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/ColdEmail/action-plans/2026-08-13-danubisoft-growth-system.md
J/K, to open it.1 delegate, 2 drop, 3 idea, 4 ask first, 0 clear, J/K to move between tasks and open the one you land on.!.Create one dated evidence ledger for current campaign state, sender readiness, deliverability signals, and business outcomes.
Current guidance is split across live platform fields and older snapshots, so campaign and sender readiness are not yet one trustworthy view.
Scaling before reconciliation can place more volume on unhealthy senders or repeat stale and misleading result claims.
You get a source-linked picture that separates sending activity, replies, opportunities, meetings, and revenue.
Every campaign and sender is inventoried with timestamps, metric definitions, gaps, and an independent read-only QA verdict. 80/20. Current sender readiness plus sent, reply, positive-reply, and campaign-state evidence.
flowchart LR
A[Scattered evidence] --> B[Current audit] --> C[Trusted decisions]
flowchart TD
S1[Ground sources]
S2[Inventory campaigns and senders]
S3[Reconcile outcomes]
S4[Independent QA]
S1 --> S2
S2 --> S3
S3 --> S4
S4 --> D{Evidence verified}
D -- yes --> DONE[Task complete]
D -- no --> S1
Build a concise Hungarian management presentation from verified evidence, with results, current health, lessons, and next decisions.
The promised leadership report does not yet reflect the newest meeting, email, sender evidence, and complete forward plan.
The client lacks a clear view of delivered value, current risk, and the decisions needed to move forward.
You receive a polished, sendable report that makes progress and next steps easy to understand.
A verified 16:9 Google Slides deck and PDF use source-linked facts, render cleanly, and have the intended permissions. 80/20. A five-to-eight-slide story covering result, health, direct campaign, partner pilot, and decisions.
flowchart LR
A[Raw campaign evidence] --> B[Executive story] --> C[Client clarity]
flowchart TD
S1[Read audit receipts]
S2[Build management narrative]
S3[Add sources and caveats]
S4[Export and visual QA]
S1 --> S2
S2 --> S3
S3 --> S4
S4 --> D{Evidence verified}
D -- yes --> DONE[Task complete]
D -- no --> S1
Turn the supplied company workbook into traceable eligible, excluded, invalid, duplicate, and research-required cohorts.
The client-supplied list is not yet proven safe, complete, or free from prior-contact and suppression conflicts.
Unsafe preparation can cause duplicate outreach, client conflicts, invalid uploads, or silent loss of useful companies.
You get a trustworthy base cohort and an exact queue of companies needing contact research or replacement.
Every source row is reconciled through immutable inputs, normalized identity, suppressions, validation, quarantines, and balanced counts. 80/20. Exact overlap, duplicate, invalid, missing-contact, and provisionally eligible counts.
flowchart LR
A[Unproven workbook] --> B[Reconciled cohorts] --> C[Safe campaign base]
flowchart TD
S1[Freeze source]
S2[Normalize identities]
S3[Check overlap and suppression]
S4[Validate and quarantine]
S5[Reconcile all rows]
S1 --> S2
S2 --> S3
S3 --> S4
S4 --> S5
S5 --> D{Evidence verified}
D -- yes --> DONE[Task complete]
D -- no --> S1
Connect the existing lead-bank components into one maintainable path from acquisition to a frozen, validated, paused-upload batch.
Strong components exist, but the client needs one short operating path with clear inputs, gates, reruns, and receipts.
Bespoke preparation increases duplicate data, silent row loss, inconsistent personalization, and unsafe provider uploads.
You get a fast, testable process that can grow lead volume while preserving provenance and quality.
A documented command path and tests cover acquisition, identity, contact research, verification, suppression, personalization, frozen batches, and dry-run destinations. 80/20. A deterministic dry run takes the research queue to a validated frozen batch without external writes.
flowchart LR
A[Separate components] --> B[Verified pipeline] --> C[Repeatable scale]
flowchart TD
S1[Map existing modules]
S2[Add missing adapter seams]
S3[Add gates and receipts]
S4[Run end to end fixture]
S5[Code review and QA]
S1 --> S2
S2 --> S3
S3 --> S4
S4 --> S5
S5 --> D{Evidence verified}
D -- yes --> DONE[Task complete]
D -- no --> S1
Use the approved segment and tested pipeline to create the largest defensible eligible cohort and load it only to paused destinations.
