Luminary Diagnostics AI LLC

Words in [brackets] are blanks you fill in with your own details when you adopt this sample in Supreme Plan AI.

Executive Summary

Luminary Diagnostics AI LLC is a company in North Carolina developing Luminary Rad Assist, an AI software tool that helps radiologists detect early signs of lung nodules in chest CT scans. The tool analyzes each scan, highlights regions of interest, scores the likelihood that a finding warrants follow-up using a published scoring system and presents a structured report for the radiologist to review and modify. The radiologist makes every clinical decision. The tool is designed to support, not replace, physician judgment.

Early detection of lung cancer is one of the clearest opportunities to save lives through imaging AI. Lung cancer is the leading cause of cancer death, and low-dose CT screening can find it at a stage when treatment is far more likely to succeed. Radiologists face high scan volumes, long shifts and pressure to read quickly. AI that reliably reduces miss rates and helps prioritize the reading queue can make a meaningful difference, if it is accurate, well-tested and properly integrated into a clinical workflow that keeps the physician in charge.

Our Year 1 plan is to complete IRB-approved retrospective validation on a de-identified dataset, begin prospective evaluation at one health system partner and advance our FDA regulatory strategy. Revenue is planned at about $200,000 from hospital and health system contracts and about $60,000 from data licensing and validation services in Year 1, growing to about $2,000,000 and $500,000 by Year 5. Our Year 1 budget is about $1,705,500. Grants and SBIR or STTR awards are planned to fall from about 85% of revenue in Year 1 to about 25% in Year 5 as commercial revenue grows.

We do not promise that our tool will reduce miss rates or improve outcomes in any specific deployment. Our clinical study is designed to measure that honestly, and we will publish results whether they support or complicate our hypothesis.

I am Dr. Sophia Park, Chief Executive Officer, a radiologist and AI researcher. The numbers in this proposal are estimates based on the assumptions listed at the end, not promises.

Company Description

Luminary Diagnostics AI LLC is a limited liability company (LLC) in North Carolina formed to develop and commercialize AI software for diagnostic imaging. We are a for-profit small business with fewer than 25 employees. We will provide our formation documents, a good standing certificate, our federal and state tax registrations, our insurance certificates and our most recent financial statements with this proposal. We will confirm that we meet the Breakthrough Health Technology Fund's size and eligibility rules before we accept any award.

Our mission. To give radiologists a reliable second reader, so that more lung cancers are found at a stage when lives can be saved.

What we believe. AI in clinical settings must earn trust through evidence, not marketing. A tool that assists a physician is only as good as the study that tests it, the transparency of its reasoning and the rigor of its regulatory pathway. We will not make claims we have not measured, and we will publish our validation results whether they are positive, negative or mixed.

How we started. Our founders, a radiologist and an AI researcher, began collaborating during a fellowship on a dataset of de-identified chest CTs. They trained a model that performed well on internal data and then tested it on an external dataset, where performance was measurably lower in a specific demographic group. That finding shaped our research agenda: we now design every study to test subgroup performance, and we designed the model's interface to show the radiologist exactly what the tool found and why it scored it, rather than giving only a number. A hospital partner who saw the tool offered a clinical evaluation agreement. This proposal asks for support to complete that evaluation and to advance our FDA pathway.

What we offer. A software module, integrated into existing radiology workflow software, that reads a chest CT, identifies regions of interest, scores each finding using the Lung-RADS system and generates a structured addendum that the radiologist reviews and edits before it becomes part of the report. The tool records every case it processes, every score it assigns and every change a radiologist makes, so we can measure its accuracy and the radiologist's level of agreement continuously. Performance data is available to the health system at any time.

Who we serve. Hospitals, health systems and radiology groups that run lung cancer screening programs and want independent, measured AI assistance. We do not sell directly to patients. Our contracts are with the health system, and patients interact only with their own physician.

What we are not. We are not a medical device that makes a diagnosis. Every output of our tool is a recommendation that a licensed physician reviews and can overrule. We are not cleared by the FDA for commercial sale, and we will not sell the tool for clinical use until we have the appropriate FDA authorization. Our current work is under IRB-approved study protocols.

How we are run. A founding leadership team of a CEO and a CMO, a board of advisors that includes a radiologist in private practice, an oncologist, a health system CIO, a regulatory attorney and a biostatistician, and a Clinical Advisory Board of three practicing radiologists. An outside accountant reviews our books each year.

