Proposal · April 2026

A plan for Rachel's agency

Automation and custom tooling to take the weight off the employee side of the business, while protecting the personal touch that makes the agency work.

FromAyman Hajja · NovaBrains LLC

ForGabi & Rachel

Follow-upMonday, April 27

What I heard in our meeting

Before jumping to ideas, here's my understanding of the business so you can correct me if I've misread anything.

Rachel runs a high-touch domestic staffing agency (~1,620 candidates in the pool) placing nannies, housekeepers, and similar roles.
The differentiator is a personal, human connection, especially with employers, who treat their homes as sacred and won't tolerate "robotic" service. Word-of-mouth drives most of the pipeline, so reputation is load-bearing.
Operations today live across Google Drive, Google Contacts, Google Sheets, and a physical notebook, with zero automation between them. Rachel and one VA handle everything.
Rachel flagged automated follow-ups with existing employees as a major pain point. Candidates who registered months ago call for updates, and she has to manually dig through the Drive to respond.

Strategy

Automate the employee side aggressively. Keep the employer side human by default, with selective augmentation.

The ideas, organized

Grouped by area, with a clear read on which I recommend vs. which I'm throwing out to spark discussion. These apply regardless of which of the two implementation paths we choose (see Two paths forward below).

Replace the paper-based application with a digital one on a proper website.

A clean, simple application form candidates fill out online.
Payment integration (Stripe) for the $75 application fee, flowing directly to a business bank account.
Because the personal brand matters, the site itself can emphasize Rachel, not a faceless agency. When Gabi or Rachel shares the link (or when word-of-mouth referrals visit), the personal touch is communicated before the first call.
After payment, the candidate is auto-scheduled for a video call with Rachel or the VA (Calendly-style). No manual scheduling back-and-forth.

Technically possible: a voice agent could conduct the candidate interview and collect all the info. Accuracy is good.

But — based on how strongly Rachel emphasized meeting every candidate personally, I don't think this fits. Mentioning it only in case it sparks a related idea (e.g., a lighter pre-screen before Rachel's video call). Happy to demo if curious, but not proposing it.

This directly addresses Rachel's primary pain point.

All ~1,620 existing employees live in one system instead of scattered across Google Contacts, Drive, and Sheets.
Every candidate gets a status (new, vetted, candidate, interviewing, trial, hired, etc.).
Automated message sequences triggered off status and time elapsed. These are just examples — we'll agree on the actual content and cadence together:
- After vetting: "You've been added to the pool."
- +2 weeks: check-in message.
- Monthly: "Just letting you know we still have you in the system, reply if anything's changed."
Monthly (or configurable) prompts to update availability / preferences / resume changes so info doesn't go stale.

Net effect: candidates stop calling for updates because the system already told them where they stand.

Don't rip out Google Contacts

My assumption is that Google Contacts matters because it's what makes Rachel's phone show caller names on incoming calls, and that matters for the personal-touch brand — happy to discuss further if there's a different reason. The plan is to make our new system the source of truth and set up a one-way sync back to Google Contacts, so her phone (and the VA's phone) stay as useful as they are today. She never has to maintain Contacts manually again, but incoming calls still show "Maria" instead of "Unknown."

Today employers call Rachel directly, which is fine. A couple of additions:

Optional booking link on the website. Employer picks a slot, joins a Google Meet (or dials in by phone number).
Automatic call notes/transcription so Rachel has a record without extra effort. Her handwritten notebook still works; this augments it.
Notes flow into the data home, tied to the employer record.

The phone-number-join-Meet approach is important: it feels like a normal phone call to the employer but gives us transcripts on our side.

Alternative worth exploring

Keep things exactly as they are, but record the phone calls themselves. Move Rachel's business line onto a service like OpenPhone or Dialpad: she keeps her number, employers call her phone like they do today, but every call is automatically recorded, transcribed, and AI-summarized into the data home. NY is a one-party consent state, so this is legally straightforward — a short recording disclosure at the top of the call is good practice regardless.

