J&M AI Services — Business Plan

Built 2026-07-02 from the AI Operator Academy and AI Agency Inner Circle curricula (see sources/), cross-checked against J&M’s existing offer strategy. Every non-obvious claim traces to a lesson or Slack message cited inline.

0. Prime Directives

The three rules that outrank everything else in this plan:

1. The antidote to overwhelm is prioritization, elimination, and consolidation. When the business (or a client) feels buried, the move is never “more” — rank it, cut it, or merge it.
2. Repeatable work becomes saved, deterministic automation with a single source of truth. Humans keep the judgment calls; the machine keeps the repetition.
3. Use the tools internally first. The internal machine is the product being built — every tool J&M sells has to be one J&M lives on.

1. The one-sentence business

J&M AI Services helps owner-operators of established local & service businesses reclaim 5–15 hours a week and stop leaking revenue, by auditing their workflows and then building AI systems with them — starting at a fixed-fee $999–1,999 diagnostic and continuing as a monthly done-with-you retainer. (Internal framing. Buyer-facing, the 5–15 hours is never claimed up front — it’s discovered on the call and stated back in their numbers.)

We are not “an AI automation agency.” That is the commodity trap — the moment you say “we build AI automations” you sound like every $50/hour freelancer on Upwork and price becomes the only lever left (AAIC, 1.2 Creating a Remarkable Offer). We sell a named diagnostic and a named build system, priced on the outcome (hours and dollars recovered), guaranteed.

2. Why this works now

This section is the short form of the north-star thesis in 00-thesis.md — read that first.

• This is 0→n work, not 0→1. The labs invent the tools; J&M gets them used inside real businesses. J&M is a distribution/implementation company, so every new tool is inventory, not a competitor.
• The adoption gap is the market. Per Menlo Ventures’ 2025 State of Consumer AI: 61% of US adults have used AI but only ~19% daily, and only ~3% of ~1.8B global users pay for premium (ChatGPT converts ~5% of weekly users). Competent, wired-in use — J&M’s level — is a sliver of that, realistically well under 1% (directional estimate). The skill of “using AI well” is scarce among exactly the people who most need it: busy operators doing $30k+/month with no time to learn it.
• The delivery cost is near zero. The whole front-end runs on Claude (incl. Claude Design for the report), cal.com/Calendly, Zoom+Fathom, and Notion — tools J&M already pays for. Four to five hours of work turns into a $999–1,999 sale (AOA, The AI Assessment Playbook / 01).
• Tool proliferation is a tailwind, not a threat. Every new “$15k/mo agency → <$100/mo skill” launch is more capability J&M can wield and more overwhelm for the DIYer. The moat isn’t the tools (anyone can buy those) — it’s being the expert who’s already hunted, learned, and wired them, running them behind the scenes, sold done-with-you on the decisions. You could change your own oil, or pay an expert; J&M is for the ones who pay the expert (AAIC 1.2, 1.3).

3. Ideal Client Profile (ICP)

Five questions define the ICP (AAIC 1.1 How to Define Your ICP):

Inverted-ICP filter (J&M house rule): they must have both money and one visible fixable gap you can name before the call. If you can’t point at the leak, they aren’t ready. And the leak is the opener: lead every first touch by handing them the observation free (“you’re not answering after-hours calls — I counted 4 missed this week”), before any ask. Personalized, unexpected, specific to THEM — the audit ask comes second.

4. The offer ladder

Three rungs. Each rung earns the right to the next. This is the core insight both communities share: the assessment is the natural front door to recurring revenue (AOA Concierge / 01: From Assessments to Recurring Revenue).

