Independent-AI-Tasks

Background

The development and ideas related to the new business myBetterRates (mBR), is out of control. The mBR Obsidian KB, at d:\FSS\KB\MBR\ contains some of the ideas. More are in Dynalist, which I use for idea capture.
- Review these, to get some context for this business. Note that all of these are stale to some degree.
  - D:\FSS\KB\MBR\_WorkingOn\Projects\MBR-Business-Dev\ROADMAP.md
  - D:\FSS\KB\MBR\02_Strategy\Executive-Summary.md
  - D:\FSS\KB\MBR\02_Strategy\Strategic Plan.md
A recent effort to cleanup these ideas, in D:\FSS\KB\MBR\_WorkingOn\Tasks\mbr-cleanup.md, is itself incomplete, and more brainstorming ideas have been added.
I have AI capacity that will expire soon, that I would like to take advantage of, but only if it doesn’t create more overwhelm and a bigger mess that needs to be cleaned up later.
A few of the highest-level mBR tasks include
- Websites for MBR and Cancer50Pledge
- Data infrastructure that addresses all components of the business needs
  - Integration with CRM, accounting, email, marketing, A/B testing, and other workflows
    - Need to specify existing Tech Stack decisions for each component
  - Foundationally important
  - Difficult to get this right without a clear context of all known system needs now, and probable in future
- Rate scanner development, for all products, world class robustness.
- Referral system mechanics, automated
- Clean up of all ideas related to MBR, organized in a single source of truth, for both KB context, and project and task management
- Foundational business docs, strategy
  - Mission, vision, ideal client profile, business models, KSFs, strengths weaknesses opportunities threats, etc
- Creation of AI agent team for higher level AI productivity
- Better organization of operating system for managing two or more businesses, including shared knowledge and processes
  - D:\FSS\KB\MBR_WorkingOn\Tasks\myMBR-OS.md, and other docs/notes/ideas on this concept
  - integrated with AI Agent team (OpenClaw or better)

Tasks

Identify any tasks that can be done by AI that are clean and independent, that do not create more of a disorganized mess to be cleaned up later.

Claude Response — 2026-04-22

Summary

Reviewed ROADMAP.md, Executive-Summary.md, Strategic Plan.md, and all task files in Tasks/
Assessed each task for: independence (no mid-task Talbot input required), cleanliness (well-scoped output, doesn’t create new mess), and Phase A→B criticality
Produced a tiered shortlist of AI-executable tasks with rationale

Assessment Method

What disqualifies a task:

Requires Dynalist ideas before it can proceed (content you haven’t shared yet)
Creates strategy/naming docs that might conflict with existing SSoT files
Depends on open Talbot decisions that haven’t been made (e.g., CanadaHelps API availability, Supabase region)
Produces partial implementations that leave the codebase half-broken
Requires creative/personal input only you can provide (e.g., your voice in the origin story)

What qualifies:

Full spec already written — AI knows exactly what to build
Output is a concrete artifact (working code, deployed site, KB doc)
No blocking decisions mid-task
Moves Phase A→B forward directly

Tier 1 — Fully Independent, Execute Immediately

These tasks have full specs, no blockers, and produce clean deployable artifacts. Zero Talbot input needed during execution.

1. Hassle Threshold Engine MVP (2026-04-08_Hassle-Threshold-Engine.md)

What: Build the core MBR product differentiator — Python CLI that reads rate_changes from SQLite, evaluates user-defined thresholds (e.g., “alert me if net benefit > $150/yr”), and produces alert payloads
Why it’s clean: Full spec with data model, algorithm, tests, and file structure. Follows the same uv + Typer + Pydantic + pytest pattern already established in rate-scanner. Committed to git on completion.
Output: /home/ta/projects/mbr/hassle-engine/ — passing tests + CLI + committed
Status: Unstarted — no Claude Response yet
Phase impact: Phase A→B critical path — this is the product

