A ninety-day social plan — positioning, a hook library, a content calendar, and the production set that makes it run.
Most agents flatten themselves into a category. Robbie has a separate life that makes the category interesting.
Recommend line one — "The Restorer." It travels across all five pillars without rewriting. It permits short-form humour, long-form detail, and the kind of testimonial that doesn't sound like every other agent's testimonial.
Each lane runs to a different audience. None is optional. The mix produces the personality.
Signature pillar. Robbie in his garage, on a build, at a show, walking a property. Each piece draws a line between a car detail and a property detail — patina, factory line, original glass, original chrome, original brick. Twice per week. The pillar that makes him uncopyable.
Listings shown the way a connoisseur shows a car — exterior line, interior detail, what the architect did, what the previous owner did, what the river adds, what the price reflects. Auction commentary in his MC register.
One graphic per week showing one inner-Brisbane number — Teneriffe unit growth, New Farm median, days on market, auction clearance rate. Plus the sold-flex — every settlement a piece, said quietly.
The MC voice as educator. How a vendor should choose. How an auction is run. What a bid means. What an under-bid means. What an off-market is and is not. These are the videos that turn passive watchers into appraisal requests.
Used sparingly. Johannesburg. The first Australian house. Why classic cars. Why New Farm specifically. Family. Mentors. The restraint pillar — too much story is the agent who never sells; too little and the personality never lands.
| Pillar | Per week | Per month | Per quarter |
|---|---|---|---|
| The Restorer (cars × houses) | 2 | 8 | 24 |
| On the Patch | 2 | 8 | 24 |
| The Number | 2 | 8 | 24 |
| Operator's Notebook | 1 | 4 | 12 |
| The Long Way | 0.5 | 2 | 6 |
| Total short-form pieces | 7.5 | 30 | 90 |
Plus one long-form YouTube property tour per month and one suburb-deep-dive per quarter.
A hook is the first three seconds. It must stop a thumb without lying about what comes next. Use these literally or as scaffolds — but say them in his voice, not the script's.
Hook engineering rule. Three opens that work for Robbie's voice — the declarative ("Original chrome. Original brick. Same conversation."), the contrarian ("Most renovations get this exactly wrong."), the numbered list with low-friction count ("Three things a restorer notices a buyer's agent misses."). Avoid the question-open and the "you won't believe" open. They flatten the personality.
Pieces are scheduled by pillar, not by mood. Cadence is the moat — most agents post in bursts; pattern beats burst.
| Week | Day | Format | Pillar | Working title |
|---|---|---|---|---|
| Week 1 — re-introduction | ||||
| 1 | Mon | Reel · 30s | The Long Way | "Johannesburg to New Farm — 13 years on one patch" |
| Tue | Carousel | The Number | Three numbers from inner Brisbane this month | |
| Wed | Reel · 20s | The Restorer | "Original chrome on the Porsche. Original glass on the Queenslander. Same conversation." | |
| Fri | Reel · 60s | On the Patch | Walk-through of current Newstead listing | |
| Sat | Story set | On the Patch | Open-home behind the scenes | |
| Week 2 — establish the lanes | ||||
| 2 | Mon | Reel · 25s | Operator's Notebook | "The first thing I tell a vendor — and it isn't price" |
| Tue | Reel · 20s | The Restorer | Saturday workshop hands. Sunday appraisal hands. Same eye. | |
| Wed | Single graphic | The Number | "Teneriffe units up 15.68% — the highest in inner Brisbane" | |
| Thu | Reel · 30s | On the Patch | Auction-day vlog if applicable; if not, current listing tour | |
| Sat | Sold post | The Number | Quiet sold-flex with one-line testimonial | |
| Week 3 — pattern lock | ||||
| 3 | Mon | Reel · 20s | The Restorer | "Three things a restorer notices that a buyer's agent misses" |
| Tue | Reel · 45s | Operator's Notebook | "What a bid actually means at a Brisbane auction" | |
| Wed | Reel · 60s | On the Patch | New listing — full walk and condition read | |
| Thu | Single graphic | The Number | "Auction clearance is at 62% — what 62 means and what it doesn't" | |
| Sat | Story set | On the Patch | Saturday open-home reel — voice-led | |
| Week 4 — the long-way piece | ||||
| 4 | Mon | Reel · 75s | The Long Way | "The first house I owned in Australia" |
| Tue | Reel · 25s | The Number | "Sold. Vernon Terrace. Off-market. Four-day campaign." | |
| Wed | Reel · 30s | The Restorer | Working on the car. Talking through what it teaches him about price restraint. | |
| Thu | Reel · 60s | On the Patch | Listing two — interior cinematic + price-rationale voiceover | |
| Sat | Carousel | Operator's Notebook | "Three rules I do not break, ever" | |
Same five-piece-per-week shape. Pillar mix tightens to 2 + 2 + 2 + 1. The Long Way drops to one piece per fortnight. Auction-week supersedes everything when applicable — auction-day vlogs replace whatever else was scheduled. By week 8, the Restorer pillar should have produced 12+ pieces, the Patch pillar 12+, and Robbie's Reels view-count median should be lifting visibly.