Rung 1 — The Assessment · $999 (solo) / $1,999 (team) · one-time, 1 week · credited toward what’s next

J&M’s version of the AOA “AI Tools Assessment” — a leverage audit of the owner’s workflows. Three deliverables: 1. 45-min Discovery Call — a SPIN-structured diagnostic that pulls 5–7 concrete pain points out of the owner, quantifies at least the top two in their numbers (hours × their stated hourly value — the Implication), and gets the owner to say what fixing #1 would be worth (the Need-payoff — their words go verbatim into the report) (see 03-audit-spin-scripts/). 2. Custom Report (built in Claude Design — AOA canceled Gamma for it; drop the discovery-call transcript into the template’s chat and say “fill in the assessment”) — 3–7 specific recommendations, each with the problem in their words, the exact tool/fix, weekly time saved, effort tier, and a 4-day quick-start plan. Priced and dated (“current as of”). Report spec (9 slides, per Corey’s open-sourced template — see sources/reference/corey-ganim-audit-template-video.md): financial-impact slide first (monthly net ROI staring at them on open), the effort-vs-impact matrix as the lightbulb slide (report prescribes top-left quick wins; top-right “major projects” is the named upsell slide), then quick wins → tool stack → 4-day plan → next steps. 3. 30-min Walkthrough Call — walk the report (expect ~half the call on the tool-stack slide), then close with the three questions in this order: Which of these will you tackle first? (they pick — a self-generated commitment, which beats anything you assert) → How urgent is that one for you? (they state the implication of waiting) → Do you want my help implementing it? (the ask lands only after they’ve committed and named the urgency). AOA reports >50% take the done-for-you/with-you upsell from exactly this close — the answer routes them into Rung 2 or 3. If yes, book the kickoff on the call — never “I’ll send the proposal.” If maybe, leave with a designed advance that has a date on it: “I’ll hold your credit until Friday; let’s put 20 minutes on Thursday to decide.”

Credited forward: the $999–1,999 is credited toward a Custom Build or the client’s first month of Concierge. It de-risks the front door and pulls buyers up the ladder — the diagnostic pays for itself the moment they continue. (Behaviorally this is a commitment device: they’ve paid, chosen a first fix out loud, and hold a credit — three consistency pressures toward Rung 2. Protect all three: never waive the fee, always make them pick the first fix, always mention the live credit at close.)

Guarantee (falsifiable, time-bound): If we don’t find at least one Quick Win you can ship in 7 days that recovers 5+ hours/week, the Assessment is free. (AOA 07/01.)

Is the 5-hour guarantee realistic? Yes — for this ICP, and the wording is what keeps it safe. - Finding 5+ hrs/week is ~easy for the target buyer ($30k/mo+, still in the day-to-day). It usually takes one task to clear the bar: hand-built quotes at 90 min/day = 7.5 hrs/wk; manual data re-entry 3–8 hrs/wk; lead follow-up + inbox triage 3–6 hrs/wk; weekly reporting 2–4 hrs/wk. If someone genuinely has no 5-hour pocket, they’re a lean/non-repetitive business you’d want to disqualify anyway — so the guarantee doubles as a qualifier, not a liability. - The promise is “find a fix worth 5+ hrs you can ship,” not “you will bank 5 hrs.” That distinction is deliberate — it makes the guarantee falsifiable on the fix’s designed savings, not the client’s subjective feeling. - The real risk is delivery, not finding. Three things to manage so the promise holds: 1. Measure the baseline on the call. Time the actual task (“your Tuesday quote block is 90 min”) so the guarantee is adjudicated on designed savings, not a vibe. The proposal generator already computes from their stated hours — keep it conservative (under-promise). 2. Anchor the week-1 win on a fix that needs no third-party approval (a quote skill, an email follow-up skill, an internal automation). Do not make an SMS fix (missed-call text-back) the 7-day anchor — A2P 10DLC registration takes days-to-weeks and will blow the window; ship SMS fixes once it clears (see §9). 3. Frame it as time-back-on-a-task, not lifestyle change — owners refill reclaimed hours, so quantify the task (“90 min → 5 min”), which is true and defensible, rather than promising freedom. - If you ever want to be more conservative, “3+ hours” is trivially defensible — but 5 holds for this ICP and reads stronger.

