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 fallbackVoice 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.
Plan: /Users/agency/Documents/Agty/KERTÚJÍTÓK/action-plans/2026-08-09-kertujitok-rescope-review.md
J/K, to open it.1 delegate, 2 drop, 3 idea, 4 ask first, 0 clear, J/K move.! to make it binding.Identify the real Cloudflare build and establish what can safely be promised before migration.
Hosting, ownership, SEO, forms, tracking, email, backup, and rollback parity are not yet proven.
A migration offer could hide a continuity risk or interrupt an important client service.
A factual migration case, explicit blockers, and a reversible route for the offer and Loom.
The exact build and staging target are identified; parity, ownership, SEO, forms, tracking, email, backup, rollback, and review integration are evidenced. 80/20: identify the build and the few blockers that change the client promise.
flowchart LR A[Unverified migration] --> B[Continuity audit] --> C[Safe client promise]
flowchart TD
S1[Identify build] --> S2[Check parity and ownership]
S2 --> S3[Check tracking email and rollback]
S3 --> D{Evidence complete}
D -- yes --> DONE[Migration recommendation]
D -- no --> S2
Make each lead easier to understand and manage, especially on a phone.
The promised app needs a clearer overview and faster lead-management flow.
A client promise stays open and follow-up remains harder than necessary.
The client can quickly see the enquiry, note, stage, reminder, and next action.
A staging flow passes synthetic end-to-end and responsive checks at desktop, tablet, and mobile sizes. 80/20: make the mobile lead card understandable and actionable without extra views.
flowchart LR A[Scattered lead details] --> B[Clear mobile lead card] --> C[Faster follow up]
flowchart TD
S1[Reproduce client flow] --> S2[Build smallest clarity fixes]
S2 --> S3[Test actions and layouts]
S3 --> D{Staging passes}
D -- yes --> DONE[Production gate ready]
D -- no --> S2
Reproduce the client-visible problem and repair only confirmed dashboard defects in staging.
The dashboard is described as broken, but the failure may be availability, stale data, or misleading metrics.
The client cannot rely on the reporting surface or the reduced monthly service.
A stable view with honest source, freshness, and limitation states.
A defect ledger, staging repair, tests, source states, and responsive checks pass. 80/20: reproduce the actual failure and restore one trustworthy client view.
flowchart LR A[If the issue is access] --> B[Trace route and authorization] --> C[Restore safe access]
flowchart LR A[If the issue is stale data] --> B[Trace source and freshness] --> C[Show current truthful data]
flowchart LR A[If the issue is wrong metrics] --> B[Reconcile definitions] --> C[Correct client view]
Finalize the commercial scope and draft one persuasive Hungarian migration and monthly-service offer.
The direction exists, but scope, terms, verified claims, and GHL consequences remain incomplete.
The current engagement may end without a clear replacement.
One factual client-ready offer with explicit inclusions, exclusions, and next steps.
Approved terms, verified migration claims, accurate GHL wording, bonus scope, and an unsent Hungarian draft exist. 80/20: lock the scope and produce one short factual email.
flowchart LR A[Unclear next period] --> B[Verified scoped offer] --> C[Client decision]
flowchart TD
S1[Ground migration claims] --> S2[Resolve commercial terms]
S2 --> S3[Draft and verify offer]
S3 --> D{Matt approves terms}
D -- yes --> DONE[Unsent final draft]
D -- no --> S2
Prepare a concise walkthrough of the verified setup, benefits, controls, and rollback.
The client needs a concrete, non-technical explanation before considering migration.
The offer stays abstract and harder to trust.
A reviewable explanation aligned with the verified offer.
A playable local draft and exact Hungarian script cover the verified architecture, ownership, tracking, SEO, bonus, backup, rollback, and unknowns. 80/20: an approved script and short staging walkthrough.
flowchart LR A[Abstract migration] --> B[Verified visual walkthrough] --> C[Lower perceived risk]
flowchart TD
S1[Turn evidence into script] --> S2[Record verified staging]
S2 --> S3[Check playback and claims]
S3 --> D{Review safe}
D -- yes --> DONE[Loom draft ready]
D -- no --> S1
After explicit final approval, send the exact approved offer once and preserve the receipt.
The must-do outcome remains incomplete while the offer is internal.
The client receives no replacement offer before the commercial change.
One reviewed proposal reaches the correct recipient with auditable proof.
The immutable approved copy, recipient, channel, and attachments are sent once and read back with a timestamp and message identifier. 80/20: exact approval before sending.
flowchart LR A[Approved internal offer] --> B[Single controlled send] --> C[Verified receipt]
flowchart TD
S1[Freeze copy and recipient] --> D{Exact approval}
D -- yes --> S2[Send once]
D -- no --> S1
S2 --> DONE[Read back receipt]
Request the current workbook and examples, then scope one grounded automation concept after permission and files arrive.
The upsell is generic until the current quotation workflow is understood.
A proposed tool could remove useful functions or offer unclear value.