When an employer's requirements come in, the system surfaces candidates matching obvious criteria (live-in/live-out, language, role, availability), the filtering part that doesn't need judgment.

The caveat: Rachel's "feel" for who fits is a big part of the value. The system shouldn't pretend to make the match. It should narrow the pool so Rachel only applies judgment to a short list. If over time she's comfortable letting the system do more, we expand it.

This is the one feature that requires a bit of custom work regardless of which path we pick.

Once an employer-side status flips to "hired," Stripe auto-generates and sends the invoice. Stripe handles payment reminders natively, so we're not reinventing that logic. Kills the manual chasing that currently eats time.

Heard Rachel's dislike of AI agents on the phone. I'm not proposing one for business hours.

The pattern some agencies use: enable it only after hours / weekends, with:

Immediate escalation option for urgent calls.
Full transcripts + sentiment visible to Rachel the next morning.
Smart routing: caller's number matches an employer in the data home → forward to Rachel's cell. If employee → agent handles, escalates only if needed.

Easy to turn off if she hates it after a week. On the table, not pushed.

Two paths forward

Here's the honest part. There are two reasonable ways to build this, and I want to lay out both fairly so you can pick the one that fits the business and your appetite for risk.

Data home

Airtable / ClickUp / Go High Level

Where the 1,620 employees and employer records live. I'm actively researching which of these three is the best fit for this niche — each has tradeoffs around relational data, custom UI, ease-of-use, and built-in automation — and I'll land on a recommendation before we start. All three are mature, well-funded platforms with strong ecosystems. Your data stays yours (CSV export anytime, no lock-in), and Rachel works in a clean custom interface built on top so she never sees a raw database grid.

Automations

n8n

Follow-up sequences, monthly check-ins, invoicing triggers, syncs (including Google Contacts sync). Every automation writes back to a "Communications Log" in the data home so Rachel can see exactly what went out and when.

Custom tools

Only where needed

The candidate-matching feature (#5) and the public-facing website + digital application (#1) are the pieces that genuinely need custom code. Built small and focused, reading/writing to the data home via its API.

Why I'm recommending this one

Lower dependency on me. All three platform candidates have huge communities. Any freelancer can pick up the automations. The custom pieces are small and scoped.
Easier exit story. Your data lives in a well-known platform. Worst case, you export it and walk.
Costs are transparent. Platform + automation tooling together run roughly $40–120/month depending on which platform we pick. Precise number once we lock in the platform.

The honest tradeoffs

Platform UIs are good but not infinite. Rachel might hit a UI ceiling in month 2 or 3 and want something more tailored. If/when that happens, we build a custom frontend for that specific screen. Budget for this possibility.
You're dependent on whichever platform we pick. All three options I'm considering are mature companies, but platform dependency is still a dependency.

A small internal web app: employee records, employer records, statuses, follow-ups, matching, invoicing, all in one UI with no seams.

Why this is attractive

Perfect fit. Every screen designed for Rachel, not adapted from a generic tool.
Single tool to log into. No switching between Airtable, n8n, etc.
No platform fees. Just hosting (~$20-50/month) and third-party services (Stripe, Twilio).
Cleaner long-term. Everything's in one codebase.

The honest tradeoffs

Higher dependency on a developer. A custom codebase needs a developer to make changes — a standard consideration for any custom build. Day-to-day maintenance itself is minimal. One reason this option is feasible for a small business like this at all: most of the build is now AI-assisted ("vibe-coding"), which has made full custom viable for operations that couldn't have justified it a year or two ago. More small businesses are adopting this route because of it.

How I'd mitigate the dependency concern

You own everything from day one. GitHub repo under your account. Database under your billing. I have access as a contractor, not owner.
Boring, popular stack. Next.js + Postgres + standard auth. Any mid-level developer can pick this up.
Documentation as a deliverable. README, architecture doc, handoff video. Another dev can clone the repo and run it locally in under an hour.
Not a single point of failure. Bus-factor covered: codebase in your GitHub, data in your database, docs in the repo.