Where things stand today. We have a clinical evaluation agreement with one health system. We have an IRB-approved retrospective study protocol and have begun data de-identification. We have a regulatory consulting firm under contract for our FDA 510k pathway analysis. We have not yet received FDA clearance or authorization, and we do not sell for clinical use without it.

Problem & Opportunity

Lung cancer kills more people each year than any other cancer. The majority of lung cancers are still found at a late stage, when curative treatment is far harder. Low-dose CT screening is recommended for high-risk adults, and the evidence shows it saves lives. The challenge is not the technology of the scan: it is what happens after the images are acquired.

Radiologists read thousands of scans. A missed lung nodule at an early stage can mean a patient is not offered follow-up imaging until the cancer has grown. False positives send patients through unnecessary procedures, with their attendant costs, stress and, occasionally, complications. The literature documents that radiologists miss a meaningful share of small nodules, particularly under time pressure and toward the end of a reading shift. At the same time, readers vary: what one radiologist recommends for a three-millimeter nodule, another may classify differently using the same published guidelines.

AI tools for chest imaging have been proposed and built for more than a decade. Several are on the market. The majority have been validated primarily on datasets that do not reflect the full diversity of the patient population, including patients with different body sizes, different imaging equipment and patients from demographic groups underrepresented in large academic datasets. When a tool performs differently in a community hospital than in the academic center where it was trained, the difference is often attributed to the hospital rather than the tool. We believe this is backwards, and we design our validation to find it.

We want to be honest about the risk. An AI tool for radiology that is marketed without rigorous validation, or that is validated on a homogeneous dataset, can increase miss rates in some populations rather than reduce them. A tool whose interface gives only a probability score, without showing why it scored the case that way, can lead a radiologist to defer to the tool even when the tool is wrong. We design against both of these failure modes, and we will say so in our published results.

Products & Services

Luminary Diagnostics AI LLC will complete IRB-approved retrospective validation, run a prospective evaluation at one health system partner and advance our FDA regulatory pathway in the grant period. The grant funds the clinical study, the FDA regulatory work and the engineering needed to support both.

How the work proceeds:
1. Retrospective validation. We use de-identified CT data from our health system partner under an IRB-approved protocol and a signed HIPAA business associate agreement. We test the model on data not used in training, measure sensitivity and specificity by Lung-RADS category, and break results down by age, sex, body mass index and imaging equipment type. An independent biostatistician analyzes the data. We register the study protocol and primary endpoints before data analysis begins.
2. Subgroup analysis and model adjustment. Where subgroup performance is measurably lower, we investigate the cause: whether the training data was unrepresentative, whether the model relies on a feature that varies across groups, or whether the scoring threshold needs adjustment. We document every finding and decision.
3. Prospective evaluation. Under a separate IRB protocol and clinical evaluation agreement, radiologists at our partner site use the tool alongside their normal reading, with a randomized read design that the biostatistician designs to reduce workflow confounds. We measure radiologist agreement, flagged findings, read time and any errors in both directions.
4. FDA pathway. Our regulatory consulting firm is analyzing the appropriate FDA pathway, likely a 510k predicate or a De Novo request, and will advise on whether our clinical evidence meets FDA's standard for substantial equivalence or a new class. We will not commercialize the tool until we have the appropriate FDA authorization.
5. Publication. We will submit results for peer review regardless of direction and publish a model card that describes performance by subgroup.

What the grant will pay for. One Machine Learning Engineer for further model development and testing, IRB fees and clinical site payments for the retrospective and prospective studies, regulatory consulting for the FDA pathway and additional cloud compute for model training and the secure medical data environment. The grant does not pay for marketing or general sales.

Human decisions. Every output is presented to a radiologist as a recommendation. The radiologist may agree, disagree or modify it, and the tool records the decision. We do not allow the tool to send a recommendation directly to a patient, to a referring physician or to any clinical record without radiologist review.

Independence of the study. Our biostatistician works under contract, is not a shareholder and is paid the same whether results are favorable or not. The IRB at our partner site reviews and approves every protocol change. We will not adjust analysis methods after seeing the data.

Market Analysis

Who benefits: radiologists who read lung cancer screening programs and the patients whose cancers are caught earlier.

