Repeated Conversation Practice Tools Are Entering Education and Training Programs

Teaching and practice are starting to separate

Foundry, Harvard Business School’s space for supporting entrepreneurs, introduced AI conversation partners modelled on its faculty. Participants can practise investor pitches, sales calls, and board meetings as often as they want, while receiving questions and feedback.

At the time of reporting on August 22, 2026, the programme cost $699 for eight weeks. It did not offer a degree or academic credit. It combined weekly classes led by real instructors with repeated practice using AI.

Harvard’s current official page describes Deep Tech Bootcamp as a four-week online programme, with no price listed. Course length and pricing can change, but Foundry’s description retains the basic structure: people teach the classes and AI handles repeated training.

The important change is not that a professor’s face appears on screen. It is that simulated conversations, which can be restarted at any time, address a problem that one human instructor could not solve: listening to dozens of participants deliver pitches repeatedly.

The currently available information does not show that this approach improves fundraising rates. Foundry did, however, operate a paid programme that combined AI-based repeated practice with weekly live classes led by experts.

The new buyer is the person running the training

Consider a training manager at a regional startup support organisation responsible for investor-pitch training for 24 aspiring entrepreneurs. This person does not teach pitching directly. Their work is to recruit instructors, collect participant materials, run mock pitches, and report training results.

Today, they download pitch files from email, match them to names, and send them on to instructors. On mock-pitch day, each person presents once, receives several questions, and leaves with notes or an evaluation form.

Even after revising a pitch, it is hard to bring the same investors back for another practice session. The manager cannot tell who actually practised again. Instructors spend time on repeated basic questions and cannot sufficiently address issues that require judgment, such as market potential or funding plans.

With AI practice added, participants can upload their pitch materials and repeat a five-minute presentation plus a 10-minute question-and-answer session. The tool can first flag whether an answer was too long, avoided the question, used numbers without evidence, or conflicted with an earlier explanation, before the participant tries again.

In an admin screen, the training manager sees each participant’s practice count, completion status, and questions where they most often got stuck. Before the final mock pitch, instead of teaching everyone the same thing, the manager can pass only the shared mistakes—such as market size, competitors, and revenue evidence—to the human instructor.

The new spender here is not the individual presenter but the training manager at the startup support organisation. To secure approval, the product should not be sold as an “AI pitch tool.” It should be sold as a training operations tool that shows whether 24 participants practised three times, where they got stuck, and how much the instructor’s final evaluation changed.

Human work remains. Automated feedback cannot judge whether a business is truthful, give advice suited to a specific industry, or introduce an entrepreneur to other investors. The product promise should therefore not be “we will help you raise investment,” but “we will help you complete the basic practice before meeting an expert.”

Expansion abroad is happening through institutional sales

PAM, created by Founderland in Europe, is a service where users upload pitch materials and record a presentation, then face questions from six AI investor roles. It separates feedback on the pitch content, on-screen materials, and delivery style. Its main customers are women entrepreneurs of colour, as well as universities and startup support organisations.

PAM does not publicly list a standalone price. Founderland says that more than 50 programmes are on its waiting list, but this is a company claim and needs independent verification. The key point is that it sells to institutions with training programmes rather than recruiting entrepreneurs one by one.

PitchBob, which began in Estonia, asks entrepreneurs structured questions and offers tools for creating pitch materials and fundraising documents. It currently offers a one-time review of existing pitch materials for $99, while products that institutions can operate under their own name start at $99 per month.

PitchBob says it has been used by more than 35,000 entrepreneurs and more than 70 startup support organisations, investment firms, and other organisations. Its fundraising outcomes have not been independently verified, but it shows a sales path: attract users with a low-cost individual product, then expand into an institutional management product.

United Rentals, a US equipment rental company, used Second Nature’s AI customer role for sales staff training. Employees practised asking questions and making proposals with virtual customers whose conditions changed. Managers did not rely only on automated assessments and retained human review.

According to the company case study, the tool was deployed to 6,000 employees and about 5,000 took part in practice. Practice-time and sales-related performance figures are from a supplier-published case study and need separate verification. Still, the case clearly shows that a sales leader can spend budget on the number of real conversation practices, rather than only on lecture attendance.

Four things to build from this

1. An investor-question practice room for startup support organisations

• What it does: Users upload pitch materials. An investor role then asks, in sequence, about the market, competitors, revenue, and planned use of funds, and identifies answers to revise.
• Who uses it: A training manager at a regional startup support organisation buys it for 20 to 30 aspiring entrepreneurs preparing their first investor pitch.
• Why now: Human reviewers can focus on the final mock pitch, while participants complete the repeated questions they need beforehand on their own.
• First screen: Include pitch-material upload, industry selection, investment-stage selection, a start-practice button, and a submission date for the training manager.

2. Complaint-handling training for restaurant franchise locations

• What it does: Staff practise conversations with a customer role in situations such as refund requests, delivery delays, allergy questions, and threats of negative reviews.
• Who uses it: A training manager at a restaurant franchise headquarters distributes it by location where new part-time staff join frequently.
• Why now: Store owners do not need to repeat the same explanations to every employee, and staff can pass required situations before meeting real customers.
• First screen: Let users choose one situation to practise today: refunds, delays, hygiene, or unreasonable demands.

3. Price-negotiation practice for small manufacturers

• What it does: A procurement-manager role raises unit-price cuts, shorter delivery times, test reports, and competitor quotes. Sales staff answer and practise again.
• Who uses it: The owner of an industrial-parts manufacturer with around 15 employees, who has newly hired three sales staff, buys it.
• Why now: The owner can enter client information and promises that must not be made in advance, then assign real customer meetings only to staff who have completed the basic responses.
• First screen: Ask users to enter the product being sold, minimum price, achievable delivery time, and conditions that must never be promised in four fields.

4. Job-interview practice for university career centres

• What it does: Based on a job posting and personal statement, an interviewer role asks about experience, understanding of the role, and examples of failure, then flags vague answers.
• Who uses it: A university career centre provides it by semester for engineering students at regional universities who are about to graduate and are preparing for jobs at mid-sized companies.
• Why now: Rather than repeatedly hearing students’ first answers, counsellors can review records of multiple practice sessions and work only on the final answers together.
• First screen: Place fields for pasting a job posting, uploading a personal statement, and starting a 10-minute interview.

What to check today

Call one training manager at a startup support organisation and ask whether a service where 20 participants practise a pitch and question-and-answer session three times, while the manager receives completion status and shared weaknesses, could be included in the next training programme. If the manager can immediately name the programme budget line they would use and asks for a pilot price, it may be worth testing. If they only demand a guarantee of fundraising outcomes, narrow the promise and the customer again.

Why this matters where you are

Check who runs repeated practice in your local training programmes and whether they need to report completion or common weaknesses. The relevant institutions, budget lines, and participant needs may differ from those in these examples. A practical first step is to test whether a training operator will pay for visibility into practice activity before an expert-led session.