Why charge for the Assessment instead of giving it free: the price filters for serious buyers, funds the relationship, and makes the report a paid deliverable they act on rather than a free pitch they ignore — and crediting it forward removes the only real objection to paying.

Rung 2 — The AI Concierge · $1,500–2,500 / month · the core business

Done-with-you monthly retainer. The recurring engine (AOA Concierge / 01, 03). - Two 45-min strategy calls per month. Screen-shared working sessions where J&M builds with the client: Claude skills, context files, automations, live. ~1.5 hrs of call time/month. - Unlimited async support (Voxer/Slack) with a 12-business-hour SLA. - Hands-on-their-machine work (fixing a workflow where it lives, installing tools, debugging) runs over RustDesk attended sessions — client launches it, reads the code, watches J&M work. Beats screen-share-and-dictate for anything longer than 5 minutes, and it’s how the “priority remote support” line item gets delivered (see 06-solution-stack/stack-table.md). - Shared Notion hub — every call logged with recording, top-3 takeaways, two-sided action items, and a quantified list of everything built. This is the renewal mechanism: the ROI is on paper, so they renew. - Run AOA — Audit → Optimize → Automate — on every workflow, every time. Fix the process before automating it. Never automate a broken process (AOA Concierge / 03, 04). - Why done-with-you is the core, not done-for-you: the client ends up owning the know-how, not just the systems — that’s the durable “run it without me” promise and the referral engine. Language discipline: on calls and in the hub it’s “we/our” — “what we built this month,” “our next fix” — and the founder’s own SE-Idaho owner-operator identity is stated, not hidden. Co-creation is the strongest unity move available; the two working sessions per month ARE it — name them that way (“our build session”). Clients who explicitly want hands-off get Rung 3 (Custom Build) instead; don’t blur the two. - Pricing discipline: $1,000/month floor, target $1,500–2,500. At $1,500 for 1.5 hrs of calls that’s a ~$1,000/hour effective rate. If you close every pitch, you’re too cheap — raise until you hear “no.” - No seat cap — capacity is the experiment. (Revised 2026-07-05; supersedes the earlier “cap at 6” rule.) The deterministic back end — everything new and novel turned into saved code, one-click deployments, documented and pushed to GitHub/the VPS — is the capacity engine, and its whole point is that delivery cost per client keeps falling. So take every qualified discovery call and project and find the real upper bound empirically instead of asserting one (see §7a). A cap returns only if the quality tripwires in §7a say so — set then, from data, not now, from fear. Scarcity language discipline follows: never quote a seat count. The only scarcity that’s still true is calendar scarcity — “my next open build slot is [date]” — plus loss framing on their leak. Both stay honest at any client count.

Rung 3 — The Custom Build (DFY) · $5,000–30,000 project (+ optional care plan)

For clients who want it done for them, not with them. This is the AAIC high-ticket lane: custom AI systems sold via the 4-call framework at $10–30k (AAIC 3.1). Also houses the productized Local Visibility & Lead-Capture System (missed-call text-back, AI receptionist, review engine, booking) for local businesses, plus the demand-gen plays — Answer-Engine Visibility (AEO/GEO — get cited when buyers ask AI) and Paid Social Ads & Creative (Meta, via a Claude-native skill stack). These turn on after front-office capture is fixed (never pour demand into a leaky funnel); playbooks in 05-delivery-plans/. Sold only after an Audit or Concierge relationship exists, or via the full 4-call process for cold high-ticket. The 4-call runbook — Discovery → Strategy → Scoping → Contract, with the between-call actions — is in 03-audit-spin-scripts/four-call-close-runbook.md.

Paperwork (contracts): three ready MSA templates live in sources/contracts/ — pick by who owns the build: client-owns (one-time, they own it), agency-owns / no annual fee (you keep the IP, charge once), agency-owns / with annual fee (you keep the IP + recurring license — the model for a Concierge-style dependency). Fill the [blue bracket] fields and get a state attorney to review the first use. (Templates + mapping in sources/contracts/README.md.)

