The leverage of a team. The cost profile of software.

RAD has two sides, and each is sized the same honest way: hours of low-judgement work returned, tool subscriptions consolidated, and — deliberately left out of the arithmetic — the upside of more getting done and the downside avoided by never letting an unapproved message reach the outside world. AI does the work; a human owns the decision and presses go.

Below, the case is laid out in two parts — RAD Business, then RAD Professional — followed by the risk value common to both and what neither replaces. Every number is illustrative and labelled as such; plug in your own.

---

RAD Business — value to your business

Most small and mid-sized teams know what good marketing looks like — consistent content, prompt follow-up, real nurture, campaigns that actually go out — and rarely have the hands to do it week after week. RAD Business changes the maths by making one engine draft across every channel: replying to leads, writing content, running nurture, drafting campaigns, answering visitors via chatbot, capturing survey feedback, watching the market — staging every outbound message for a person to approve. The same team ships far more, marketing finally happens consistently, point-tool subscriptions collapse into one engine, and there's no brand risk, because nothing reaches a customer until a human approves it.

The case rests on three independent levers, each sized with an illustrative range — they vary enormously by business, team size and channel mix. Treat the ranges as prompts to estimate your own numbers, not as findings.

1. Output / throughput — the same people ship far more

The engine drafts; your team edits and approves. Because drafting is the slow part, the number of content pieces, follow-ups and campaigns one person can ship per week goes up sharply — without hiring.

Work the engine drafts	Illustrative lift per person	What stays human
Blog / SEO posts (grounded, multi-channel repurposing)	from ~1 → 3–6 per month	the brief, the edit, hitting publish
Social / channel content packs (LinkedIn, X, dev.to, newsletter)	a handful → dozens of drafts/month	voice check + approve
Lead follow-ups & nurture touches	the ones you get to → most of them, on cadence	approve the reply that goes out
Bulk-email campaigns	quarterly → as often as you have something to say	approve the send

Illustrative; varies. The lever is drafting throughput — review time is the new bottleneck, and it is a far smaller one.

2. Tool consolidation — one engine instead of a stack of subscriptions

The capabilities below are commonly bought as separate monthly SaaS subscriptions, each with its own seat fees, its own login, and its own copy of your customer data. RAD Business covers the same ground as one engine on your own cloud:

Job to be done	Typically a separate subscription	In RAD Business
Email marketing / bulk campaigns	a dedicated ESP	bulk-email (Listmonk)
Website chatbot / lead capture	a chatbot SaaS	Flowise RAG chatbot
Surveys / feedback	a survey SaaS	surveys (Formbricks)
CRM sync	a CRM + connectors	crm-sync (Twenty)
Blog / newsletter	a CMS + newsletter tool	Ghost
Social scheduling	a scheduling SaaS	content packs + Buffer drafts
Market / competitor monitoring	a monitoring tool	market-intelligence + private search

Illustrative; varies. Several point subscriptions (each commonly tens to low-hundreds of dollars a month, more with seats) consolidate into one engine whose main running cost is the GCP infrastructure it sits on — which scales to zero when idle — plus metered LLM usage routed through a cost-governed gateway. We quote no specific tool prices here; use your own current invoices.

3. Time saved — hours of drafting, follow-up and admin returned

Every approved draft is an hour your team didn't spend staring at a blank page or chasing an open loop. The hours returned are illustrative and vary by role — and they're returned after accounting for review-and-approve time, which is real and deliberately not zero.

Recurring task the engine absorbs	What it does	Illustrative hrs/week
Drafting content (blog, social, newsletter)	first drafts, grounded + repurposed per channel	3–8
Lead reply & follow-up	classify inbound, draft warm replies, run nurture cadence	2–6
Campaign assembly	draft bulk-email, segment, stage for send	1–3
Inbound support triage	classify and draft replies to the support inbox	1–4
Market / competitor watch	weekly scan + drafted digest	1–2
Illustrative total		~8–23

Illustrative; varies. The point of the table is not the total — it is that most of these hours are production overhead, not the strategy and judgement only your team can do.

An illustrative value model

A deliberately simple, transparent calculation — a thinking tool for a conversation about a specific business, not a quote and not a promise. Plug in your real numbers; add the levers that matter, ignore the ones that don't.

monthly value = ( hours returned/week × loaded hourly rate × ~4.33 weeks )  +  consolidated tool subscriptions no longer needed/month  +  value of incremental output (extra leads worked, posts shipped — often largest, left unmodelled)

Stated assumptions (all illustrative — change them):

• Hours returned/week — not the full task time above, but the share genuinely given back after review and approval. We model a conservative 5 hrs/wk and a moderate 12 hrs/wk, below the absorption band on purpose.
• Loaded hourly rate — the all-in cost of the marketer/owner doing the work (comp + overhead), or the value of their best alternative use of the hour. Illustratively $40–$100/hr.
• Tool savings — sum of the point subscriptions you'd retire. Use your own current invoices; we invent no prices.
• We do not count the output/upside lever in the arithmetic, even though it is often the largest. Leaving it out keeps the floor credible.