One concrete, low-risk automation idea grounded in the real workflow.
An unsent privacy-safe request exists and, after inputs arrive, one evidence-based optional scope is reviewable. 80/20: prepare the request and let missing files gate the concept.
flowchart LR A[Generic software idea] --> B[Real workbook evidence] --> C[Grounded optional scope]
flowchart TD
S1[Draft file request] --> D{Inputs received}
D -- yes --> S2[Map workflow and exceptions]
D -- no --> WAIT[Wait without guessing]
S2 --> DONE[Optional concept ready]
Classify the earlier plan and prepare one source-linked future-ideas registry.
Old work should stop driving the current scope without losing evidence or opportunities.
Future agents may repeat work or mistake proposals for commitments.
A small active scope with prior context still easy to recover.
Every earlier item is classified and linked; one concise CRM-ready draft and placement proposal await review. 80/20: complete the local registry without external writes.
flowchart LR A[Mixed old scope] --> B[Classified evidence registry] --> C[Clean active plan]
flowchart TD
S1[Inventory prior tasks] --> S2[Classify and link evidence]
S2 --> S3[Draft registry and placement]
S3 --> D{Complete coverage}
D -- yes --> DONE[Ready for review]
D -- no --> S1
Close the duplicate-conversion, paid-session attribution, and changing-landing-page gaps from the 30-day report.
Ads events, form activity, phone clicks, connected calls, leads, and sales do not yet reconcile.
Campaign and landing-page decisions can optimize toward duplicated or soft events.
One auditable funnel from paid visit to contactable lead, with stable landing-page reporting.
Goal cleanup is re-verified; form events are deduplicated; clicks and calls stay separate; paid interactions reconcile to three stable landing pages; tests, rollback, and next-report queries exist. 80/20: make one paid landing page report an unambiguous interaction rate.
flowchart LR A[Mixed platform events] --> B[Deduplicated paid funnel] --> C[Trustworthy campaign decisions]
flowchart TD
S1[Reconcile current sources] --> S2[Define unique business event]
S2 --> S3[Restore paid attribution and stable pages]
S3 --> D{Tests and rollback pass}
D -- yes --> DONE[Next report measurable]
D -- no --> S1
Refresh and correct no-result search spend without optimizing against broken conversions.
Some search activity consumed meaningful spend with weak reported outcomes while measurement and delivery remain unreliable.
Money can keep flowing to weak queries or inactive delivery without a trustworthy result signal.
An approval-ready allocation based on search-term quality and verified business events.
Delivery, billing gate, queries, negatives, spend, and reconciled outcomes are refreshed; approved changes have before-state, rollback, read-back, and monitoring evidence. 80/20: identify the few sources of most avoidable spend.
flowchart LR A[Unreconciled search spend] --> B[Fresh query and outcome audit] --> C[Approved efficient allocation]
flowchart TD
S1[Refresh delivery and outcomes] --> S2[Rank waste and intent]
S2 --> D{Matt approves exact changes}
D -- yes --> S3[Apply and read back]
D -- no --> HOLD[Keep account unchanged]
S3 --> DONE[Monitoring receipt]
Turn the strongest existing organic topic into a verified, migration-safe SEO improvement.
One service page carries most of the site's current organic opportunity but still sits outside the strongest result positions.
The clearest low-effort SEO opportunity may be lost during migration.
A focused improvement with a preserved URL and measurable post-migration benchmark.
Current baseline, intent gap, technical and internal-link checks, staging change, migration rules, rollback, and a 60-day measurement plan exist. 80/20: fix the strongest evidence-backed on-page or linking gap in staging.
flowchart LR A[Visible topic outside top results] --> B[Focused migration safe improvement] --> C[Higher qualified visibility]
flowchart TD
S1[Refresh query baseline] --> S2[Find smallest supported gap]
S2 --> S3[Build and test in staging]
S3 --> D{URL and signals preserved}
D -- yes --> DONE[Measurement plan ready]
D -- no --> S2
Compare the blocked Google forwarding-number route with suitable third-party call tracking and recommend one reversible pilot.
Phone-link clicks still cannot prove a connected call, source, duration, or business outcome.
Phone demand remains invisible and campaign value stays uncertain.
A defensible call-tracking route with attribution, integrations, privacy controls, reliability, and known cost.
Current GFN proof and blocker are refreshed; an official-source vendor matrix covers local numbers, dynamic insertion, call details, integrations, consent, portability, reliability, cost, and rollback; one pilot or no-go is recommended. 80/20: prove whether any vendor meets the essential requirements better than GFN.
flowchart LR A[Unproven call attribution] --> B[GFN and vendor comparison] --> C[Reversible evidence based pilot]
flowchart TD
S1[Refresh GFN proof state] --> S2[Compare official vendor evidence]
S2 --> S3[Design privacy cost and rollback gates]
S3 --> D{Option beats current path}
D -- yes --> DONE[Approval ready pilot]
D -- no --> HOLD[Keep GFN or no go]