Side-by-side

	Option A (Hybrid)	Option B (Full Custom)
Time to Phase 1	~2 weeks	~2 weeks
Ongoing platform cost	~$40–120/mo	~$20–50/mo
Dependency on Ayman	Low (any freelancer can pick up)	Moderate (need a developer)
UI polish	Good (limited by platform's UI builder)	Excellent (fully tailored)
Flexibility to change later	High	Highest
Rachel's learning curve	Small (mature, intuitive platform)	Small (tool designed for her)

My recommendation

Option A. For this business, the reduced dependency and platform stability outweigh the UI ceiling tradeoff. And if Rachel ever outgrows the chosen platform, we can evolve into Option B later — the data migrates cleanly because it was always structured.

That said, Option B is completely reasonable if you'd rather have one unified tool and are comfortable with the dependency structure above. Happy to go either direction.

Proposed phasing

Same phasing works for both options. The difference is what's under the hood.

Phase 1

~2 weeks

Data home + follow-up automation

Stand up the data home (on the chosen platform for Option A, or as a custom app for Option B).
Migrate all ~1,620 employees from Google Contacts / Drive / Sheets.
Set up one-way sync back to Google Contacts so phones keep showing caller names.
Build the day-to-day UI Rachel will actually use.
Agree on specific message sequences (text, email, cadence) and wire them up.
Rachel-only access.

Heads-up on one dependency

A2P 10DLC registration (required for any business sending automated SMS in the US) takes 2–3 weeks through Twilio or similar and is outside our control. We'd start this on day one so it doesn't block the texting piece. Email sequences can go out immediately.

Fair warning on migration

The existing data lives in three or four places with inconsistent formatting, so cleaning it up is itself a small project. I'll budget for that honestly rather than promising a pristine pool on day one.

Phase 2

Employer side + matching + invoicing

Employer records + status tracking.
Candidate-surfacing UI for new employer requirements.
Stripe invoicing tied to hired status.

Phase 3

Website, digital application, booking

Public-facing site with Rachel's personal brand.
Digital application + Stripe application fee.
Calendly-style auto-scheduling after payment.

Phase 4+

Optional

After-hours voice agent + ongoing automations

As the business grows and new ideas emerge.

Happy to reshuffle this — let's talk through it and decide together.

Commercials

Phase 1 timing

~2 weeks from the moment we agree on scope.

Working with Rachel

2–3 half-hour sessions during the first couple of weeks to walk through the system and fine-tune. Not overhead — that's true regardless of which option.

Maintenance retainer (optional)

A small monthly retainer — low-hundreds range for ~5 hours/month. Technically optional, but I've found most clients prefer it for peace of mind — someone on call without thinking about it. Cancelable any time.

Ownership and exit

Option A

Platform workspace + n8n instance under your billing. Your data, your accounts. Walking away means canceling two subscriptions and exporting your data.

Option B

GitHub repo, database, and all credentials under your accounts from day one. Written exit clause: code + docs + handoff within 5 business days if you ever want out.

I'll give you a precise build number once we agree on Phase 1 scope and which path you'd like to go.

Next steps

1

Review this and push back on anything that doesn't fit.
2

Pick a path

Option A is my recommendation, but I want you comfortable with whichever we choose. Ask me to help pick if useful.
3

Access to Google Drive & Contacts

So I can start looking through the existing data, understand its shape, and prep for migration. I'll likely have follow-up questions once I've had a chance to look through it.
4

Sync this week if helpful

Happy to meet Thursday or Friday if you'd like to talk through this before the Monday 4/27 meeting. Otherwise we lock scope and approach then.
5

Once scope is locked, we begin the build.