Our direct customers are hospitals, health systems and large radiology groups that run established lung cancer screening programs. Our first commercial target is the health system we are studying with now. Secondary targets are regional and community health systems that run screening programs and do not have in-house AI resources.

Market size. The United States has several hundred hospital systems and several thousand radiology practices. We will use CMS data and published estimates of lung cancer screening program volume in our final application and will not overstate the market. Adoption of AI in radiology is real but slower than many projections suggest, because radiologists, hospitals and payers each have adoption timelines and because FDA clearance is a prerequisite for most hospital purchasing decisions.

Pricing. We charge a per-study fee or an annual platform fee, depending on the health system's volume and contract structure. We do not charge patients directly. We expect the tool to be revenue-positive for a health system only if it demonstrably supports faster, more consistent reads at a measurable level of quality, and we design our evaluation to test that.

We do not guarantee any customer a clinical outcome, a reduction in miss rates or a return on investment. We offer a measured tool, published evidence and the ability for the health system to see the performance data in real time.

Competitive Landscape

Other AI lung nodule tools on the market and where we are positioned.

- Established commercial AI tools for chest CT. Several are FDA-cleared and in use at major academic centers. They are well-resourced and widely distributed. We differ by focusing on subgroup equity in validation, by showing the radiologist the specific finding and the reason it was scored, and by providing continuous performance monitoring.
- General-purpose AI platforms for radiology. Large imaging AI companies offer platforms with many models, including lung. They serve large systems well and require significant IT integration. We are a focused tool, easier to integrate and to study.
- Internal AI programs at large academic health systems. Some large systems build their own models. This is not a path most community hospitals can take.
- No AI, current practice. This is the dominant choice. The cost of current practice is not zero: it is miss rate, variability and radiologist workload. We aim to show that the cost of a well-validated tool is lower.

What makes us different.
- Subgroup performance is a design criterion, not an afterthought.
- The tool shows its reasoning, so the radiologist can disagree intelligently.
- Continuous performance monitoring so the health system knows if accuracy changes.
- Pre-registered study endpoints and independent biostatistics.
- A commitment to publish results regardless of direction.

What we do not claim. We have not yet shown in a prospective study that our tool reduces miss rates in a community hospital setting. That is what the grant is for.

Marketing & Sales Strategy

How we will reach customers, clinical partners and funders.

Clinical partners. We will reach health systems and radiology groups through the radiologists and hospital leaders we know from our founders' clinical and research networks, national radiology society meetings and through the lung cancer screening program infrastructure at the American College of Radiology. Every new site starts with a clinical evaluation agreement before any commercial contract.

Regulatory path. Achieving FDA authorization is a condition for hospital purchasing in most systems. We will complete our regulatory analysis and submit our application as soon as our clinical evidence meets the standard.

Payers. We will engage Medicare, Medicaid managed care plans and commercial insurers about coverage for AI-assisted reading in the lung cancer screening benefit. We will not promise customers that payer coverage exists until it does, and we will document coverage status for each customer.

Publication and scientific community. We will submit our validation results for peer review in a radiology or clinical AI journal, and present at radiology and health AI conferences. We believe that published, peer-reviewed evidence is the most credible form of marketing for a clinical AI tool.

Investors. We will use the grant period to build the clinical evidence and regulatory progress that support our first equity fundraising round. The grant does not fund investor relations activity.

Operations Plan

How the work will be carried out.

Timeline for Year 1:
- Months 1 to 3: Hire the Machine Learning Engineer and the Clinical Research Coordinator, finalize the retrospective study data de-identification under the HIPAA BAA and IRB protocol and engage the regulatory consulting firm for the FDA pathway analysis.
- Months 4 to 6: Complete retrospective validation, conduct subgroup analysis and document all findings, including those that require model adjustment.
- Months 7 to 9: Begin the prospective evaluation with randomized read design. Start FDA 510k application preparation based on regulatory consulting guidance.
- Months 10 to 12: Complete prospective study enrollment, analyze results with the independent biostatistician and submit results for peer review.

Products and infrastructure. We run on HIPAA-compliant cloud infrastructure with a signed business associate agreement, encryption at rest and in transit, role-based access and audit logs. Clinical data is stored in a dedicated, access-controlled environment, used only for the purposes in the IRB protocol and the data-use agreement and deleted on the schedule the agreement requires. Cloud computing and data storage costs about $100,000 in Year 1.

