How to start your financial forecasting model

Start with the business, then the fields

A forecast begins with how customers pay, what it costs to deliver, who you need to hire and how much cash you have. You can start with estimates, as long as you label them. AI can help organize the information and reduce typing; you still decide which assumptions belong in the model.

Choose how much help you want

Manual entry gives you a close look at every input. AI preparation helps turn scattered notes into a structured checklist. Browser assistance can help with repeated entry when your agent supports it. These are alternative starting points, not a sequence you must finish.

Separate preparation from prediction

Ask an agent to expose missing information and explain unit conversions. Do not accept an impressive-looking table as evidence that its assumptions are right. A copied value and a typed value need the same review.

Build once, keep improving

After entering assumptions, inspect the model and adjust what you learn. You can start with AI preparation, enter some values manually and use browser help for another section. Keep your sources and revisit the forecast when your plans or evidence change.

1. 01

   Build your first model step by step

   Enter the assumptions yourself and learn what each part of the model means. A good choice for your first setup or a small update.

2. 02

   Prepare your forecast with an AI agent

   Use your own AI agent to turn business notes into an input checklist. Review its output, then copy values into Tessly one field at a time.

3. 03

   Fill your model with a browser-capable AI agent

   Delegate repetitive typing to a third-party agent that can control your browser. You choose the model, review the assumptions and confirm the final actions.

Northstar Notes is a fictional B2B subscription business. Every number below is a teaching assumption, not a benchmark or a recommendation. Use a separate practice model; replace these values before making business decisions.

Operating currency

USD

Current monthly subscription revenue

10,000

Current paid company customers

100

Monthly new paid company customers

10

Monthly paid-company retention

98%

Monthly non-headcount OpEx, excluding marketing

2,000

Monthly non-headcount marketing

500

Team / monthly base salary per person

2 people / 3,000

Current cash

100,000

Check the units: 10,000 ÷ 100 = USD 100 per customer per month. Non-headcount spend is (2,000 + 500) × 12 = USD 30,000 per year, excluding payroll. Base salaries total USD 6,000 per month before employer costs. Retention of 98% means 2% churn for the same monthly cohort; do not enter 2 in a retention field. These checks do not predict the final forecast, which also depends on timing, costs and other assumptions.