Proposal prepared by Ayman Hajja novabrains.com

Proposal · April 2026

A plan for Rachel's agency

Automation & custom tooling

Ayman Hajja

NovaBrains LLC · For Gabi & Rachel

What I heard

The business, in four lines

High-touch domestic staffing · ~1,620 candidates in the pool
Personal connection is the differentiator · word-of-mouth drives the pipeline
Data lives across Drive, Contacts, Sheets, and a notebook — zero automation
Automated employee follow-ups is Rachel's major pain point today

The strategy

Automate the employee side aggressively.
Keep the employer side human.

Protects the personal touch that built the business. Offloads the work that doesn't need Rachel's judgment.

Seven ideas

At a glance

1Digital application + paymentRecommended

2Automated candidate interviewNot recommending

3Data home + automated follow-upsPhase 1

4Employer intake via websiteRecommended

5Candidate matching / surfacingRecommended

6Automated invoicingRecommended

7After-hours voice agentOptional

1 Digital application + payment

A website that does the paperwork

Digital application form replaces paper
Stripe integration for the $75 fee · straight to the bank
Rachel's personal brand front-and-center on the site
Auto-schedule the video interview after payment

3 Data home · Phase 1

Migrate 1,620 candidates. Automate the follow-ups.

One system for every employee · statuses, notes, history
Auto-messages triggered by status + time (week 2, monthly, etc.)
Periodic prompts to keep availability & preferences fresh
Candidates stop calling for updates — the system already told them

Don't rip out Google Contacts. One-way sync keeps Rachel's phone showing "Maria" instead of "Unknown."

4 Employer intake

Notes and transcripts, without extra effort

Path A · Google Meet

Booking link on the website. Employer joins via video or phone dial-in. Transcripts flow into the data home under the employer's record.

Path B · Record the phone line itself

Move Rachel's business line to OpenPhone or Dialpad. Keep the number, keep the experience — every call recorded, transcribed, and summarized automatically.

NY is one-party consent · a short recording disclosure is good practice either way.

Ideas 5 · 6 · 7

The rest, in one view

Matching & surfacing

System narrows the pool by obvious criteria. Rachel's judgment does the rest. (This one requires custom work regardless of path.)

Automated invoicing

Status flips to "hired" · Stripe sends the invoice and handles reminders.

After-hours voice agent

Only outside business hours. Smart routing. Full transcripts. Easy to turn off.

The honest part

Two paths forward.

Two reasonable ways to build this. Laying them out fairly so you can pick what fits the business and your appetite for risk.

Option A Recommended

Hybrid: proven platform + automation + targeted custom

Data home Airtable / ClickUp / Go High Level

Where the 1,620 employees live. Still researching which of the three fits this niche best — final pick before we start. Rachel works in a clean custom interface built on top.

Automations n8n

Follow-ups, sync to Google Contacts, invoicing triggers. Full communications log in the data home so Rachel sees what went out and when.

Custom Only where needed

Matching logic + public website/application. Small and focused. Reads/writes to the data home via API.

Option B Alternative

Full custom build · one integrated tool

Why it's attractive

Perfect fit — every screen designed for Rachel
Single tool to log into
No platform fees, just hosting + services
Cleaner long-term · one codebase

The mitigations

You own everything from day one (GitHub, DB, credentials)
Boring, popular stack · Next.js + Postgres
Documentation is a deliverable
Written 5-business-day exit clause + handoff call

Tradeoff: higher dependency on a developer — a standard consideration for any custom build.

Side-by-side

How they compare

	Option A (Hybrid)	Option B (Custom)
Time to Phase 1	~2 weeks	~2 weeks
Platform cost	~$40–120/mo	~$20–50/mo
Dependency on Ayman	Low	Moderate
UI polish	Good	Excellent
Rachel's learning curve	Small	Small

Recommendation: Option A. Reduced dependency + platform stability outweigh the UI ceiling. Can evolve into B later if Rachel outgrows the platform — data migrates cleanly.