IRB and research ethics. Every use of patient data, de-identified or otherwise, is covered by an IRB-approved protocol, a signed data-use agreement and, for individually identifiable data, a HIPAA business associate agreement. We use only retrospective de-identified data in the training phase and IRB-approved prospective data in the evaluation. We do not use patient data for training a model that is separate from the study. The IRB at our partner site is independent of our company.

FDA regulatory pathway. We are not FDA-cleared and we do not sell for clinical use without the appropriate FDA authorization. Our regulatory consultant will determine the correct submission type (510k, De Novo or other). We follow FDA's framework for AI and machine learning as a software as a medical device (SaMD). We will disclose our regulatory status to every health system partner and to the Breakthrough Health Technology Fund.

HIPAA and cybersecurity. We follow HIPAA Security Rule requirements and conduct an annual risk analysis. A HIPAA Security Officer is designated. We carry out an independent cybersecurity audit and penetration test before any prospective clinical evaluation involving identifiable data. Cybersecurity, HIPAA and compliance costs about $40,000 in Year 1.

Clinical advisory board. A group of practicing radiologists advises on the clinical design of the study, the tool interface and the interpretation of results. Advisory board members are compensated fairly, disclose any financial relationships with the company and recuse themselves from any evaluation where a conflict exists. Their fees are about $30,000 in Year 1.

Legal and intellectual property. Our attorney reviews all clinical site agreements, data-use agreements, HIPAA BAAs and IP assignments. We file patent applications where counsel advises, and we disclose all open-source components we use. Legal costs are about $50,000 in Year 1.

Insurance and records. We carry general liability, professional liability (errors and omissions), cyber liability, clinical trial liability and directors and officers coverage. Insurance is about $40,000 a year in Year 1. Our outside accountant reviews our books every year, and we keep grant money in a separate account.

Management & Organization

The people who will carry out the project.

Dr. Sophia Park, Chief Executive Officer. A board-certified radiologist and AI researcher with more than ten years of experience reading chest imaging and building imaging AI models. She leads clinical design, fundraising, partnerships and hiring, reports to the board and is the main contact for funders. (Replace this paragraph with your own leadership's background.)

Chief Medical Officer and Principal Investigator. A physician researcher who is the IRB-listed principal investigator for our clinical studies, oversees the clinical evaluation design and is the primary relationship with our health system partner's radiology leadership.

Machine Learning Engineers (two in Year 1, three in Year 2, four in Year 3, five in Year 4 and six in Year 5). Build and maintain the imaging AI model, the inference pipeline, the performance monitoring dashboard and the integration with radiology workflow software.

Clinical Research Coordinators (one in Years 1 and 2, two from Year 3). Manage IRB submissions, data-use agreements, patient consent (where applicable under the prospective protocol), site coordination and clinical data quality.

Regulatory Affairs Specialist (one in Years 1 to 3, two from Year 4). Owns the FDA submission, monitors regulatory developments for AI-based medical devices and manages regulatory correspondence.

Software Engineers (two in Years 1 and 2, three in Year 3, four in Year 4 and five in Year 5). Build the product interface, the structured reporting module, the performance monitoring dashboard and hospital system integrations.

Bookkeeper (part-time). Keeps the books and works with our outside accountant.

Staff pay in this plan includes payroll taxes and benefits. Salaries are CEO $165,000, CMO $200,000, Machine Learning Engineer $155,000, Clinical Research Coordinator $75,000, Regulatory Affairs Specialist $110,000, Software Engineer $135,000 and part-time Bookkeeper $40,000.

Milestones Narrative

Our goals for each year and how we will measure them.

- Year 1: Complete IRB-approved retrospective validation and begin prospective evaluation at one health system, advance FDA 510k analysis and submit results for peer review. A Year 1 budget of about $1,705,500 in revenue, with profit after income tax of about $47,400. 9 paid staff.
- Year 2: Complete prospective evaluation, submit FDA application, publish peer-reviewed results and sign a second health system partner. A budget of about $1,927,000 in revenue, with profit after income tax of about $48,900. 10 paid staff.
- Year 3: Receive FDA authorization and begin first commercial deployments. Expand to a third health system and build the continuous monitoring dashboard. A budget of about $2,372,000 in revenue, with profit after income tax of about $53,900. 13 paid staff.
- Year 4: Grow to about five health system deployments and begin the next model development cycle, targeting a second imaging application. A budget of about $2,864,500 in revenue, with profit after income tax of about $60,900. 16 paid staff.
- Year 5: Reach about ten health system deployments, with grant and SBIR revenue covering a minority share of the budget as commercial revenue grows. A budget of about $3,334,500 in revenue, with profit after income tax of about $66,900. 19 paid staff.

