Best harness stacks at $5K, $7K–$15K, $100K monthly revenue

The wrong harness stack is the most expensive mistake a Boss makes in their first year on AI. Not because the tools are pricey. Because the wrong stack quietly trains the team into habits that have to be unlearned at the next revenue band.

A solo on the team stack burns money on collaboration features no one is collaborating with. A small team on the solo stack spends Mondays fixing what should have been audited. An established team on the small-team stack finds the compliance officer won't sign off on anything without SSO and an audit trail.

Three bands. Three stacks. We name the tools, the prices, the tradeoffs, and the moment you outgrow each one. We sell BossMode. We are not pretending to be neutral. We've written this the way we wish other companies wrote theirs about us. When the other side is the right answer, we say so.

Why does revenue dictate the harness?

Revenue is a proxy for three things that decide what stack you can run.

Time supply. A solo at $5K/mo has time to sit with an install, read the standing orders, tune the Pack. A team at $100K/mo doesn't. The founder's hour is the most expensive line item in the company.

Team coordination load. One person doesn't need shared memory or an approval queue. Twelve people do. Features that look like padding at the solo tier are load-bearing at the team tier.

Risk surface. At $5K/mo, the worst case is a misfired email. At $100K/mo, that email goes to a regulator, a big-co buyer, or a public-company partner. The audit trail is the price of admission.

Pick the stack for the band you're in, not the band you wish you were in. Bosses who buy ahead of their stage under-use the tools and then blame the tools.

What's the $5K/mo stack?

Solo Boss. Time-rich, cash-poor. Annual revenue between $50K and $250K. The bleed is leads dying in the inbox, follow-up that takes two days, invoices that ship while money doesn't.

The stack:

• Model surface: Claude Pro at $20/mo. ChatGPT Plus at $20/mo if you write a lot of long-form; otherwise skip.
• Coding harness: Claude Code Pro, included with the $20 Claude Pro subscription. No Cursor yet. You're not shipping enough code for the second harness to pay for itself.
• Workflow runtime: n8n Community Edition, self-hosted, free. Or Make's Core plan at $9/mo if you don't want to run a server.
• Memory + vault: Notion Free, or Obsidian free if you prefer file-based. Pick the one your hand already knows.
• Project management: whatever you're already using. Linear's free tier works. Notion projects work. A text file works.
• Cockpit: BossMode Free. One workspace, one device, local dashboard. No cloud sync, no team, no audit.

Total: $20–$50/mo plus sweat.

Pack the Free Packs that ship with BossMode Free: Day One Operator and Lead Rescue System. Add one standalone Pack to stop the worst bleed: Outbound Engine at beta $197 (was $497), Get-Paid Engine at Case Call-scoped, or Lead Qualifier Engine at Case Call-scoped.

What this stack sacrifices:

• No SSO. You're the only login.
• No audit log. You'll feel this the day a buyer asks "what did your AI send my team?" and you can't answer with receipts.
• No team sharing. The memory is yours. If a contractor joins for a project, you're emailing them screenshots.
• No retainer support. You're the one keeping the Pack tuned. The Tune retainer at Case Call-scoped DWY is more than your tooling stack costs combined; you can't justify it yet.

This stack works because the Boss reads the playbook. It does not work for a Boss who wants the tools to do the thinking.

What's the $7K–$15K/mo stack?

Small team. Three to twelve people. Annual revenue between $300K and $1.5M. Someone on the team can write a prompt without crying. The bleed has moved upstream: signed deals stalling before kickoff, discovery sessions going in unprepped, content getting created and never multiplied.

The stack:

• Model surface: Claude Code Max at $100/mo for the Boss. ChatGPT Plus at $20/mo for content work. Gemini if you're heavy on Google Workspace; otherwise skip.
• Coding harness: Cursor Pro at $20/mo for one or two engineers, or Pro+ at $60/mo if any ship code daily. Claude Code Max stays primary; Cursor handles IDE-shaped work.
• Workflow runtime: n8n Cloud Pro at €60/mo (~$65). 10,000 executions, unlimited users and workflows. The monthly hour you'd spend maintaining your own n8n server costs more than the tier.
• Memory + vault: Notion Team plan, or a shared Obsidian vault on sync. Linear's team plan handles project memory if engineering is the gravity center.
• Project management: Linear team plan if you ship product. Notion projects if not.
• Cockpit: BossMode Cockpit access through Fly AI for two or three people, or Studio at $149/mo for a real five-person team. Studio adds shared Packs and shared memory the team can edit. Operator caps at one operator with three workspaces; Studio is where teams live.