Phase 1

Data home + follow-up automation

Stand up the data home · migrate all ~1,620 employees
One-way sync back to Google Contacts (phones keep working)
Build the day-to-day UI · wire up message sequences
~2 weeks, regardless of which option

Two honest caveats:
· A2P 10DLC registration takes 2–3 weeks through Twilio — start day one so it doesn't block SMS.
· Data cleanup is itself a small project. Data lives in 3–4 places with inconsistent formatting.

Roadmap

Four phases · reshufflable

Phase 1

Data home + automations

The pain-point fix. Migration + follow-ups.

Phase 2

Employers + matching + invoicing

Employer records, candidate surfacing, Stripe invoicing.

Phase 3

Website + digital application

Public site. Stripe fee. Auto-scheduling.

Phase 4+

After-hours agent + new automations

As the business grows.

Commercials

How the money works

Maintenance (optional)

Small monthly retainer · low-hundreds for ~5 hrs/month · cancelable any time. Technically optional, but most clients prefer it for peace of mind.

Ownership

Option A: Platform + n8n under your billing. Option B: GitHub + DB + credentials under your accounts from day one.

Exit

Option A: cancel two subscriptions. Option B: 5-business-day handoff clause + one-hour handoff call.

Next steps

Five moves

Review & push back on anything that doesn't fit.
Pick a path · Option A is my recommendation; either is fine.
Drive & Contacts access · so I can start looking at the data.
Optional sync Thu / Fri · before the Monday 4/27 meeting.
Scope locked → we begin the build.

Proposal prepared by Ayman Hajja

novabrains.com

1 / 16

Proposal · v1 · April 24, 2026

One system Rachel lives in

A custom internal CRM that becomes the home for every part of the agency — employees, employers, calls, communications, dashboards. Built to be the foundation Rachel runs the business from, and malleable enough to grow into new sectors later.

FromAyman Hajja · NovaBrains LLC

ForGabi & Rachel

Refined afterApr 23 discovery call (Pt. II)

What's different in v1

Three clear decisions came out of the second discovery call. v1 reorganizes the plan around them.

1

Focus on the data home — idea #3 from v0

This is the centerpiece. It's Rachel's #1 daily pain (chasing follow-ups across Drive, Sheets, Contacts, the notebook), the foundation everything else snaps onto, and the call's strongest agreement. Phase 1 of the build is exactly this.
2

Public website + digital application — idea #1 from v0 — on a parallel track

Separate work stream so it doesn't gate the CRM and the CRM doesn't gate it. Brand presence, digital application form, $75 Stripe fee, auto-scheduling. Submissions land in the data home as new candidate records.
3

Build it as a custom internal system, not a hybrid platform stack

v0 leaned on Airtable / n8n / custom. v1 commits to a single custom build. Reasons below.

Why custom over hybrid

The agency's workflows are too specific to bend into a generic CRM. Trying to live inside one would re-create the seams Rachel already feels between Drive, Sheets, Contacts, Voice, Canva, and the notebook.
Rachel's stated goal — expanding into medical offices and other professional sectors — needs a malleable base. A custom platform lets us add a new vertical without ripping out the old one.
Rachel decides what's automated and what stays manual — per task, per client. Auto-invoice this employer, manually negotiate the next. Auto-text new candidates, hand-write the tricky ones. A custom build makes that fine-grained control natural; off-the-shelf platforms tend to be all-or-nothing.

The vision

A simple test: Rachel opens the laptop in the morning, logs into one URL, and everything she needs to run the agency that day is in front of her. No Drive tab. No Sheets tab. No Contacts. No Canva. No notebook on the side.

→

Every employee & employer in one place

All ~1,620 candidates, every employer, every data point — searchable, filterable, editable from one screen.

→

Every conversation captured

Email threads, text messages, WhatsApp, phone calls — recorded, transcribed, attached to the right person automatically.

→

Click-to-call — and inbound calls too

Click an employee → call from the browser. Inbound calls land in the system, matched to the right record. Recording + transcript saves itself either way.

→

Cards generate themselves

The polished candidate cards Rachel currently builds in Canva are auto-generated from the record. Update a preference, the card regenerates. No more Canva.

