AI funding: preparing a first introduction.

A useful first introduction explains the customer workflow, the evidence available today and the next result the company intends to demonstrate.

Artificial intelligence is part of our stated investment focus. The outline below helps founders and intermediaries describe an AI business without sending customer records or confidential technical files.

Name the customer workflow

Identify the person using the product, the task they need to complete and who pays. Describe the input, the output and any human review before that output reaches a customer or triggers an action.

State whether the product is a demonstration, an unpaid trial or a paid service. Give the reporting period for customer and usage figures, and define what you count as a customer or a completed task. Label planned deployments separately from current use.

Make performance claims checkable

Name the product and model versions, test dates and intended operating conditions. Explain the measure of success, the comparison method and the number of cases assessed. State whether those cases were used to train or tune the system.

Include failures, human corrections and situations the team has not tested. If a result comes from a controlled test, label it as such. For a time-saving claim, explain the starting point and include any review or rework.

The NIST AI Risk Management Framework, under MEASURE 2.1, calls for recording test datasets, metrics and evaluation tools.

Explain data and dependencies

Summarise where the data comes from and the permissions or licences the business relies on. Identify rights or customer permissions that remain unconfirmed. A first introduction can describe the arrangement without attaching contracts or personal data.

Name external models, hosting services and other important suppliers. Explain where customer data is processed, which parts your team controls and how the service behaves if a provider becomes unavailable.

Show the cost of a completed task

Choose a unit that matches the service, such as a resolved support request or a document reviewed. Give the currency, period and volume behind each cost estimate. Include model or API charges, storage, monitoring and human review; describe onboarding and support costs separately.

Distinguish recorded costs from forecasts. State whether temporary credits or discounts affect the figures and what assumptions would change at a higher volume. Do not present the price of one model call as the full cost of serving a customer.

Connect the funding to a milestone

State the amount and currency sought, the expected timing and the work it would fund. Describe the next result, how the team would measure it and what remains uncertain.

Illustrative outline: a document-search company seeks funding for a controlled customer pilot. It plans to prepare an authorised evaluation set, integrate access controls and assess answers with human reviewers. The pilot would report answer quality, failure types, review time and cost per resolved request. This is a fictional example, not a Zethong investment or requirement.

Prepare one non-confidential page

Include the company name, website, your role and a business reply address. Summarise customer use, available evidence, data dependencies, costs and the funding request. The capital introduction guide provides the contact address and an email outline.

Keep identification documents, bank details, credentials and confidential files out of the first message. An introduction does not imply financing. We have not published a minimum investment amount or a standard response time.