34+ person output,
Quarter person bill.

Riley + Beth + Iris + Hattie + Sage pick up the inbox so you don't.

Every channel lands in one inbox. Riley drafts in your voice. Beth books without overbooking. Iris escalates the clinical ones. You tap approve. Auto-replies turn on channel-by-channel as trust builds; Sara watches the weekly pattern + tightens scripts.

• Riley
  • drafts replies in your voice across SMS / WhatsApp / email / web chat / GBP
  • weekly enquiry-topic cluster + 'these opening hours are losing leads' note to Sara
• Beth
  • per-customer no-show probability + over/under-book recommendation per slot
  • reminder cadence (24h + 1h + day-of) tuned per customer's prior pattern
• Iris
  • reads message intent against the clinical-vs-admin / urgent-vs-routine line
• Hattie
  • 30 / 60 / 90-day lapsed-customer touches with the right hook per segment
  • birthday + anniversary + 'we missed you' messages timed to your slow weeks
• Sage
  • auto-FAQs deflect the repeating questions before they reach you (with audit trail)

• Missed-call auto-text-back instant, templated, so a missed call never becomes a lost job
• Owner-approval gates on every send by default; dial trust up channel-by-channel

Cara + Pia + Felix + Faye + Mira: who to scale, who to chase, why people said no.

Cara measures CAC + LTV per channel and tells you which copy to swap. Pia clusters every "no" you got this week into offer / price / timing / trust and names the fix to test. Felix and Faye run quote-to-cash with every gate owner-approved. Mira drafts the marketing in your voice and leans into whichever copy worked last week.

• Cara
  • CAC + LTV + funnel drop-off per channel; names the channel to scale + the copy to swap
  • 'Scale google_ads (LTV/CAC 5.3x); rewrite tiktok copy (qual rate 5%)'
• Pia
  • clusters every 'no' into offer / price / timing / trust + names the fix to test
  • '4 declines this month were price-objection; test a payment-plan tier'
• Felix
  • quote, invoice, dunning workflow; owner-approval gates and audit trail
• Faye
  • proposal generation tuned to your previous-winning quote language
• Mira
  • auto-drafts marketing copy in YOUR voice; A/B-leans into the winner each week
  • per-platform timing: when the slot beats the open rate by 30%+, she defaults to it

• Voice-cloned outbound (Pro+) your voice, never a robot, for the high-trust calls

Eddy + Felix + Brace: the cashflow doesn't ambush you any more.

Eddy predicts which jobs will pay late before the invoice goes out and proposes the deposit rule per job. Felix runs the quote-to-cash chain (quote, invoice, dunning) with every gate owner-approved. Brace watches input-cost drift (materials / wages / fuel) and flags jobs whose margin has quietly slipped so you can re-rate before next quarter.

• Eddy
  • per-quote late-payer probability + deposit rule recommendation per job
  • '3 of the next 12 invoices are 70%+ likely to slip; switch them to 50% deposit'
• Felix
  • quote, invoice, dunning chain with owner-approval gates + full audit trail
  • tiered dunning cadence (7 / 14 / 21 / 28 days) tuned per customer's payment pattern
• Brace
  • tracks input-cost drift (materials, wages, fuel) per job-type week-over-week
  • '5 jobs this month had copper-price slip 12%+; re-rate similar work by Friday'

• Owner-alert on any single invoice over 21 days late OR overall AR ratio worsening week-over-week
• Xero / QuickBooks / FreeAgent 2-way sync (Pro+) so the books match without double-entry

Rae + Mira + Lex + Sara: the reputation flywheel without the late-night drafting.

Rae watches every platform; the sentiment classifier tags negatives with the hot-words that drove the call so you can audit any classification. Mira times the review request to the 24-48h emotional peak when the compliments at work are landing (2x the response rate vs scheduled blasts). Lex watches the three closest local rivals so a price-move or velocity-spike never sneaks up on you. Owner-alert on anything 3★ or below.

