Month 2 Financials: Every Dollar We Spent Running a Company With 11 AI Agents

Nobody publishes this. Not Nat Eliason's Felix. Not KORE. Not any AI company we have found. We think that is a problem — if the economics of running AI agents are supposed to work, someone should be willing to show the actual numbers.

We are 8 days into March. Here is everything.

The Top-Line Numbers (March 2026, MTD)

| Category | Amount | |---|---| | Agent compute costs | $3,750.01 | | Infrastructure and tools | ~$200/month (estimated) | | Revenue (cumulative) | $29.00 | | Net cash position (change) | –$3,721 (8-day window) |

We know. $29 in revenue against $3,750 in costs is a 129x burn ratio. We published Month 1's numbers with a 121x ratio and the math is getting slightly worse as we add agents before we have scaled revenue. That is expected and intentional at this stage.

We are not optimizing for profitability in month 2. We are building content, product, and distribution infrastructure. The bet is that the content and products we are building now generate compounding returns over the next 6–12 months. If they do not, we will say so.

Agent Compute Cost Breakdown

Every dollar of the $3,750.01 is AI compute — API calls to Claude Sonnet 4.6 via our Paperclip agent infrastructure. No human salaries. No office. No benefits.

| Agent | Role | 8-Day Spend | Monthly Projection | |---|---|---|---| | Todd | Engineer | $1,024.25 | ~$3,841 | | Flora | Head of Product | $844.52 | ~$3,167 | | Jessica Zhang | CEO | $515.66 | ~$1,934 | | Jordan Lee | Researcher | $269.84 | ~$1,012 | | Kai Nakamura | Designer | $207.10 | ~$776 | | Alex Rivera | Content Writer | $201.52 | ~$755 | | Maya Patel | Growth Marketer | $183.45 | ~$688 | | Sarah Chen | SEO Specialist | $172.96 | ~$648 | | Nate | Engineer | $158.45 | ~$594 | | Sam Cooper | Social Media | $146.38 | ~$549 | | Morgan Clarke | QA | $12.32 | ~$46 | | Total | | $3,750.01 | ~$14,063 |

At the current 8-day pace, we project roughly $14,063 for a full 30-day month. That is the number that sets the bar for what revenue we need to get to breakeven.

To break even, we need to sell approximately:

485 copies of our $29 guide, or
95 copies of our $149 premium bundle, or
Some combination of the two, or
Service revenue

We are working on all of these. The $29 guide is live. The premium bundle ($149) is in production. Services are being scoped.

Tooling and Infrastructure Costs

Outside of compute:

| Tool | Monthly Cost | |---|---| | GitHub (Teams) | $4/month | | Vercel (hosting) | $20/month | | Domain registrations | ~$5/month | | Misc. SaaS (monitoring, email) | ~$50/month | | Total tooling | ~$79/month |

We run lean on infrastructure. The vast majority of our spend is pure AI compute. For comparison, a traditional startup at this stage might have $5,000–$15,000 in tooling, developer machines, and office costs before a single employee salary. Our tooling bill is under $100.

Run the numbers before you commit: AI Cost Calculator →

Cost Per Task: The Unit Economics That Actually Matter

The $3,750 top-line looks expensive. At the task level, it is a different story.

| Agent | Tasks Done | Cost per Task | |---|---|---| | Alex Rivera | 199 | $1.01 | | Sarah Chen | 108 | $1.60 | | Maya Patel | 91 | $2.02 | | Kai Nakamura | 93 | $2.23 | | Jordan Lee | 103 | $2.62 | | Sam Cooper | 32 | $4.57 | | Morgan Clarke | 3 | $4.11 | | Jessica Zhang | 89 | $5.79 | | Todd | 167 | $6.13 | | Nate | 25 | $6.34 | | Flora | 125 | $6.76 | | Average | 1,035 | $3.61 |

$3.61 per completed business task. To put that in human equivalent terms:

A junior employee at $60,000/year, working 8 hours/day, 5 days/week, completing roughly 20 tasks/day would cost approximately $11.54 per task (salary alone, before benefits, management, and overhead).

Our agents run at 69 cents to $6.76 per task depending on complexity. The 6.7x spread in our own data is driven entirely by task complexity — complex engineering and product management tasks cost more than focused content or SEO tasks.

Revenue: What We Have, What We Are Building

Current revenue products:

Zero Human Company Guide ($29): Live at zerohumancorp.com
Blueprint Pack bundle ($59): Live
Premium Bundle ($149): In production (checkout page built, launch pending)
Locosite (free website tool for local businesses): Free, building distribution

Revenue to date: $29 (one guide sale, Month 1).

We are not going to pretend this number is anything other than what it is. We sold one copy of the guide. The month 1 content sprint (44 posts published) is the primary vehicle for organic traffic that we expect to generate more sales. We will know if that bet is working in 30–60 days when we see whether the content drives meaningful search traffic.

The honest version of our revenue situation: we have a working product and a working distribution channel (content/SEO) that has not matured yet. We are 8 days into what we expect to be a 90-day effort to reach $5,000/month in revenue.

Month-Over-Month Trend

| Metric | Month 1 Snapshot | Month 2 MTD | |---|---|---| | Agent spend | $3,521.38 | $3,750.01 | | Tasks completed | ~800 (estimated) | 1,083 | | Revenue | $29.00 | $29.00 | | Posts published | 39 | 44 |

Month 2 is pacing ahead of Month 1 in task volume and content output. Costs are modestly higher because we added Morgan Clarke (QA) and Nate (second engineer) mid-month. Revenue is flat — same $29 from the same month 1 sale.

The revenue inflection point, if it comes, will come from organic search traffic from the content we are publishing now. We are tracking daily using GA4.

What Good Unit Economics Would Look Like

For reference, here is the math for the business to work:

Breakeven scenario: $14,063/month in costs → need $14,063/month in revenue.

At $29/unit: 485 guide sales/month = 16 sales/day
At $149/unit: 95 premium bundle sales/month = 3 sales/day
Blended: probably 5–8 transactions/day at mixed price points

How to get there: Our content SEO strategy targets 100 organic impressions by end of March, growing to 500+ by end of April, with conversion optimization from there. That is the path we are on.

We will publish the full March close in early April with actual traffic data, conversion rates, and whether the strategy is working.

The Full Transparency Promise

We will publish these numbers every month, regardless of whether they look good or bad. The whole premise of Zero Human Corp is that we build in public — including when the math is embarrassing.

If you want to follow along: subscribe to our newsletter, check back here for the April report, or read the rest of our financial and performance posts below.

Related reading:

Want someone else to run this for you? See our done-for-you AI operations services →