→

Calendar & Gmail live inside

Google Calendar and Gmail piped in: see what's scheduled, draft replies, send messages — without leaving the system.

→

A morning dashboard

Today's calls. New applications. Candidates due for follow-up. Open employer requests. One screen. Rachel knows what to do.

→

VA-ready

Multi-user with roles. Rachel's VA — and any future hire — can step in without a week of onboarding. Secure: only the people Rachel approves get in.

→

Exports, always

Basic CSV export on every record type from Phase 1 — your data stays yours and leaves cleanly any time. Imports and richer export options (column picking, filtered exports, scheduled drops) come in later phases.

The phased plan

Phases are guidelines, not contracts. We meet ~twice a week and reshuffle as we learn. Each phase invoiced on completion. Rough timelines below assume we start within a week of locking scope.

Rachel and Gabi have signaled real urgency on this — "ready to do it yesterday." If you want, we can compress the whole plan to roughly half its standard length. Same scope, same quality bar; the only thing that changes is the pace and how much of my week is on this.

Phase 1

~~~2 weeks~~ → ~1 week

Each subsequent phase

Half the standard duration

Meetings with Rachel

~~~2 / week~~ → 3–4 / week

My time on this

Heavier — I'm on it intensively, not splitting attention

Standard pace is the default in this proposal. Just say the word and we run the accelerated track instead.

Phase 1

~2 weeks

Foundation + data home

The shell of the system, with all the data already in it. The thing Rachel can start using on day one of week three.

Authentication, roles, secure access. Rachel + VA accounts ready.
Employee and employer record models — every data point captured (preferences, references, status, history, documents).
Migration scripts pull from Drive, Sheets, and Contacts. Ayman runs the migration. Migration is complete by end of week 2 — included in Phase 1, not a separate stream.
Search, filter, bulk update — fast. The kind of search that surfaces "live-in housekeeper, Spanish-speaking, available May" in one query.
Basic CSV export on every record type. Imports and richer export options come in subsequent phases.
Basic morning dashboard (what's new, what's pending, what needs follow-up).
Foundation laid for Phase 2 messaging — record-level email/SMS fields, provider integrations stubbed in, message-log table ready. Actual automated sends start in Phase 2.

Started day one

A2P 10DLC registration (required for any business sending automated SMS in the US) takes 2–3 weeks. We file day one so SMS is unblocked by the time we're ready. Email and WhatsApp (post-verification) can carry the load until then.

Honest note on migration

The current data lives in 3–4 places with inconsistent formatting. Cleaning it as we go is part of Phase 1, not a footnote. We'll surface anomalies for Rachel to triage rather than guess.

Phase 2

~2 weeks

Communications + auto-generated cards

Everything Rachel sends out — and the polished outputs employers see — lives inside the system.

Email automations live from day one of Phase 2 (no carrier registration required). SMS goes live as soon as A2P 10DLC clears (filed at start of Phase 1, typically ready by now). WhatsApp follows post-verification.
All channels sent from inside the CRM, threaded onto each record.
Workflow automations Rachel can override per record: "When status flips to Active → text the candidate." Easy toggle to skip on records that need the personal touch. Examples (we'll agree on real cadences):
- Status → Active: welcome SMS / email.
- +1 week: "We have everything we need from you for now."
- Monthly: "Anything changed? Reply to update."
Auto-generated candidate cards replace Canva. The card is a live render of the record — change a preference, the card updates. Send to employer via email or WhatsApp in two clicks.
Communications log on every record — Rachel can see exactly what went out, when, and what came back.

Phase 3

~2 weeks

Calls — make them, record them, extract from them

The phone moves into the CRM. Calls become structured data instead of disappearing into the air.