The funnel: $999 Audit → Concierge retainer → Custom Build upsell. Most revenue and enterprise value lives in Rung 2’s recurring base; Rungs 1 and 3 feed and expand it.

5. Positioning & the unique mechanism

Name the process so it stops sounding like a commodity and passes the “can I put ‘The’ in front of it?” test (AAIC 1.3). J&M’s live positioning (jacobdart.com) already names it — keep this language consistent across site and plan: - The overarching promise: turn your business into an operating system — your real work, turned into pieces that run themselves. - The diagnostic: The Assessment. - The mechanism: one-click skills (agents that do the repetitive work) on a knowledge spine (the business’s memory, connected across tools, compounding each month), delivered via AOA — Audit → Optimize → Automate. - Lead with pain removal, then the mechanism — and cold, it’s a question first: “How much of your day is admin you’d never hire for but can’t drop?” Only after they’ve named the hours does the claim land: “I take the guesswork and the 2-hours-a-day of admin off your plate — I turn your business into an operating system you own.” A benefit only exists against a stated explicit need. (AAIC 1.2.) - Authority move: volunteer a limit early — “This won’t fix your hiring problem, and if I can’t find you 5 hours I’ll tell you and it’s free.” Admitting the boundary before the pitch is what makes the rest believed. - Always tie back to dollars. Every recommendation is either money saved (hours × their stated hourly value) or money made (faster lead response, more closes). Get their hourly value on the call so the ROI math is theirs, not yours (AOA discovery Q6). Frame the ROI both ways, loss first: “staying manual costs you ~$1,800/month” lands harder than “this saves $1,800/month” — same math, stated as the running leak.

6. Go-to-market (channel sequence)

Sequenced, not scattershot (std-gtm; AOA 07/09 Seven Channels): 1. Months 0–3 — warm + local + outbound. SE-Idaho network and referrals first (highest trust, lives the node). Simultaneously: LinkedIn presence + cold outreach and Upwork for volume and reps. Scripts in 02-outreach-scripts/. 2. Months 1–6 — referrals & proof. Every delivered Audit/Concierge produces a quantified result → testimonial → the next deal. The Notion hub is the case-study generator. Match proof to prospect: same trade or same town beats a bigger number from elsewhere — a Rexburg plumber’s 6 hrs/week moves an Idaho Falls electrician more than a Seattle SaaS logo ever will. Tag every testimonial by trade + town in the Notion hub so the right one is retrievable per pitch. 3. Months 3–12 — inbound. LinkedIn content (the AAIC “0→26k followers” system), lead magnets, and local SEO/GEO compound so leads come inbound.

Primary metric: booked qualified discovery calls per week. Everything upstream (content, outreach) is judged on whether it produces those. Every touch has a designed advance: cold outreach → a 15-min “one thing I noticed” call; discovery call → a paid Assessment booked; walkthrough → kickoff dated on the call. A reply of “sounds great, send info” is a Continuation and counts as zero.

7. Unit economics & targets

Illustrative, conservative (not a forecast — validate against real close rates):

The math that matters: every Concierge client at $1,500 is ~$18k/year recurring off ~2 hrs/month of delivery — 6 of them ≈ $108k/year, 12 ≈ $216k/year. The old plan stopped at 6; the deterministic back end exists precisely to make 12 (and the audit/build volume feeding it) deliverable by one person. Rung 2 is the whole game — fed by Rung 1, expanded by Rung 3, bounded only by what §7a actually measures.

7a. Capacity: the bottleneck experiment & the hiring ladder

(Added 2026-07-05, replacing the fixed 6-client cap.) The strategy is deliberate: run intake wide open and find the true upper bound of a one-person shop running on deterministic code. Every repeatable thing — audits, reports, proposals, deploys, follow-ups, this plan itself — becomes saved, version-controlled, one-click infrastructure (Prime Directive #3; Agent OS in §12 is the cockpit). Each automation raises the ceiling, so the ceiling is a moving target that must be measured, not guessed.