Illustrative monthly time-saved value (hours/week × loaded rate × 4.33):

	$40/hr	$70/hr	$100/hr
5 hrs/wk (conservative)	~$870	~$1,520	~$2,170
12 hrs/wk (moderate)	~$2,080	~$3,640	~$5,200
20 hrs/wk (higher)	~$3,460	~$6,060	~$8,660

Illustrative only. Rounded. Your mileage will differ — the value is the method, not the cells.

How to read it: the time-saved lever alone, at a conservative 5 hrs/week, lands around ~$1–2k/month — before adding retired tool subscriptions and before any credit for extra leads worked or content shipped. Those sit on top as headroom the model deliberately leaves out, so the floor stays honest.

Compared to hiring or an agency

The realistic alternatives to "get marketing done consistently" are a hire or an agency retainer. RAD Business is not a replacement for either's judgement — it is the production engine underneath.

	A marketing hire	A marketing agency	RAD Business
Cost	salary + benefits + tools	a monthly retainer	software on infrastructure you already own + metered LLM usage
Speed to consistent output	ramp + one person's hours	weeks of onboarding; their queue	drafts across every channel from day one; review is the only bottleneck
Control	high, capped by their hours	low — you're in their process	total — every send waits for your approval
Where your data lives	in your systems	scattered across the agency's tools	in your own Google Cloud, your own Directus
Scales with volume	hire again	bigger retainer	the same engine drafts more; add review capacity, not headcount

The strongest configuration is the engine plus a person: a marketer (or the owner) freed from the production grind to spend judgement on strategy, brand and offers. For a team that can't justify a full hire or retainer, the engine provides a floor of marketing leverage that wasn't economic before.

---

RAD Professional — value to your leaders

A senior executive's time is the scarcest, most expensive asset in the organisation — and much of it is spent not on judgement and decisions, but on the administrative tail: triaging an overfull inbox, scheduling back-and-forth, assembling context before meetings, drafting routine email, chasing follow-ups. RAD Professional returns those hours to high-value work — it reads the exec's world, drafts in their voice, and prepares each action so they start from done, not blank. It does this without the risk of turning AI loose on someone's email and calendar, because nothing goes out in the exec's name until they approve it in the chat thread.

Where the time goes

These are the recurring, low-judgement tasks the assistant absorbs or compresses. The weekly hour ranges are illustrative and vary enormously by role, seniority and inbox volume — a founder closing a round looks nothing like a stable-state COO. Treat the ranges as a prompt to estimate your own numbers.

Task the assistant absorbs	What it does	Illustrative hrs/week	What it compresses, not eliminates
Inbox triage	prioritises, summarises threads, surfaces what matters, drafts replies	3–8	you still decide what to send
Scheduling back-and-forth	proposes times, detects conflicts, drafts holds/reschedules	1–4	you still approve the booking
Meeting prep	assembles a brief — attendees, history, relevant docs and research	1–4	you still run the meeting
Drafting (email/docs/memos)	first drafts in your voice from your prior correspondence	2–6	you still own the words
Research / "ask my workspace"	private web research and cited answers from your own files	1–4	you still draw the conclusion
Follow-up & task capture	tracks open loops, surfaces what's owed and to whom	1–3	you still do the work owed
Illustrative total		~9–29	varies by role

The point of the table is not the total — it is that most of these hours are administrative overhead, not the work only the executive can do. That is exactly the band the assistant targets.

An illustrative value model

A deliberately simple, transparent calculation — a thinking tool for a conversation with a specific executive, not a quote and not a promise. Plug in their real numbers.

monthly value = hours returned/week × loaded hourly value × ~4.33 weeks/month

Stated assumptions (all illustrative — change them):

• Hours returned/week — not the full task time above, but the share the assistant genuinely gives back after review and approval. We model a conservative 3 hrs/wk and a moderate 8 hrs/wk, well below the task-absorption band, on purpose.
• Loaded hourly value — total cost of the executive's time (compensation + overhead), or the value of their best alternative use of an hour. Illustratively $150–$400/hr; many senior roles sit higher, which only strengthens the case.
• We do not count second-order gains (faster responses, fewer dropped balls, better-prepped meetings) — real, but unmodelled here, so the figures stay conservative.

Illustrative monthly value (hours/week × loaded rate × 4.33):

	$150/hr	$250/hr	$400/hr
3 hrs/wk (conservative)	~$1,950	~$3,250	~$5,200
5 hrs/wk (moderate)	~$3,250	~$5,400	~$8,700
8 hrs/wk (higher)	~$5,200	~$8,700	~$13,900

Illustrative only. Rounded. Your mileage will differ — the value of this table is the method, not the cells.

How to read it: even the most conservative corner — 3 hrs/week at $150/hr — returns around ~$2k/month of executive time. The model deliberately leaves the headroom — higher hour-bands, unmodelled second-order gains, avoided-mistake value below — out of the arithmetic so the floor stays credible.