| Week | Anchor piece | Pillar split | Outcome target |
|---|---|---|---|
| 5 | "The cheapest river view in Newstead this week" | R2 / P2 / N2 / O1 | First reel above 10K views |
| 6 | "Why I work New Farm specifically, and not the rest of Brisbane" | R2 / P2 / N1 / O1 / L1 | First inbound DM appraisal request |
| 7 | "This building has nine apartments. Three sold this year. Watch what the price did." | R2 / P2 / N2 / O1 | Building-level case-study reel |
| 8 | "The reason I restore cars — and what it has to do with houses" | R3 / P2 / N1 / O1 | The pillar piece that pins the personality |
Volume holds. Quality lifts. By week 12, Robbie should have 90+ short-form pieces published, one long-form YouTube tour, and one quarterly suburb-deep-dive (Newstead apartment market 2026). The first vendor-lead conversion attributable to social should land in this window. The week-12 piece is the year-one anchor — a long-form reel mixing his garage, an inner-Brisbane property, the same voice across both, with the closing line already locked.
| Week | Anchor piece | Pillar split | Outcome target |
|---|---|---|---|
| 9 | "Ten years of New Farm prices in one chart" — long-form YouTube + reel cut-down | R2 / P2 / N3 / O1 | Long-form establishes him as a market voice |
| 10 | "Settled this week. Held the price. The vendor wrote the testimonial." | R2 / P2 / N2 / O1 | First testimonial-led reel |
| 11 | "Newstead Apartment Market 2026 — the quarterly read" | R1 / P2 / N3 / O1 / L1 | The deep-dive piece — market-voice positioning |
| 12 | "The Restorer — twelve weeks in" | R3 / P2 / N1 / O1 | Anchor piece. Pin-to-profile. |
A content engine that depends on Robbie filming every day will fail. The set produces in batches, on a fortnightly cadence.
Half-day shoot in Robbie's workshop. 8 Restorer-pillar reels captured against the car. Same shirt, same lighting, same single shot — micro-variations across the eight pieces. Produces two weeks of Restorer content in three hours.
One Newstead/Teneriffe/New Farm walking shoot per week, timed against current listings or open-homes. Captures: one ninety-second listing tour, two thirty-second on-street reels, one auction-day vlog if applicable. Two-hour window, gimbal + iPhone, no crew.
One forty-five-minute talking-head session per week. 3 Operator's Notebook reels captured per session, single-take, single shirt. Three weeks of Notebook content in one session.
Built without Robbie on camera. JRVS produces one graphic + one voiceover per week from inner-Brisbane data — Robbie records the voiceover from his phone at the end of the Notebook session. Two-week buffer maintained.
One longer-form sit-down per month. Sixty to ninety seconds. Lit, considered, single take. Pinned moments — origin, mentor, why-the-patch, the hardest sale.
| Activity | Cadence | Time per session | Per fortnight |
|---|---|---|---|
| Garage Day | Fortnightly | 3 hrs | 3 hrs |
| Patch Walk | Weekly | 2 hrs | 4 hrs |
| Notebook session | Weekly | 45 min | 1.5 hrs |
| Voiceover top-up | Weekly | 15 min | 0.5 hrs |
| Long Way piece | Monthly | 1 hr | 0.5 hrs |
| Robbie's time per fortnight | ~9.5 hrs |
Everything past Robbie's hand — editing, captioning, scheduling, distribution, performance review — runs through JRVS infrastructure (Remotion video pipeline, Whisper auto-caption, Buffer / Later scheduling, Metabase performance dashboard).