2. mybetterrates.com Website MVP (2026-04-08_MBR-Website-MVP-Brief.md)

What: SvelteKit static site at /home/ta/projects/mbr/mbr.com/ — rate table, embedded Benefit Snapshot widget, Cancer50Pledge story, Tier 2 waitlist signup, deployed to Cloudflare Pages
Why it’s clean: Full spec, architecture already decided (SvelteKit in mbr/ repo), Brand.md available for colors/typography, rates.json already exported daily, cancer50pledge.pages.dev already live to link to. No backend, no auth, no user data — pure static.
Output: mybetterrates.com live on Cloudflare Pages
Status: Unstarted — no Claude Response yet
Phase impact: Phase A→B critical path — this is the public face of MBR

3. Ops Dashboard → Cloudflare Pages (from 2026-04-08_Rate-Scanner-Layer1-Hardening.md)

What: Deploy the already-built ops dashboard (/home/ta/projects/mbr/ops-dashboard/index.html) to Cloudflare Pages so it’s accessible from any device, not just locally on WSL
Why it’s clean: Dashboard already exists and works (verified in prior session). One action: create new Cloudflare Pages project pointing to ops-dashboard/ as build output.
Output: Dashboard live at mbr-ops.pages.dev or similar
Status: Talbot asked for this; Claude asked for confirmation — confirmation can be assumed given the explicit request

Tier 2 — Mostly Independent, Minor Caveat

4. Cancer50Pledge Phase B (2026-04-08_Cancer50Pledge-Phase-B.md)

What: Real pledge form (Cloudflare Worker), Founding Pledgers wall, shareable certificate, Fathom analytics, custom domain activation
Caveat: Brief says “use whichever backend was chosen in Task 3 (MBR website waitlist)” — but Google Sheets is the stated fallback and is a valid choice for Phase A→B. If default is set to Google Sheets at the start, this task proceeds fully independently.
Output: cancer50pledge.ca live with real backend
Status: Unstarted — no Claude Response yet

5. Mortgage Fixed Rates — Add to rates.yaml (from 2026-04-08_Rate-Scanner-Layer1-Hardening.md)

What: Research current Big 6 fixed mortgage rates (2yr, 3yr, 5yr, insured vs. uninsured) and add product rows to rates.yaml. Update ops dashboard to display them.
Caveat: Mortgage rates change frequently — data will need periodic refresh. Each product should be tagged needs_verification: true.
Output: Updated rates.yaml + committed to git
Status: Talbot explicitly requested this in the last Talbot Response

6. Referral Fee Mechanics → KB Document

What: The comprehensive AI Studio referral fee report is already embedded in Tasks/Referral-Fee-Mechanics.md. The AI task is: extract the report, save it as a proper KB artifact at D:\FSS\KB\MBR\03_Processes\Referral-Fee-Mechanics.md, update cross-references.
Why it’s clean: The research is already done. This is purely a file organization task.
Output: Properly filed KB document, task file updated/closed
Status: Content already written, just needs to be properly filed

Tier 3 — Needs Your Input First (Not Recommended for AI-Only Sprint)

These tasks have real value but require Talbot’s judgment or content before AI can execute cleanly:

Task	Blocker
Origin Story SSOT	Your personal voice, updated story beats, and review of upgraded copy — this is the foundation of the brand narrative and shouldn’t be AI-generated without your active involvement
Filter+Focus Refresh	Requires your current thinking on priorities; AI refresh without Dynalist context would be stale
CRM — Resolve Open Questions	Requires your decisions: CanadaHelps API, Supabase region, QBO Stripe Sync availability
myMBR-OS	Architectural standards decisions — needs your input on what “done” looks like
mbr-cleanup loose ends	Blocked on Dynalist content you haven’t shared yet

Recommended Execution Order

If you want to use AI capacity productively right now, without creating mess:

Hassle Threshold Engine MVP — /task-start 2026-04-08_Hassle-Threshold-Engine.md
mybetterrates.com Website MVP — /task-start 2026-04-08_MBR-Website-MVP-Brief.md
Cancer50Pledge Phase B (with Google Sheets as default backend) — /task-start 2026-04-08_Cancer50Pledge-Phase-B.md
Ops Dashboard deploy + Mortgage fixed rates — both small enough to bundle into one session on the Rate Scanner task (/task-continue 2026-04-08_Rate-Scanner-Layer1-Hardening.md)
Referral Fee Mechanics filing — smallest task, can be done anytime

Tasks 1 and 2 are the highest-leverage use of AI capacity — they’re Phase A→B critical path items, fully specced, and will produce the two visible artifacts that define MBR’s public presence.

Next Steps for Talbot

Confirm which task(s) to run first — Tier 1 items can be started immediately 1.
For Cancer50Pledge Phase B: confirm Google Sheets as the waitlist/pledge backend (or specify alternative) 1.
For Ops Dashboard deployment: confirm you want it live on Cloudflare Pages (vs. staying local) 1.

Talbot Response

Tasks 1, 2, and 4 are good candidates. Use the /task-prep skill, if helpful, to review the task files to ensure that all appropriate context is available, especially for the website in task 2, as I will deploy this to Cursor.
Then I will deploy each.

Claude Response — 2026-04-22

Summary

Task 1 (Hassle Threshold Engine) — already complete, flagged below
Task 2 (mybetterrates.com Site) — prepped with locked context, status: ready
Task 4 (Cancer50Pledge Phase B) — prepped with locked context, status: ready

Task 1 — Already Built

The Hassle Threshold Engine was completed in a prior session (2026-04-09). Evidence in the task file:

/home/ta/projects/mbr/hassle-engine/ exists with full project structure
13/13 tests passing
hassle-engine evaluate, status, and --dry-run all working
Net-benefit calculation manually verified

This task does not need to be started — it is done. Recommend running /task-complete 2026-04-08_Hassle-Threshold-Engine to close it properly.

Task 2 — mybetterrates.com Site (status: ready)

Key locked decisions embedded in the task file (prior sessions resolved all architecture debates):

Astro site in monorepo/sites/mbr/ — not SvelteKit, not mbr/mbr.com/
Cloudflare D1 for waitlist backend
Email-first alert delivery (no PWA)
MyBetterRatesEntity.astro component already exists in template — use it
Best chequable rate as of 2026-04-20: EQ Bank Personal Account 2.75% (not 3.65% — that’s a GIC)

Context block embedded at end of task file with: locked decisions table, resolved file paths, brand colors, Astro monorepo pattern, D1 setup snippet, execution constraints.

Ready for Cursor. Open the file at D:\FSS\KB\MBR\_WorkingOn\Tasks\2026-04-08_MBR-Website-MVP-Brief.md and run /task-start.

Task 4 — Cancer50Pledge Phase B (status: ready)

Key updates embedded in context block:

Backend = Cloudflare D1 (not Google Sheets/Airtable as brief states)
Tailwind v4 is already installed — do NOT downgrade to v3
Cancer50Pledge domain (cancer50pledge.ca) DNS = Talbot’s job; AI creates DEPLOY.md with steps only
Fathom PUBLIC_FATHOM_SITE_ID needed from Talbot’s account — site builds without it
Founding pledgers stub = empty array; Talbot populates manually

Ready for Cursor. Open D:\FSS\KB\MBR\_WorkingOn\Tasks\2026-04-08_Cancer50Pledge-Phase-B.md and run /task-start.

One Thing to Provide Before Starting

Before running Cancer50Pledge Phase B: if you want Fathom analytics live at launch, provide your Fathom Site ID (PUBLIC_FATHOM_SITE_ID). The site will build and run without it, but analytics won’t track until it’s set.