Click-to-call from any record, right in the browser. Outbound calls go through Rachel's business number.
Inbound calls land in the system, matched to the right employee or employer record automatically.
Optional recording + transcription on inbound and outbound. Short legal disclosure at the top of the call (NY is one-party consent — straightforward).
Auto-extraction from transcripts: AI pulls the key facts out of the conversation and drops them into the right fields on the record. Rachel reviews and approves, doesn't re-type.
Employer intake supports both modes: a regular phone call (recorded via the platform) or a Google Meet (also recorded). Rachel picks per client.
One-way sync back to Google Contacts so phones still show caller names (incoming calls show "Maria" instead of "Unknown"). Lives here since it's tied to the call experience.

Phase 4

~2 weeks

Matching engine and other integrations

Once the data is clean and the records are rich, matching becomes worth doing well — and Rachel's existing Google tools snap into the same screen.

Note: basic filtering by an employer's hard requirements (live-in/live-out, language, role, hours, location radius) is already available from Phase 1 — it's part of the search/filter Rachel uses every day. Phase 4 is where we sharpen it: dedicated matching UI, ranked shortlists, and smarter logic if needed (e.g. weighting preferences, surfacing edge cases, learning from past placements). Only as deep as it earns its keep.
Surface the top N candidates with a clear "why this one" trail.
Rachel's judgment still makes the call. The system narrows the pool — it doesn't pretend to know who fits a family.
Iterative: we'll meet to define the matching logic together and tune over the first few real placements.
Calendar integration — Google Calendar synced into the CRM (and back out), so scheduling lives in one place.
Gmail integration — read and reply to threads tied to each record, without leaving the system.

Phase 5+

Ongoing

Dashboards, invoicing, and whatever comes up

Once the foundation is real and the work has shape, we add what proves useful.

Richer dashboards — placement velocity, follow-up health, revenue, anything Rachel wants visibility on.
Stripe-based auto-invoicing tied to "hired" status — with a per-client override so Rachel can edit, cancel, or hand-negotiate the cases that need it (chargebacks, partial fees, employees who leave early). For clients who'd rather skip credit cards, we can also surface Zelle / bank transfer / debit as an option on the invoice — not the default, just available where it helps avoid chargeback risk.
Guided interviews — a structured walkthrough of Rachel's existing 5–6 question employer intake script, captured field-by-field as the call progresses (instead of relying purely on transcript extraction). Same idea can apply to candidate intake later.
After-hours voice agent (optional, on the table — not pushed).
New verticals (medical, professional services) when ready — same platform, new schemas.

Parallel track — online presence

The public website + digital application is a separate work stream. It doesn't gate the CRM, and the CRM doesn't gate it. Both can be built at once because they're different surfaces.

Track A · Internal

The CRM

What Rachel and the VA use every day. Phases 1–5+ above.

Track B · External

The website

Public-facing site that establishes the brand. Digital application form. Stripe for the application fee (with Zelle / bank / debit available as alternative options for candidates who'd rather skip credit cards — addresses the chargeback concern Rachel raised). Auto-scheduling into Rachel's calendar after payment. Submissions land directly in the CRM as new candidate records.

If bandwidth is tight, the website can lag the CRM by a phase or two — but starting it in parallel costs little and gets Rachel an online presence sooner. Open to either.

How we'll work together

Cadence

Two short meetings per week, especially in the first month. The build is iterative — Rachel sees the system every few days and shapes it as we go.

Rachel uses it early

As soon as Phase 1 is live, Rachel starts using the CRM for new candidates and employers. Existing pipeline migrates in parallel — no big-bang cutover.

Migration

Handled by Ayman. Scripts pull from Drive / Sheets / Contacts. Cleanup pass with Rachel for ambiguous records. No spreadsheet homework for you.

Stack

Boring and popular. Next.js + Postgres + standard auth. Hosted under your accounts. Any mid-level developer on the market can pick it up.

Commercials

Billing

Per phase, invoiced on completion. No upfront. You see Phase 1 working before Phase 1 gets paid.

Phase 1 timing

~2 weeks from the moment we agree on scope.