| Channel | Pillars served | Native posting cadence | Posting window |
|---|---|---|---|
| Instagram (Reels + Feed + Stories) | All five | 5–6 / wk | Mon–Sat 7am, 12pm, 5pm AEST |
| TikTok | Restorer, Patch, Notebook | 4–5 / wk | Cross-posted from IG with TikTok-native cuts |
| YouTube Shorts | Restorer, Patch, Notebook | 4 / wk | Cross-posted; long-form tour monthly |
| Notebook, Number, Long Way | 2–3 / wk | Tue, Wed, Thu — vendor-side audience | |
| Facebook (Robbie page) | All five — repost from IG | 4–5 / wk | Cross-posted; comment engagement managed |
| McGrath internal channels | Patch, Number — sold-flex | Weekly | Whatever the franchise distribution requires |
| Metric | Cadence | Day-30 target | Day-60 target | Day-90 target |
|---|---|---|---|---|
| Pieces published | Weekly | 20 | 45 | 90 |
| Median IG Reel views | Weekly | 1.5K | 5K | 15K |
| Top piece IG Reel views | Weekly | 10K | 50K | 150K |
| IG follower growth | Weekly | +200 | +800 | +2,500 |
| TikTok follower growth | Weekly | +150 | +1,000 | +5,000 |
| Inbound DMs (qualified) | Weekly | 2 | 8 | 20 |
| Appraisal requests sourced from social | Monthly | 1 | 3 | 6 |
| Listings sourced from social | Monthly | 0 | 1 | 2 |
Targets are honest. The day-90 follower numbers are conservative against the inner-Brisbane personality-agent ceiling — Drew Davies cleared 100K over years; Robbie does not need to. He needs 3 appraisal requests per month directly attributable to social by month four — that is the unit economics that keep the engine self-funding.
One Metabase dashboard built day one — pieces published, views by pillar, follower growth, DM volume, appraisal-request attribution. One weekly five-line note from Nathan to Robbie — what worked, what didn't, what next week looks like.
The OAIC opened a sector-specific privacy compliance sweep in January 2026. Real estate is named. The voice can be loose; the privacy posture cannot.
| Day | Action | Owner | Output |
|---|---|---|---|
| 1 | Engagement signed. Robbie kick-off call. | Nathan + Robbie | Voice samples, photo references, three classic-car photos, current listing list |
| 2 | Privacy policy refresh. APP 1.4 audit. | JRVS engineering | Updated policy live on every Robbie surface |
| 3 | Brand-voice fine-tune captured (LLM tone profile) | JRVS engineering | Robbie-toned listing-copy generator live |
| 4 | Garage Day shoot — eight Restorer reels | Robbie + cameraperson | Eight reels — fortnight-one supply |
| 5 | Notebook session 1 — three reels | Robbie + cameraperson | Three Notebook reels banked |
| 6 | First Patch Walk — current Newstead listing | Robbie | One ninety-second tour, two on-street reels |
| 7 | The Number engine live — first graphic + voiceover scheduled | JRVS engineering | Two-week Number buffer banked |
| 8 | Buffer / Later schedule loaded for week 1 + week 2 | JRVS engineering | Twenty pieces queued |
| 9 | Metabase dashboard live | JRVS engineering | Real-time view of every metric in section 8 |
| 10 | Week 1 Mon piece publishes — "Johannesburg to New Farm" | Auto | First piece live |
| 11 | Inbound DM monitor live — Ollama AI inbox first-touch | JRVS engineering | After-hours DM auto-acknowledge with qualified hand-off to Robbie |
| 12 | LinkedIn refresh — Notebook content lane opens | Robbie + JRVS | First LinkedIn long-post published |
| 13 | YouTube channel set up — long-form scheduled | JRVS engineering | Channel live, branding aligned |
| 14 | End-of-fortnight review — first ten pieces against day-30 targets | Nathan + Robbie | Five-line note. Adjustments. Week-3 cued. |
Before we automate anything, we map what he actually does by hand. The audit answers one question — where is his time being spent on work a system should be doing.
How we use this. Forty-five-minute discovery call — Robbie talks, Nathan listens, transcript captured via Whisper. The output is a process map showing every manual step, the time it costs, and the failure modes. The automation map in the next section is then weighted against the actual answers.
A streamline removes a manual step. A scale puts an output engine in front of a single creative input. Most agents conflate the two — and end up with neither.