Next Steps for Talbot

Run /task-complete 2026-04-08_Hassle-Threshold-Engine to close the already-done task 1.
Open website task in Cursor and run /task-start: D:\FSS\KB\MBR\_WorkingOn\Tasks\2026-04-08_MBR-Website-MVP-Brief.md
1. A brief review of this task revealed that a key architectural decision was required. The entire purpose creating the Astro monorepo was to establish a foundation to be used for all business websites, especially those that are mostly static content. Thus unless there is a very strong reason To deviate, the MBR and cancer 50 pledge websites should use this mono repo. Interactivity can be done with the established frameworks, including spelled components when alpine is insufficient. Additionally, Astro sites deploy very easily to Cloudflare.
Open Cancer50Pledge task in Cursor and run /task-start: D:\FSS\KB\MBR\_WorkingOn\Tasks\2026-04-08_Cancer50Pledge-Phase-B.md 1.
Provide Fathom Site ID for Cancer50Pledge analytics (optional — site builds without it) 1.

Talbot Response

Modify the brief for the MBR website, so it can be deployed to Cursor.
Note that you confused tasks 3 and 4. The cancer 50 pledge was task 3. Deploying the ops dashboard to Cloudflare was task 4. Ensure that both of these tasks are ready to be started and/or continued, without further involvement from me at this time.

Claude Response — 2026-04-22

Summary

MBR website brief rewritten — replaced the outdated SvelteKit/mbr/mbr.com/ content with a clean Astro/monorepo spec; all architecture decisions locked at the top, no re-debating required
Task numbering corrected — Cancer50Pledge = task 3, Ops Dashboard deployment = task 4
Cancer50Pledge Phase B — already prepped (status: ready) from prior session; context block at bottom overrides outdated Google Sheets backend with Cloudflare D1
Ops Dashboard deployment — context block added to 2026-04-08_Rate-Scanner-Layer1-Hardening.md; exact deploy command and daily-run.sh integration steps embedded

Three Tasks Ready for Cursor — No Further Input Required

#	Task	File	Action
2	myBetterRates Site	`2026-04-08_MBR-Website-MVP-Brief.md`	`/task-start`
3	Cancer50Pledge Phase B	`2026-04-08_Cancer50Pledge-Phase-B.md`	`/task-start`
4	Ops Dashboard → Cloudflare Pages	`2026-04-08_Rate-Scanner-Layer1-Hardening.md`	`/task-continue`

What’s in Each Context Block

MBR Website (rewritten brief):

Framework = Astro, location = monorepo/sites/mbr/, waitlist = Cloudflare D1
Brand colors, wordmark component path, rates.json current top product (EQ Bank 2.75% chequable)
D1 setup commands, wrangler.toml snippet, deploy command
Rate table testing note: best chequable = 2.75%, not 3.65% (GIC)

Cancer50Pledge Phase B:

Backend = Cloudflare D1 (overrides “Google Sheets/Airtable” in original brief)
Tailwind v4 already installed — don’t downgrade
Fathom site ID needed from Talbot; site builds without it
Domain DNS = Talbot’s job; AI writes DEPLOY.md only

Ops Dashboard deployment (Rate Scanner task context):

wrangler pages deploy /home/ta/projects/mbr/ops-dashboard/ --project-name mbr-ops-dashboard
Add deploy as Step 6 in daily-run.sh
Verify URL loads with current data

Next Steps for Talbot

Start task 2 in Cursor: /task-start on 2026-04-08_MBR-Website-MVP-Brief.md
1. started in Cursor
Start task 3 in Cursor: /task-start on 2026-04-08_Cancer50Pledge-Phase-B.md 1.
- Continue task 4 in Cursor: /task-continue on 2026-04-08_Rate-Scanner-Layer1-Hardening.md
  1. continued in CC
Provide Fathom Site ID for Cancer50Pledge when ready (optional — builds without it) 1.