The client needs meaningful reach, but raw volume is unsafe without identity, verification, suppression, personalization, and membership proof.
Campaign timing slips without volume, while rushed uploads create deliverability and client-trust risk.
You receive a sizeable ready-to-review audience with every exclusion, cost, and provider membership accounted for.
Frozen cohort checksums reconcile to paused provider membership, exact spend, exact drop counts, and independently verified zero sends. 80/20. One sender-capacity-sized pilot cohort passes all gates and has exact paused-state readback.
flowchart LR
A[Insufficient safe volume] --> B[Verified paused cohort] --> C[Launch readiness]
flowchart TD
S1[Freeze segment and budget]
S2[Process and quarantine]
S3[Freeze strict ready contacts]
S4[Upload to paused destination]
S5[Read back and QA]
S1 --> S2
S2 --> S3
S3 --> S4
S4 --> S5
S5 --> D{Evidence verified}
D -- yes --> DONE[Task complete]
D -- no --> S1
Connect current and historical copy to available outcomes, validate proof boundaries, and prepare prioritized direct-campaign tests.
Suggestions and anecdotes are mixed together, while reply rate alone can reward messages that do not create qualified opportunities.
New campaigns can repeat weak positioning, use unsupported proof, or optimize for the wrong result.
You get clear keep, change, test, and reject recommendations tied to meaningful downstream outcomes.
A versioned copy inventory, outcome join, claim ledger, sample-size caveats, prioritized tests, and reusable AI handoff are source-linked. 80/20. Identify the best-supported message, weakest pattern, proof boundary, and next two tests.
flowchart LR
A[Mixed copy signals] --> B[Evidence audit] --> C[Better tests]
flowchart TD
S1[Inventory versions]
S2[Join downstream outcomes]
S3[Separate findings and hypotheses]
S4[Validate claims]
S5[Draft tests and handoff]
S1 --> S2
S2 --> S3
S3 --> S4
S4 --> S5
S5 --> D{Evidence verified}
D -- yes --> DONE[Task complete]
D -- no --> S1
Assess the EU partner or reseller opportunity and produce a traceable go, revise, or no-go recommendation with a bounded pilot.
The idea has strategic appeal but no verified partner profile, offer, market, contact role, list source, or economics.
The team may ignore a useful channel or spend time and money scraping a market that cannot support a focused campaign.
You and the client get an evidence-backed feasibility decision and a cheap pilot design before committing resources.
Luna Max research lanes, authenticated Browser and Perplexity work, official Apify comparison, sample validation, costs, risks, and no-go rules are synthesized. 80/20. Validate one partner archetype in one geography with a small evidence sample.
flowchart LR
A[Vague partner idea] --> B[Evidence and sample] --> C[Pilot decision]
flowchart TD
S1[Dispatch Luna Max lanes]
S2[Run Browser and Perplexity research]
S3[Compare Apify and current paths]
S4[Validate sample]
S5[Synthesize pilot gates]
S1 --> S2
S2 --> S3
S3 --> S4
S4 --> S5
S5 --> D{Evidence verified}
D -- yes --> DONE[Task complete]
D -- no --> S1
Package verified work, operating instructions, the executive report, recommendations, next approvals, and an unsent meeting-request email.
Technical progress does not create client confidence when evidence, ownership, and next decisions are scattered.
The client may miss the value delivered, the next campaign can stall, and unsafe reruns can follow from incomplete handoff.
You get one management and operating package plus a ready Hungarian email and meeting agenda.
A QA-checked handoff index links verified artifacts, labels completed and gated work, includes restart instructions, and supplies the unsent email and agenda. 80/20. Executive deck, decision summary, next approvals, email, and meeting agenda in one package.
flowchart LR
A[Scattered deliverables] --> B[Verified handoff] --> C[Confident next meeting]
flowchart TD
S1[Collect receipts]
S2[Build handoff index]
S3[Add report and campaign state]
S4[Finalize email and agenda]
S5[Check links and claims]
S1 --> S2
S2 --> S3
S3 --> S4
S4 --> S5
S5 --> D{Evidence verified}
D -- yes --> DONE[Task complete]
D -- no --> S1