Total: $250–$400/mo, depending on seats and Cockpit tier. Add a Tune retainer at Case Call-scoped DWY if you want the Packs maintained as the business changes; total goes to $2,750–$2,900/mo all-in.

Pack the load-bearing recipes for this band: PM Engine at beta $197 (was $499) self-install or DFY scoped on a Case Call, Client Kickoff System when the Case Call names that bleed, Content Multiplier at Case Call-scoped, LinkedIn Authority Engine at $497. Add Outbound Engine and Lead Qualifier if outbound is the bleed.

What this stack sacrifices:

• Dedicated infrastructure. n8n Cloud Pro and BossMode Studio run on shared multi-tenant infra. Fine for 99% of Bosses in this band. If your data classification says otherwise, you're already shopping the next tier up.
• Custom limits. n8n Cloud Pro caps at 10,000 executions; Cloud Business jumps to 40,000 at €800/mo. No in-between tier. Bosses in the upper half of this band sometimes hit the ceiling and make the jump early.
• SSO, unless you're on Scale. Operator and Studio don't include SSO; it arrives at Scale ($499/mo). If your buyers ask for SSO, you're moving up early.
• Multi-region data residency. Same answer as dedicated infra: not until Scale or Enterprise.

This stack is where most Bosses land once they admit they have a team but not an AI ops lead. The Tune retainer at Case Call-scoped DWY is the line item that decides whether the stack stays sharp or quietly drifts. Most teams who skip it come back six months later asking why their Outbound Engine stopped booking calls. The world changed. Their Pack didn't.

What's the $100K/mo stack?

Established team. Ten to fifty people. Annual revenue between $1.2M and $6M. You have a CFO, a head of ops, and at least one regulator you take seriously. The bleed isn't a single broken workflow; it's that the whole shop is held together by tribal knowledge and the founder's calendar.

The stack:

• Model surface: Claude Code Max at $200/mo across the team. Multiple seats. You've stopped counting the model bill the way you've stopped counting the office coffee bill.
• Coding harness: Cursor Pro+ at $60/mo per seat for engineering. Some add Cursor Ultra at $200/mo for lead engineers who live in it. Heavy users see API overage; budget $500–$1,500/mo of headroom.
• Workflow runtime: n8n Cloud Business at €800/mo (~$870). 40,000 executions, SSO, audit log. Or Make Teams plan at $29/mo per user, depending on which side of the n8n vs Make line your team chose earlier. Regulated workloads land on n8n Cloud Business or self-hosted Enterprise; marketing-shaped workloads stay on Make.
• Memory + vault: Linear team plan plus Notion team plan. Linear holds the work; Notion holds the playbooks. Some teams swap Notion for Obsidian Sync at $4/mo per user for IP custody.
• Project management: Linear team plan, integrated with Slack and the harness. The PM Engine Pack at $197 beta self-install (was $499)-install or DFY scoped on a Case Call runs on top of Linear and converts commitments into chased and closed work.
• Cockpit: BossMode Scale at $499/mo. Unlimited team members, advanced approvals, audit and SSO ready, priority support. Or custom Enterprise for dedicated infrastructure, custom limits, security review, launch support.

Total: $1.5K–$3K/mo for tooling, plus model and API costs that vary by team activity. Add an Operator retainer at $5,000/mo or a Bespoke retainer at $10,000/mo if you want a human on call. All-in stack runs $7K–$15K/mo depending on retainer and team size.

Pack the Trust Pack at $14,997+ DFY: 12 component Packs wired as a unified OS, 48+ standing orders, 90 days of operator support. Add the PM Engine and any specialty Packs the Case Call names. Bosses in this band rarely buy Packs one at a time; they buy the wired stack.

What this stack sacrifices:

• Still not Enterprise dedicated infrastructure, unless you're on a custom plan. Scale at $499/mo is multi-tenant, audit-ready, SSO-ready. It is not air-gapped, single-tenant, or security-reviewed for your specific compliance posture. If your auditor needs that, you're on Enterprise.
• No public price on Enterprise. Bespoke engagements start at $7K–$15K and climb based on integration count, regulated-industry overhead, and IP custody arrangements.
• The DFY install fee, starting at $10K. The cost of not learning the tools yourself. For a Boss at $100K/mo, a week of revenue. For most teams at this band, the right call. For some, not.

