BrokerLens

Transparent business valuation, backed by real market research.

BrokerLens turns owner-provided financial information, business-quality factors, and sourced market research into a transparent preliminary business valuation. Instead of producing only a price, it shows the complete reasoning behind the result.

Inspiration

My inspiration for BrokerLens came from my dad.

He has sold a business before and is currently trying to sell another one. In both cases, one of the hardest questions was also one of the most important:

What is the business actually worth?

Different people could suggest different asking prices or industry multiples, but it was difficult to understand where those numbers came from. Business owners may understand their operations extremely well while still lacking access to expensive valuation tools, completed-sale databases, and clear explanations of how different risks affect the price.

Watching my dad struggle with this process made me realize that he needed a tool like BrokerLens—not another calculator that produces a mysterious number, but a tool that explains how the value was calculated and which evidence supports it.

What it does

BrokerLens guides the user through a four-step valuation process covering the business, its financials, quality and risk factors, and current market conditions.

The owner provides information including:

• Industry and location
• Net profit and owner compensation
• Interest, depreciation, and verified add-backs
• Revenue growth and recurring revenue
• Owner dependence
• Largest-customer concentration
• Remaining lease term
• Inventory, FF&E, real estate, excess assets, and assumed debt

BrokerLens first calculates normalized Seller's Discretionary Earnings:

SDE = Net Profit + Owner Salary + Interest + Depreciation and Amortization + Verified Add-backs

Every business begins with the same universal 3.00× base multiple. BrokerLens then applies deterministic business-quality adjustments based on the information entered by the owner.

AI researches exactly four market factors:

1. Industry transaction multiples
2. Local demand
3. Labor conditions
4. Competition

After the research succeeds, BrokerLens reveals the final multiple, estimated market-value range, suggested asking price, likely sale range, and a complete explanation of why the multiple moved.

The final value also accounts for assets and liabilities:

Business Value = (SDE × Final Multiple) + Inventory + FF&E + Real Estate + Excess Assets − Debt Assumed

How I built it

I built BrokerLens as a full-stack web application using:

• Next.js 16
• React 19
• TypeScript
• Custom responsive CSS
• OpenAI Responses API
• Vercel

The financial valuation engine is deterministic. TypeScript calculates SDE, business-quality adjustments, valuation ranges, asset adjustments, and the final estimated value.

AI is used only as a controlled market-research layer. A server-side API route requests structured results for the four required market categories.

The server independently validates the AI response by:

• Requiring exactly four research categories
• Restricting research to user-approved domains
• Validating every source URL
• Grading transaction evidence by quality
• Enforcing adjustment limits
• Assigning a neutral adjustment to unsupported evidence
• Calculating the total AI adjustment instead of trusting an AI-generated total

Each research card lists all supporting websites used for that factor. The shortened links remain clickable and open the complete source pages.

BrokerLens also includes browser-local saved projects, dark mode, formatted financial inputs, a revenue-growth calculator, custom-industry matching, responsive design, and server-side protection for the OpenAI API key.

Challenges I faced

One of the biggest challenges was deciding how much control AI should have over a financial valuation.

I did not want BrokerLens to be a thin AI wrapper that generated a confident-looking price without showing how it reached the result. I separated the application into two layers: deterministic code handles the financial calculations, while AI gathers and interprets current market evidence.

Another challenge was evaluating the strength of transaction evidence. A completed sale for the same type of business is much stronger than a broad industry average or an active asking price. I created evidence tiers and server-enforced limits so weak evidence cannot create an unjustifiably large adjustment.

I also needed to prevent stale research. If a user changed a financial input after completing research, the old evidence could incorrectly remain attached to the new valuation. BrokerLens solves this by immediately locking the results whenever an input changes and requiring fresh research.

Finally, public business listings do not always disclose every financial detail. This reinforced the importance of showing assumptions clearly and treating BrokerLens as a preliminary valuation rather than a certified appraisal.

What I learned

I learned that AI is most useful when it operates inside clearly defined boundaries.

The AI does not need to calculate the entire valuation. It is better suited for researching and summarizing market evidence, while deterministic code is better for repeatable financial calculations, validation, and adjustment limits.

I also learned how important financial normalization is. Owner compensation, depreciation, interest, add-backs, inventory, equipment, real estate, and debt can all substantially change a business's estimated value.

Most importantly, I learned that transparency is not just an explanation added at the end. It must be part of the product itself. Users need to see what changed, why it changed, and which evidence supported the decision.

Accomplishments I am proud of

I am proud that BrokerLens:

• Produces a complete valuation from financial inputs through market research
• Reconstructs SDE instead of blindly trusting a reported cash-flow number
• Makes every multiple adjustment visible
• Provides clickable evidence for every research factor
• Prevents unsupported AI evidence from affecting the valuation
• Protects against stale research
• Handles inventory, FF&E, real estate, excess assets, and assumed debt
• Works across desktop and mobile devices
• Successfully performs live AI research in production
• Passes linting and production-build validation

Most importantly, I turned a problem my dad experienced in real life into a working product that could help other business owners facing the same uncertainty.

What's next

My next steps for BrokerLens include:

• Uploading tax returns and financial statements for automatic extraction
• Supporting multi-year weighted SDE analysis
• Connecting licensed completed-transaction databases
• Backtesting estimates against verified business sales
• Adding sensitivity analysis and scenario comparisons
• Expanding industry-specific risk factors
• Supporting collaboration between owners, brokers, accountants, and buyers

The long-term goal is to make preliminary business valuation more accessible, transparent, and defensible before an owner pays for a certified appraisal or begins negotiating a sale.

BrokerLens provides a preliminary valuation and does not replace a certified appraisal. Financial inputs, add-backs, asset values, assumed debt, and comparable transactions should be independently verified.

Built With

• css
• html
• javascript
• localstorage
• next.js
• node.js
• openai-api
• react
• vercel