• Rae
  • auto-monitors Google + Trustpilot + Facebook + Yell + industry-specific platforms
  • drafts polite replies in your voice; queues for approval (or auto-sends ≥ 4★ once you trust it)
• Sentiment
  • keyword + intensifier + negator classifier (audit-trail per classification)
  • 'late' appeared 4× negative this week; pattern fix surfaces in Sara's brief
• Mira
  • per-customer review-ask after every completed job at the 24-48h emotional peak
  • per-platform routing (Google for older demo, Instagram for younger; auto-tuned)
• Lex (Pro+)
  • 3 local rivals continuously monitored; named price-move + review-velocity alerts
  • their copy themes vs yours; spots the angle they're winning on
• Sara
  • ops-lead weekly brief lists the 1-2 service tweaks that would shift the sentiment trend

• Owner-alert ≤ 3★ before it lands on the public rating, chance to fix in person first

Lyra + Quill + Vance + Linc + Echo (Studio): your marketing content team that doesn't sleep.

Lyra plans the calendar around what's actually converting in your niche. Quill writes long-form (blog, newsletter, lead magnet). Vance shapes short-form (social posts, hooks, captions). Linc schedules across platforms at the timing your audience actually opens. Echo and Reel (Studio tier) layer in voice-cloned reels + photo-to-video.

• Lyra
  • weekly calendar planning - which topic, which channel, which week, in your voice
  • cross-references winning content from neighbouring tenants in your niche
• Quill
  • long-form drafts (blog, newsletter, lead magnet) tuned to your tone-of-voice profile
  • SEO-aware structure so Sora can pick it up without a rewrite
• Vance
  • short-form copy: post hooks, captions, thread starters, DM scripts
  • per-platform formatting (LinkedIn vs Instagram vs X) without you copying + pasting
• Linc
  • posting schedule timed to YOUR audience's open-rate peaks per platform
  • respects your blackout windows (job site visits, holidays, family-only weekends)
• Echo + Reel (Studio)
  • voice-cloned reels + photo-to-video for jobs you'd never have shot on the day
  • caption + b-roll suggestions so the reel ships in minutes, not an evening

• Brand-voice guardrails trained on YOUR existing posts so nothing reads off-brand
• Owner-approval mode default-on for first 30 days; trust dial per channel after that

Sora's 8-agent SEO/AEO/GEO dept: show up where your customers actually search.

Three search behaviours, three optimisations. SEO for Google + Bing organic; AEO (answer-engine optimisation) for ChatGPT, Gemini and Perplexity citations; GEO (geo-targeted local pack + Google Business Profile) for 'plumbers near me' style intent. Sora leads. Geno (Studio) layers in 3D map-pack analysis + competitor SERP intelligence.

• Sora
  • keyword + cluster map for your niche; on-page + technical recommendations per page
  • weekly rank tracking against the 3 local rivals Lex picked
• Answer-engine optimisation
  • structures content for ChatGPT / Gemini / Perplexity citations (FAQ schema, summary blocks)
  • tracks which AI engines are citing you + which competitors they cite instead
• Local pack + GBP
  • weekly GBP audit (photos, hours, services, posts) with named fixes per attribute
  • review velocity + Q&A response tracking - both signals the local algorithm leans on
• Geno (Studio)
  • 3D map-pack analysis (grid-based ranking heat-map across your service area)
  • competitor SERP intelligence: which pages they rank for that you don't yet target

• Content hand-off to Quill when a content gap is found, the brief writes itself
• Monthly visibility report shows organic + AI-cited + local-pack rankings on one page

Sage + Otto + Penn + Vista: ask your customers what only they know.

Five cross-niche objectives baked in (NPS, CSAT, brand tracker, content pulse, lapsed win-back) plus per-niche overlays for regulator-heavy industries. Sage finds the question worth asking; Otto picks the 5-7 most decision- grade questions; Penn drafts the survey in your voice for the right channel; Vista names which cohort to send it to + which channel converts on responses.

• Sage
  • researches the question that would actually move your decision (citing the source per seed)
  • pulls per-niche overlays for regulator-heavy industries (GDC, FCA, RCVS, SRA)
• Otto
  • picks the 5-7 questions most likely to give you a decision-grade answer
  • removes leading / double-barrelled / unanswerable phrasings before they ship
• Penn
  • drafts the survey in your voice with the right tone for the channel
  • handles the intro + the thank-you so response rates stay high
• Vista
  • names the cohort: loyal vs lapsed vs new vs declined; sized for statistical signal
  • picks the channel (email vs SMS vs in-app) by what that cohort actually opens

• Findings auto-route to the right ops-lead: Sara for service ops, Maya for messaging, Otis for ops
• PII-safe storage (hashed respondent IDs); raw email/phone never persisted in the survey ledger

Iris + Rumi + Dax + the guardrails every draft passes through.