Maintenance retainer (optional)

Small monthly retainer — low-hundreds range for ~5 hours/month. Most clients prefer it for peace of mind. Cancelable any time. If a given month needs more, we talk.

Ownership & exit

GitHub repo, Postgres database, and all credentials live under your accounts from day one. I have access as a contractor, not as the owner. Written exit clause: code + docs + handoff video within 5 business days if you ever want out.

Running costs

Hosting + DB roughly $20–50/month. Usage-based services (SMS via Twilio, call recording / transcription, email sending) bill at cost — typically tens of dollars/month at this scale, scaling with volume. No platform subscription tax on top.

Precise build number for Phase 1 once scope is locked Monday.

Next steps

1

Read v1 and push back.

Anything that's not what you heard on the call, anything missing, anything you'd reorder — tell me before Monday, or we can discuss it live during Monday's call.
2

Confirm direction.

Custom CRM as the foundation, website as a parallel track, automations Rachel controls per task. If yes, we lock Phase 1 scope on Monday.
3

Drive, Sheets, Contacts access.

So I can shape the migration scripts before we start. I'll have follow-up questions once I see the data.
4

File A2P 10DLC the day we start.

2–3 week external clock. Best to start it the moment we agree on scope so SMS is ready when Phase 2 needs it.
5

Schedule a deeper-dive session on website features.

Pierre and I will bring competitor inspiration (nurse finders, Madison, Nannies R Us, All-in-One, maids.com). This doesn't have to happen before we start Phase 1 — it can run in parallel and inform the website track when we're ready for it.
6

Monday 4/27 — lock scope, start the build.

Proposal v1 prepared by Ayman Hajja novabrains.com

A plan for Rachel's agency

What I heard in our meeting

The ideas, organized

Digital application + payment (website)

Automated candidate interview

The data home + automated follow-ups

Employer intake via website

Candidate matching / surfacing

Automated invoicing

Voice agent for after-hours only

Two paths forward

Hybrid: proven platform + automation layer + targeted custom tooling

Why I'm recommending this one

The honest tradeoffs

Full custom build

Why this is attractive

The honest tradeoffs

How I'd mitigate the dependency concern

Side-by-side

Proposed phasing

Data home + follow-up automation

Employer side + matching + invoicing

Website, digital application, booking

After-hours voice agent + ongoing automations

Commercials

Next steps

Review this and push back on anything that doesn't fit.

Pick a path

Access to Google Drive & Contacts

Sync this week if helpful

Once scope is locked, we begin the build.

A plan for Rachel's agency

The business, in four lines

Automate the employee side aggressively. Keep the employer side human.

At a glance

A website that does the paperwork

Migrate 1,620 candidates. Automate the follow-ups.

Notes and transcripts, without extra effort

Path A · Google Meet

Path B · Record the phone line itself

The rest, in one view

Matching & surfacing

Automated invoicing

After-hours voice agent

Two paths forward.

Hybrid: proven platform + automation + targeted custom

Full custom build · one integrated tool

Why it's attractive

The mitigations

How they compare

Data home + follow-up automation

Four phases · reshufflable

Data home + automations

Employers + matching + invoicing

Website + digital application

After-hours agent + new automations

How the money works

Five moves

One system Rachel lives in

What's different in v1

Focus on the data home — idea #3 from v0

Public website + digital application — idea #1 from v0 — on a parallel track

Build it as a custom internal system, not a hybrid platform stack

Why custom over hybrid

The vision

Every employee & employer in one place

Every conversation captured

Click-to-call — and inbound calls too

Cards generate themselves

Calendar & Gmail live inside

A morning dashboard

VA-ready

Exports, always

The phased plan

Same work, half the calendar

Foundation + data home

Communications + auto-generated cards

Calls — make them, record them, extract from them

Matching engine and other integrations

Dashboards, invoicing, and whatever comes up

Automate the employee side aggressively.
Keep the employer side human.

Rachel opens her laptop, logs into one URL,
and the whole agency is on the screen.