which client or market, which real competitors, and which evidence standard should the first analysis use? 80/20: Analyze one verified competitor's business, offer, funnel, and strongest observable ads using source-linked evidence.

flowchart LR
  A[Persona and offer work lacks a consistent view of real competitors] --> B[Produce lightweight competitor analyses] --> C[A lighter version of the agency analysis creates usable competitive]
flowchart TD
  S1[Ground current source and evidence] --> S2[Analyze one verified competitor s business offer funnel and strongest]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • UNKNOWN. Exact question: For which client or market, which real competitors, and which evidence standard should the first analysis use?
Decision

T-031 — Record manual research and convert it into reusable skills

Undecided
Task

Record manual research and convert it into reusable skills. Smallest useful result: Record and document one competitor-research run before creating any broad automation.

Problem solved

Competitor and persona research methods live in ad hoc manual behavior rather than a repeatable process.

Cost of inaction

Research quality varies and cannot be delegated reliably to agents on the Linux development box or VPS.

Benefit

A recorded, reviewed human process can become a grounded reusable skill.

Done

Record one manual competitor-research process while using Google and, if chosen, Perplexity through the available Anthropic or Codex recording capability. Review the demonstrated steps, convert them into a reusable skill, and validate that an agent can reproduce the bounded process. Repeat later for persona research if useful. 80/20: Record and document one competitor-research run before creating any broad automation.

flowchart LR
  A[Competitor and persona research methods live in ad hoc manual behavior] --> B[Record manual research and convert it into reusable skills] --> C[A recorded reviewed human process can become a grounded reusable skill]
flowchart TD
  S1[Ground current source and evidence] --> S2[Record and document one competitor-research run before creating any]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • Which currently available screen-recording feature is approved and can capture the process without exposing credentials or private data?
Decision

T-032 — Research a sound creative and funnel A/B testing method

Undecided
Task

Research a sound creative and funnel A/B testing method. Smallest useful result: Produce a source-grounded test-design checklist for one platform and one creative comparison before building a live scoreboard.

Problem solved

The portal may show ad and funnel variants, but the comparison method is not yet trustworthy and Matt explicitly says he is not an expert in it.

Cost of inaction

Tests may be designed or interpreted incorrectly, leading to weak optimization decisions.

Benefit

A research-grounded method supports defensible testing plans and a meaningful live scoreboard.

Done

UNKNOWN. Exact question: Which ad platforms and funnel decisions need a testing method first, and what traffic and decision constraints should the method assume? 80/20: Produce a source-grounded test-design checklist for one platform and one creative comparison before building a live scoreboard.

flowchart LR
  A[If the first need is creative testing] --> B[Define a creative comparison rule] --> C[Ad choices become defensible]
flowchart LR
  A[If the first need is funnel testing] --> B[Define a step comparison rule] --> C[Journey changes can be evaluated]
flowchart LR
  A[If the first need is a scoreboard] --> B[Define metrics before building UI] --> C[Results remain interpretable]
Open questions
  • UNKNOWN. Exact question: Which ad platforms and funnel decisions need a testing method first, and what traffic and decision constraints should the method assume?
Decision

T-033 — Build the evergreen data, CRM, and reporting layer

Undecided
Task

Build the evergreen data, CRM, and reporting layer. Smallest useful result: Connect one verified metric source and one read-only CRM view for one client, then generate one plain slide summary.

Problem solved

Real data, CRM context, and recurring reports are not yet combined into a continuously useful client view.

Cost of inaction

Reporting remains manual or fragmented and the portal can become a static research artifact.

Benefit

An evergreen dashboard plus CRM context supports delivery, sales conversations, and repeatable reports.