The tripwires (leading indicators, checked weekly — the founder is the bottleneck when any two fire in the same week): 1. Response SLA slips — async client replies breach the 12-business-hour SLA more than twice in a week. 2. The calendar lies — a discovery call, workshop, or build session gets rescheduled for my capacity (not theirs) more than once a week. 3. Sessions ship nothing — a Concierge session ends without something shipped + logged (the renewal engine starving is the first real quality casualty). 4. The automation backlog grows — a week passes where nothing new got turned into deterministic code because delivery ate every hour. That’s the flywheel stalling: the exact work that raises capacity is the first thing overload cancels. 5. Lead decay — a qualified inbound lead waits >24h for first touch (the plan’s own speed-to-lead doctrine, violated at home).

When tripwires fire, the response is a pre-committed ladder, not a scramble. Work is offloaded in order of how easy it is to hand off — deterministic first, judgment last:

Rules of the ladder: the tripwire log (a weekly note in the Notion hub / Agent OS) is the source of truth — hire on data, not on a bad week. Every handoff inherits the same discipline: the contractor works from the playbooks (05/, 03/, 04/), their output is verified against the same templates, and anything they do twice gets turned into code too. Client-facing voice stays the founder’s until Rung 3 — done-with-you intimacy is the moat; back-office is what gets sold off first. And the moment a cap is warranted, it gets set from the measured bound and stated truthfully — that’s when seat scarcity becomes honest persuasion again (13-persuasion-review).

8. Validation gate & 90-day plan

Gate: don’t scale spend or systems until 3 paying Concierge clients prove the funnel end-to-end.

• Days 1–30: Finalize the Audit deliverable (Claude Design report template, question bank, analysis skill). Publish the offer page. Book & run first 5 discovery calls from warm + local. Land 1–2 paid Audits.
• Days 31–60: Deliver those Audits; convert ≥1 into a Concierge. Turn on LinkedIn content + one outbound channel. Book 8+ discovery calls.
• Days 61–90: Reach 3 Concierge clients. First quantified case study live. Decide whether to open the Custom Build lane based on demand seen in Audits.

9. Compliance & ethics (non-negotiable)

• Tool recommendations: disclose any affiliate relationship in writing in every report; never recommend a worse-fit tool because it pays more (AOA 07/01).
• Any SMS/texting automation (missed-call text-back, reminders) requires A2P 10DLC registration + TCPA consent before launch. Bake consent capture into every build.
• Client data: a simple DPA/mutual-NDA before touching a client’s systems or data.
• Remote access to client machines: written authorization in the engagement agreement first (add a remote-access clause to the MSA templates). Default to attended, per-session-code access (RustDesk, self-hosted relay); install unattended agents only for standing retainer clients with explicit sign-off. Servers via Tailscale + SSH so the client can revoke access in one click. Tooling detail in 06-solution-stack/stack-table.md.

10. What could kill this (honest risks)

• Capacity is the constraint, not demand. The Concierge caps at 6; growth past that means either raising prices or productizing into the Build lane / hiring. Plan for it early.
• Commoditization of “AI automations” is real and coming. The defense is process, named mechanism, relationships, and vertical depth — not staying on the technical frontier.
• Under-pricing out of fear. The curricula are emphatic: floors exist for a reason, raise until you hear “no.”
• Selling deliverables, not outcomes. Every artifact in this repo is built to keep the conversation on hours and dollars, not features.

Companion deliverables: 02-outreach-scripts/ (book the call), 03-audit-spin-scripts/ (run the diagnostic), 04-proposal-generator/ (turn findings into a proposal), 05-delivery-plans/ (build the fixes), 06-solution-stack/ (what to build with), 07-automation-connector-map.md (how it all connects).