| Lane | Manual today | JRVS streamlined | Time saved per week | Build days |
|---|---|---|---|---|
| Enquiry first-touch | Robbie reads every DM and email; replies in scattered windows. | AI inbox (Ollama first-touch) acknowledges every enquiry within sixty seconds, qualifies (suburb, intent, finance), routes only the warm ones to Robbie. | 3–5 hrs | 3 |
| Vendor weekly reports | Manually compiled from open-home numbers, portal stats, buyer feedback. | n8n pulls portal data + open-home sign-ins + buyer-feedback CRM notes; auto-generates the report; Robbie approves and sends. | 2–4 hrs | 4 |
| Listing copy | Written from scratch or via McGrath template; tonal flatness. | Brand-voice fine-tuned LLM produces first-draft listing copy in Robbie's voice from photo set + property facts. He edits, ships. | 2–3 hrs | 2 |
| Social scheduling | Reels uploaded one platform at a time when remembered. | Buffer / Later schedule loaded from a single Airtable record; one piece publishes IG / TikTok / YouTube / LinkedIn natively. | 2 hrs | 2 |
| Open-home follow-up | Manual SMS or call after inspection; some inspectors slip. | Sign-in app feeds n8n; templated personalised follow-up sent within thirty minutes; warm hand-off only when buyer engages. | 1–2 hrs | 2 |
Total streamline saving: 10–16 hours per week back to Robbie.
Not every pillar scales. Two of the five do — they have a clean creative input and an obvious distribution multiplier. One is a borderline case. The remaining two stay artisanal because the personality is the product.
Robbie's voiceover is the only human input. The graphic, the data, the platform-native cuts, the captions all run on infrastructure. Output multiplier — one fifteen-minute Notebook session can produce 10+ Number reels.
Single-take talking head. Same shirt, same lighting, same single shot. Once we have a Robbie AI avatar trained, the Notebook pillar becomes — Robbie writes a thirty-second script in his voice; HeyGen produces the reel; native cuts ship. The single human input is the script.
The walk-through is the personality piece — Robbie's eye on the architecture is what people watch. The avatar can reproduce his face but not his read. Use AI for B-roll generation, voiceover-only listing-summary cuts, and the auto-edit of a forty-five-minute walk-through into ten short-form pieces.
The car has to be real. The hands have to be his. The signature pillar does not scale — and should not. Two pieces per week, filmed in the workshop, stays the JRVS-supported but manually-shot lane. That is the moat.
Origin story by AI avatar is uncanny and trust-eroding. The Long Way piece must be the real human, sat down, telling a real story. Do not scale this lane.
| Pillar | Manual today | After streamline + scale | Output multiplier |
|---|---|---|---|
| The Restorer | 2 / wk manual | 2 / wk manual (protect) | 1× |
| On the Patch | 2 / wk manual | 2 / wk manual + 5 cut-downs from one filmed walk-through | 3.5× |
| The Number | 2 / wk semi-manual | 10 / wk fully automated | 5× |
| Operator's Notebook | 1 / wk manual | 3 / wk real-Robbie + 5 / wk avatar variants | 8× |
| The Long Way | 0.5 / wk manual | 0.5 / wk manual (protect) | 1× |
| Weekly output | 7.5 pieces | 27.5 pieces | ~3.7× |
The 3.7× output multiplier compounds. The first ninety days of the original plan ship 90 pieces. With streamline and scale layered in by day 30, the same ninety days produce closer to 280 pieces with Robbie's hour-budget reduced, not increased.
All five tools below produce avatar-led short-form video at scale. The first one was purpose-built for real estate.
| Tool | Strength | Real-estate fit | Pricing signal | Recommendation |
|---|---|---|---|---|
| HeyGen — Ask the Agent heygen.com/agencies/ask-the-agent |
Purpose-built real-estate avatar product. Photo + short voice recording produces a usable avatar; supports interactive avatars holding live conversations. | Highest. Built for the use case. | Free trial; tiered subscription from ~$30 USD/mo with avatar limits, pro tier ~$180+ USD/mo. | Lead candidate for Robbie. |
| Argil argil.ai |
Highest avatar realism per the $115K LinkedIn comparison test. Strong UGC-style output; widely used by performance marketers. | High — but UGC-ad register, not editorial. | ~$39 USD/mo entry; ~$150+ USD/mo for serious volume. | Backup if HeyGen avatar realism falls short. |
| Captions — Mirror captions.ai |
Mobile-first, fast UI; full reels pipeline (avatar + caption + B-roll). Strong on quick iteration. | Medium-high. | ~$10–$70 USD/mo per seat. | Cheap fallback for high-volume Notebook output. |
| Synthesia synthesia.io |
Enterprise-grade photorealism; 140+ languages. Strongest for educational / explainer content. | Medium — not built for short-form social. | From ~$30 USD/mo personal; enterprise from ~$1,000 USD/yr. | Use only if expanding into multi-language buyer segments (Mandarin, Cantonese). |
| Creatify creatify.ai |
URL-to-video workflow; batch testing. Best for testing many angles at once. | Medium — UGC ad bias. | ~$39–$229 USD/mo. | Use for paid Meta-ad variants, not organic content. |