This is what an established team runs when the founder has stopped trying to be the AI ops lead. The Cockpit sits over the harness, the harness sits over the work, and the work flows through Packs audited against compliance posture. The retainer is the human who keeps it honest.

What does each tier sacrifice?

The $5K/mo stack sacrifices SSO, audit, team sharing, retainer support, and any pretense of compliance posture. In return, it costs less than a domain renewal. The Boss's time is the load-bearing budget item, not their cash.

The $7K–$15K/mo stack sacrifices dedicated infrastructure, custom execution limits, multi-region data residency, and SSO. It buys shared workspaces, an approval queue, shared Packs, and a Tune retainer if you want one. The team's coordination capacity is the load-bearing budget item.

The $100K/mo stack still sacrifices dedicated single-tenant infrastructure unless on Enterprise. It buys SSO, audit, advanced approvals, priority support, and either an Operator or Bespoke retainer. The audit trail and the regulator's signature are the load-bearing budget items.

Notice what doesn't change across tiers: the Pack catalog. The standalone Packs at the $5K/mo tier are the same Packs bundled into the Trust Pack at the $100K/mo tier. The recipes don't change. The harness changes.

When do you upgrade between tiers?

Four signals, in order of clarity.

Time spent operating the harness exceeds the cost of the next tier. Cleanest signal. If you're spending eight hours a week wrangling your stack and the next tier costs $300/mo more, the math is done. You're already paying that cost in time; you haven't moved the line item to the right column. Most Bosses wait too long here.

Team grows past the seat count. BossMode Operator caps at one operator. Studio caps at five team members. Scale is unlimited. n8n Cloud Pro and Business have execution caps, not seat caps, but the practical effect is the same. The day you onboard the seat that doesn't fit, you upgrade.

Compliance or a regulated industry forces SSO and audit. A buyer asks for SSO. An auditor asks for an audit log. A compliance officer asks who can see what. The answer at the $7K–$15K/mo tier is "we can't show you that yet." At the $100K/mo tier it's "here's the export." If the question gets asked, you upgrade.

Pack stack exceeds three or four simultaneous Packs. One Pack is cheap anywhere. Three is the inflection point. Four-plus is where unified orchestration earns its keep. Bosses running four-plus Packs without the Trust Pack architecture spend more time managing the seams than running the work.

These signals arrive together. Solos who hit team-of-three usually hit the seat-count and time-spent signal in the same quarter. Small teams hitting the SSO conversation are usually within sixty days of the four-Pack threshold.

What's the migration path?

Three moves. Each takes a week to plan and an afternoon to execute.

$5K to $7K–$15K: add the Cockpit tier and the first non-founder seat. Move from BossMode Free to Operator ($49/mo) on the day a virtual assistant or contractor joins. Move from Operator to Studio ($149/mo) on the day a full-time team member joins. Add Cursor Pro for the first engineer writing more code than prompts. Move from self-hosted n8n to n8n Cloud Pro on the day you're tired of maintaining the server, not the day you're tired of the bill. The trigger is operational pain, not budget.

$7K–$15K to $100K: add the audit trail, the SSO requirement, the retainer. Move from Studio to Scale ($499/mo) the week the SSO conversation becomes real. Move from n8n Cloud Pro to Cloud Business the quarter you cross 8,000 executions per month. Hire the Operator retainer ($5,000/mo) the week the founder stops being the AI ops lead. The Trust Pack at $14,997+ DFY is the Pack-side mirror; the Case Call names which component Packs ship first.

$100K to Enterprise: have the security conversation. Not a self-serve checkout. A quoted engagement. Bespoke installs start at $7K–$15K. The Bespoke retainer at $10,000/mo includes the human who handles compliance reviews, custom integrations, IP custody, and regulated-industry overhead the audit-ready tier doesn't cover. If you're asking whether you need it, take the Case Call and find out.

The migration stays easy because the Packs travel with you. The Outbound Engine you installed at $5K/mo runs the same way at $100K/mo, with more humans approving more outputs through more workspaces. The harness changes. The Cockpit changes. The Packs are stable because the bleeds are stable.

You're the Boss. You tell the Co-pilot where to go. Pick the stack that fits the size of the operation you're running, then upgrade when the operation forces the upgrade. Not before.

Stop using AI. Start running your business on it.