What we measure:
- Clinical study progress: IRB approvals, data de-identified, scans analyzed and radiologist reads enrolled.
- Model performance: sensitivity and specificity by Lung-RADS category and by subgroup, compared against the pre-registered primary endpoints.
- Radiologist agreement rate and the rate at which radiologists override the tool, with reasons documented.
- Regulatory progress: FDA submission status and correspondence timeline.
- Revenue by source, gross margin and cash runway.
- Health system deployments and study sites signed.

How we review results. The CMO and the biostatistician review study data monthly, the full leadership team reviews outcomes and finances monthly and the board of advisors reviews scientific progress and regulatory status quarterly. We will report to the Breakthrough Health Technology Fund twice a year and in a final report at the end of the grant period.

Funding Request

What we are asking for, and how the money will be used. Our request: $200,000 from the Breakthrough Health Technology Fund for one Machine Learning Engineer, IRB fees and clinical study operations, regulatory consulting and cloud computing for the clinical studies, for July 2027 through June 2028.

The request is spread across these budget lines:
- Personnel: Machine Learning Engineer salary, $60,000 (other funds add $95,000).
- Other: IRB fees, clinical study operations and site payments, $60,000 (other funds add $20,000).
- Contractual: Regulatory consulting (FDA and SaMD pathway), $50,000.
- Other: Cloud computing, model training and medical data storage, $30,000 (other funds add $70,000).

Other money in the program. Our own revenue, other grants and SBIR or STTR awards and investment we are raising pay the other $185,000 of those lines, plus the leadership salaries, additional engineering, regulatory affairs, cybersecurity audits, clinical advisory board, legal, sales and marketing and office costs. Investment is not yet committed and is not counted as received in this plan.

Year 1 budget: about $1,705,500 in revenue. Year 1 money coming in: grants and SBIR or STTR revenue $1,445,500 (this request plus other awards we are applying for), hospital and health system contracts $200,000 and data licensing and validation services $60,000. Year 1 costs: Chief Executive Officer $165,000; Chief Medical Officer / Principal Investigator $200,000; two Machine Learning Engineers $310,000; Clinical Research Coordinator $75,000; Regulatory Affairs Specialist $110,000; two Software Engineers $270,000; Bookkeeper $40,000; cloud computing, model training and medical data storage about $100,000; irb fees, clinical study operations and site payments about $80,000; regulatory consulting about $50,000; office, lab rent and utilities about $40,000; clinical advisory board and medical expert fees about $30,000; cybersecurity, hipaa and compliance audits about $40,000; sales and marketing about $45,000; legal about $50,000; travel and conference about $20,000; recruiting about $25,000; and about $21,200 a year for the wear on equipment.

Grant money is a planning estimate: the grant line in the plan is the balance needed after commercial revenue, and it is not a promise from any funder. If a contract is delayed or not signed on schedule, we will reduce hiring and scope before we spend money we do not have.

How we planned the clinical study budget. IRB fees, clinical site payments and study operations cost about $80,000 in Year 1, or about what a single-site retrospective and prospective study at an academic health system typically costs when the study involves de-identified data, a contracted biostatistician and site coordinator support. We expect this to grow as we add sites. Regulatory consulting is about $50,000 in Year 1 for the 510k analysis, gap assessment and submission preparation.

About government contracts and SBIR. The NIH, NCI, NLM and NIST fund AI in health research, and the SBIR and STTR programs provide non-dilutive capital to small businesses doing this kind of work. SBIR Phase I awards are typically around $300,000 and Phase II around $1,000,000. We are applying and do not count any specific award in this plan beyond what is shown. These funds are competitive, peer-reviewed and tied to technical milestones, not revenue.

One-time startup costs. Startup costs of $135,000 are paid once from startup money, not from the yearly budget: $40,000 for equipment, $10,000 for leasehold improvements, $10,000 for deposits, $15,000 for opening supplies and inventory, $5,000 for licenses and permits, $35,000 for legal and professional help, $8,000 for signage, brochures and a website, $12,000 for background checks, training and first insurance.