Done

A selected client's portal refreshes from verified real data, connects or embeds the intended Google Sheet CRM, and can generate a Google Slides report from the same evidence. 80/20: Connect one verified metric source and one read-only CRM view for one client, then generate one plain slide summary.

flowchart LR
  A[If the first layer is a live dashboard] --> B[Connect one verified metric] --> C[Current results stay visible]
flowchart LR
  A[If the first layer is CRM context] --> B[Embed one read only CRM view] --> C[Results connect to enquiries]
flowchart LR
  A[If the first layer is reporting] --> B[Generate one plain slide summary] --> C[Evidence becomes shareable]
Open questions
  • Which data sources and Google Sheet are authoritative, how often must they refresh, and should the report look cleanly automated or intentionally imitate manual production? The suggested deliberate grammatical errors are preserved as an unresolved idea and are not endorsed or authorized.
Decision

T-034 — Create the prospect-specific living proposal portal

Undecided
Task

Create the prospect-specific living proposal portal. Smallest useful result: Create one internal prospect-specific prototype with a proposal first page, one research section, one design, and comments.

Problem solved

A separate proposal plus scattered research, designs, and review links fragment the sales and delivery experience.

Cost of inaction

Prospects receive disconnected artifacts, and the agency loses a strong way to demonstrate prepared expertise during and after the call.

Benefit

One prospect-specific surface can serve as proposal, research library, design review, and future client portal.

Done

A reusable template can be duplicated per prospect, prefilled with known information, shown during the sales call, and sent afterward. The first page is the proposal. Research and resources sit behind it. Future sections can be locked until purchase. Website designs and the comment or review interface live in the same portal. 80/20: Create one internal prospect-specific prototype with a proposal first page, one research section, one design, and comments.

flowchart LR
  A[A separate proposal plus scattered research designs and review links] --> B[Create the prospect-specific living proposal portal] --> C[One prospect-specific surface can serve as proposal research library]
flowchart TD
  S1[Ground current source and evidence] --> S2[Create one internal prospect-specific prototype with a proposal first]
  S2 --> S3[Complete only the confirmed scope]
  S3 --> D{Done evidence verified}
  D -- yes --> FINISH[Verified result]
  D -- no --> S2
Open questions
  • Which prospect should be the first prototype, and which sections must be unlocked before versus after purchase?
Decision

T-035 — Move toward approximately three fixed-price packages

UndecidedNeeds input
Task

Move toward approximately three fixed-price packages. Smallest useful result: Review recent proposals and cluster recurring scopes before naming or pricing packages.

Problem solved

Fully custom offers increase proposal complexity and slow selection and follow-up.

Cost of inaction

Each proposal requires more reinvention and may remain harder for the sales representative to choose or explain.

Benefit

A small package set could simplify the proposal workflow and sales conversations.

Not expanded — only the approximate count and direction are given. Scope, price, positioning, and eligibility are absent.

Open questions
  • UNKNOWN. Exact question: What customer segments, outcomes, scope boundaries, and price points should the approximately three packages represent?
Decision

T-036 — Define the free diagnostic measurement product

Undecided
Task

Define the free diagnostic measurement product. Smallest useful result: Define and manually demonstrate one useful website or measurement alert for one internal test prospect before automating the product.

Problem solved

Prospects may have no immediate proof of value or ongoing signal that the agency can detect and fix site and marketing problems.

Cost of inaction

The agency loses a low-friction entry product and a continuing reason for prospects to re-engage.

Benefit

A free setup and dashboard could create useful alerts for website errors, SEO declines, and related issues while demonstrating capability.

Done

UNKNOWN. Exact question: What measurements and alerts are included for free, who receives them, through which channel, for how long, and what paid next step should they lead to? 80/20: Define and manually demonstrate one useful website or measurement alert for one internal test prospect before automating the product.

flowchart LR
  A[If the diagnostic is site health] --> B[Monitor one actionable error] --> C[A useful alert is demonstrated]
flowchart LR
  A[If the diagnostic is SEO visibility] --> B[Monitor one ranking signal] --> C[Declines can trigger review]
flowchart LR
  A[If the diagnostic is paid measurement] --> B[Monitor one conversion path] --> C[Tracking failures become visible]
Open questions
  • UNKNOWN. Exact question: What measurements and alerts are included for free, who receives them, through which channel, for how long, and what paid next step should they lead to?
Decision