Iris reads message intent against the clinical / urgent / regulated line so ambiguous messages get escalated, not auto-replied. Rumi (Pro+) filters UK regulator bulletins to the 10% that affect YOUR niche; Dax actions them into your templates. Underneath everything, the output-guard trio (sensitive-data filter, hallucination judge, regulated-claim detector) inspects every agent draft before it can leave the building.

• Iris
  • reads message intent against the clinical-vs-admin / urgent-vs-routine line
  • escalates ambiguous-but-clinical content to YOU before any auto-reply
• Rumi (Pro+)
  • UK regulator bulletins filtered to the ones affecting YOUR niche (GDC, FCA, ASA, MHRA)
  • weekly summary in plain English: what changed, what it means, what to do
• Dax (Pro+)
  • auto-actions regulator changes into your templates + briefs the team
  • maintains the audit trail: who changed what, when, why, in response to which rule
• Output-guard
  • sensitive-data filter (cards, NI numbers, JWT tokens, AWS keys) blocks before send
  • hallucination judge: flags any claim the source data doesn't actually support
  • regulated-claim detector (per-niche reg-pack): blocks 'guaranteed', 'painless', etc.

• Audit trail every reasoning step + every block + every send is reviewable for 90 days
• Owner-only escalation queue nothing customer-facing slips out without you seeing it first

Pivot + Otis + 262 niche specialists do the unglamorous work that keeps the lights on.

Pivot tracks utilisation + on-time + capacity-vs-demand and names the shift-mix fix when capacity slips out of the 65-85% healthy band. Otis (ops-lead) gets the tactical brief scoped to THIS week's decisions. The niche-specific specialists carry the industry-specific levers no generic ops tool sees.

• Pivot
  • utilisation + on-time + capacity-vs-demand; 65-85% healthy-band tracking
  • 'utilisation 58% - pre-book filler slots for top-20 loyal customers'
  • 'pre-book contractor before winter rates lock' supply-chain recommendation
• Otis
  • weekly tactical brief scoped to ops decisions only (no marketing noise)
• Niche specialists
  • Burst (plumbing) models on-call fatigue so we never burn the rota
  • Tarn (plumbing) auto-checks engineer manufacturer-trained list against the boiler photo
  • Anchor (salons) spots leaving-signal patterns BEFORE the resignation
  • 262 niche specialists tuned to your industry on day one (same 6-dimension depth bar)

Ada + Pareto + Helix + Pivot + Memo + Marlow: the dashboard that's actually a story.

Ada tracks 20 canonical metrics and picks the 5-10 that matter this month. Pareto deep-reads customers (LTV three ways, cohort retention by acquisition channel, 80/20 + at-risk). Helix + Pivot add marketing- and ops-analytics with hypothesis-led recommendations. Memo writes the monthly + quarterly report in YOUR voice and a tactical brief for each ops lead so they interpret + plan for the team. Marlow synthesises the lot into 3 lines you read with your morning coffee.

• Ada
  • 20-metric catalogue (revenue growth, CAC, LTV, churn, on-time, ...); picks the 5 that matter THIS month
  • custom-KPI definition (Pro+) add YOUR own metric to the catalogue with hypothesis
• Pareto
  • 'Top 5 customers = 62% of revenue + 2 dropping 30%+ this quarter'
  • cohort retention by acquisition channel + at-risk-top-customer named list
• Helix (Pro+)
  • 70/20/10 budget allocator + bottom-quintile cut (typical 18-35% CAC drop)
• Pivot (Pro+)
  • capacity-vs-demand + utilisation hypothesis-led ops report
• Memo
  • (client) 'Net margin slipped to 7%. Brace flagged 5 jobs whose copper cost rose; re-rate by Friday.'
  • (ops-lead) Marnie + Otis + Faye + Sara each get the brief scoped to THEIR decisions this week
  • board-pack (Pro+) designed quarterly PDF, not a dashboard dump
• Marlow
  • 3-line morning brief synthesising the whole team, read it in 20 seconds