How we got our startup number. The Funding Need Calculator in this plan adds the startup costs ($135,000), the cash we need to cover the wait for the first payments (about $183,400) and a cash cushion of about $412,500, which is about three months of Year 1 costs. That comes to $731,000. We start with $80,000 of our own cash and $680,000 in startup support from our founders, angel investors and a health system innovation fund, $760,000 in all, which covers that amount.

The next five years. Revenue grows from about $1,705,500 in Year 1 to about $3,334,500 in Year 5 as commercial hospital contracts grow after FDA authorization. Grant and SBIR revenue falls from about 85% of revenue in Year 1 to about 25% in Year 5. We add engineering, clinical and regulatory staff in each year to support growth.

Risks and how we will handle them:
- The clinical study shows lower performance than expected in a subgroup. We document the finding, investigate the cause, adjust the model and retest before any commercial deployment.
- FDA review takes longer than planned or requires additional evidence. We plan with cash cushion, have a realistic timeline from our regulatory consultant and do not promise a clearance date to customers or funders.
- A health system does not renew or delays a contract. We maintain multiple partner conversations, do not count revenue from a contract until it is signed and have a cash cushion for delays.
- A cybersecurity or data breach occurs. We carry cyber insurance, conduct an annual penetration test, follow the HIPAA Breach Notification Rule and have a written incident response plan.
- A grant or award is not received. We stage hiring so costs follow funding and apply to multiple sources.

Savings. We plan to keep several months of costs in cash. The lowest year-end cash balance in the plan is about $533,200, at the end of Year 4. Customers and funders often pay after we invoice or submit reports, so we plan for about $238,770 owed to us at the end of Year 1.

If you are just starting. A new company can begin with a founder-funded retrospective study on a public or shared dataset, a regulatory analysis and a clinical evaluation agreement, and can apply for SBIR or STTR Phase I before seeking a larger grant. Strong, published retrospective evidence and a signed health system partner make larger awards more competitive.

Appendix Notes

Documents to attach with this proposal:

- Formation documents, a good standing certificate and the operating agreement.
- Your most recent financial statements, tax returns and current year budget.
- A list of owners, officers and advisors, with their roles and any conflicts of interest.
- Resumes for the CEO and CMO and relevant publications or prior patents.
- Your IRB approval letter and study protocol for the retrospective study.
- Your clinical evaluation agreement and signed HIPAA business associate agreement with the health system partner.
- Your regulatory consulting engagement letter and preliminary FDA pathway analysis.
- Your data-use agreement template and HIPAA security policies.
- A description of your model and the performance measures you will report.
- Proof of general liability, professional liability, cyber liability and clinical trial liability insurance, or quotes.
If you don't have any of this yet, no worries - let's continue.

The assumptions behind our numbers (change them in the Financials section to match your own):
- Grant and SBIR or STTR Revenue: $1,445,500 in Year 1, then $1,317,000, $1,232,000, $1,104,500, $834,500.
- Hospital and Health System Contracts: $200,000 in Year 1, growing to $2,000,000.
- Data Licensing and Validation Services: $60,000 in Year 1, growing to $500,000.
- Pay: CEO $165,000, CMO $200,000, ML Engineer $155,000, Clinical Research Coordinator $75,000, Regulatory Affairs Specialist $110,000, Software Engineer $135,000, Bookkeeper (part-time) $40,000, with headcount growing as shown.
- Other yearly costs in Year 1: Cloud Computing, Model Training and Medical Data Storage $100,000, IRB Fees, Clinical Study Operations and Site Payments $80,000, Regulatory Consulting (FDA 510k and SaMD) $50,000, Office, Lab Rent and Utilities $40,000, Clinical Advisory Board and Medical Expert Fees $30,000, Cybersecurity, HIPAA and Compliance Audits $40,000, Sales and Marketing $45,000, Legal (IP, Contracts and Regulatory) $50,000, Travel and Conference $20,000, Recruiting $25,000. Growing between 5% and 22% a year.
- Equipment: $40,000 at the start, then $15,000 to $18,000 a year.
- Startup costs: $135,000, paid once from $760,000 in startup money.
- About 14% of each year's revenue is still owed to us at year end.
- Tax: 21 percent federal and state income tax on profit.
- Every dollar amount is an estimate. Replace them with your own quotes and real budget.

Startup Costs