Compared to a human EA or chief of staff

A great executive assistant is invaluable, and RAD Professional is not a replacement for one — it is an augmentation. The honest comparison:

	Human EA / Chief of Staff	RAD Professional
Cost	a full salary + benefits + overhead	a fraction of that, as software on the exec's own cloud
Availability	working hours; out sick; on leave; turns over	always-on — drafts the 6am brief and the 11pm reply at the same standard
Scale across a team	one person supports one or a few execs	the same pattern deploys privately to every leader, each in their own Workspace
Onboarding	weeks to learn voice and context	learns your voice from your real correspondence; sharpest in the first couple of weeks
Judgement & relationships	its real strength — reads the room, manages people, owns ambiguity	deliberately does not do this — it drafts and prepares; the human decides

The strongest configuration is both: an EA freed from the mechanical tail (triage, scheduling, first drafts, prep packs) to spend judgement where it actually compounds. For a leader without dedicated support, the assistant provides a floor of leverage that wasn't economic before.

---

Risk value — the part most ROI cases miss

Time and money saved are only half the case. The other half is downside avoided — and for a system that acts in your name, that belongs in the case, not a footnote. It's the same guarantee on both sides of RAD.

• The gate prevents a costly wrong move. Nothing is emailed, posted, published, booked or committed until a human approves the row. AI generates → a row is staged as Pending Approval → a person reviews and edits → sets Approved → a separate scheduled sender acts. The check is a strict, trimmed status === "Approved" match, and generation and sending are always separate workflows; the conversational assistant is never given write access. A wrong price, a tone-deaf post, a campaign to the wrong list, a mis-sent email to a client or board member, a wrongly-declined meeting — these are structurally impossible to do autonomously. Zero approved rows → zero actions. That is the design, not a setting. (Posts to third-party platforms are never auto-published; drafts are staged for a human, respecting those platforms' terms.)
• An audit ledger by construction. Every proposed action and every decision — who proposed, what, when, approved or rejected — is recorded in Directus. For a regulated or governance-conscious team, that defensible record of "nothing went out ungated" has real value on its own.
• Data sovereignty. The whole engine runs on your own cloud — RAD Business in your Google Cloud project, RAD Professional reaching Workspace through the exec's own OAuth grant with least-privilege scopes. Customer emails, leads, content, mail, calendar and documents stay under your access controls — not handed to a stack of third-party SaaS vendors to be pooled, retained, trained on, or scattered across tools you'd have to chase down to delete. Even whole-web search runs through a self-hosted engine, and LLM traffic is routed through a gateway you control.

A single avoided off-brand blast, one wrong message in an executive's name to a client or regulator, or one customer database you don't have scattered across five vendors when a breach or audit comes — can outweigh a month of the savings models above.

---

What it does NOT replace

Being honest about the boundary is what makes the rest credible. On both sides, the same line holds.

• Strategy. What to say, to whom, and why — the positioning, the offers, the campaign calendar; which deal to take, how hard to push, when to say no — stays with you. The engine executes a strategy; it doesn't set one.
• Judgement. It drafts in your voice, but the call on whether a message is right — on-brand, well-timed, true — is human. That judgement is exactly what the approval gate exists to apply.
• The relationship. It can draft a warm reply, but it can't have the relationship with your customer or your counterpart. Trust, reading the room, the hard conversation — human, and staying human.
• The decision and the accountability. It never sends, publishes, books or commits without explicit approval. The leverage is real; the accountability stays entirely with your team. That is not a limitation to engineer around — it is the whole point.

If a vendor promises AI that runs your marketing, inbox or calendar autonomously, end to end with no human in the loop, they're selling you the exact risk RAD is designed to remove.

---

How to think about payback

Don't think of it as buying software; think of it as buying back hours and consolidating tool spend — with a guardrail. Set the illustrative models against the all-in cost — GCP infrastructure, which scales to zero when idle, plus metered LLM usage — using your real loaded rates, current invoices, and an honest estimate of hours genuinely returned. For most teams, the time-saved lever and retired subscriptions cover the running cost alone — with the biggest prize (extra leads worked, content shipped, executive hours returned) sitting on top as unmodelled upside, alongside the avoided-mistake and sovereignty value.

The cleaner way to frame the decision: the work that actually gets done, consistently, by the team you already have — without ever letting an unapproved message reach the outside world. Size it with the model, sanity-check it against a quarter of real use, and let the gate carry the risk.

What it actually costs

No "contact sales" wall you can't see behind first. Module deployment is credits-based — you pay for build time and, for advanced deployments, a module fee — with current pricing and subscription tiers published at radmodules.dev, including a guided-setup option if you'd rather not self-serve. A fuller RAD Business or RAD Professional engagement — brand-voice and knowledge-base configuration, multi-module deployment, handover to your team — is scoped and quoted individually, and may include an ongoing retainer for support and tuning. Full terms at terms.html.

AI does the work. A human owns the decision and presses go.

All figures on this page are illustrative and clearly labelled as such. They are a tool for sizing the opportunity for a specific business or person — not a quote, a benchmark, or a guarantee.