From prototype
to first clients.

Opening question: your website is live; what would make someone take the next step? This is a three-hour workshop for the existing cohort. A first client is an ambition, not a condition for passing.

Use the notes as a facilitation guide, not a script to read aloud. Keep most of the session practical. The assignment opening uses 40 minutes; customer research uses 50; break uses 10. In the practical hour, spend 8 minutes on tool choice, 12 on one CRM lab, 20 on conversation practice, and 20 on capture and experiment planning. Use alternatives as optional reading. Keep 20 minutes for portfolio review.

Ask one student to explain the assignment in their own words. A polished site alone does not explain the entrepreneurial process. These requirements come from the current hub; dates and exam arrangements remain in Google Classroom.

Open the existing Assignment page. Show all eight items and the submission form. The working prototype item includes both live website and repository links. The advertisement includes its creative link and copy. This workshop adds examples, not a ninth required submission.

Worked example: participation 5.0 and portfolio 8.0 gives 7.4 weighted, but the participation component still fails the minimum. Attendance and active contribution are both relevant. Use the existing rubric; do not invent a sales quota or a new grading category.

Illustrative example: a small retailer may lose enquiries when stock information is unclear. A stock-enquiry flow tests that assumption before building full inventory software. Connect the decision to Lean Startup, Business Model Thinking, or entrepreneurial experimentation.

Example task: find a suitable product and submit a stock enquiry. Check outside the developer account. A screenshot of the homepage cannot demonstrate the complete flow. A repository link does not replace the live prototype. Scope can be small if the flow is testable.

Fictional illustration: three volunteer shoppers tested the enquiry form; two missed the submit button. Keep this labelled as a teaching example. In real portfolios, use anonymised notes and relevant screenshots or recordings with permission. Separate user observations from your interpretation.

This is a fictional example, not a proven conversion improvement. Ask: what observation justified the change, what exactly changed, and how was it checked? If a retest has not happened, label it planned. Cosmetic changes can be useful, but the evidence must explain the decision.

Weak: we used AI because it is fast. Stronger: AI drafted three form labels; I checked them with two users and chose the clearest. Team work can share project evidence, but each student must identify their own contribution and reflective learning. Do not invent participation.

Ad clicks and likes do not establish demand by themselves. A useful next step could be a product enquiry, trial, booking, or volunteer test signup. The ad is already required in the hub and has no extra grading weight. Check that the creative link opens for the reviewer.

All numbers on this slide are fictional. Pairs choose one vague statement from their own draft and rewrite it with actual evidence. If they have no evidence, write an honest test plan instead. Ask another pair to distinguish fact, interpretation, and next action.

Allow five minutes of individual work and a brief share-out. Ask students to explain what a reviewer could verify from their links. Assignment deadlines come from Google Classroom. Transition: the following customer work should help close a real evidence gap.

A segment should tell students where to look and why someone might care. Small businesses is too broad. Try independent shops in one area that receive repeated stock questions, then verify rather than assuming the problem.

Ask teams to separate a sales invitation from a research invitation. If they are testing demand, do not pretend to have an established service. Pair review: could the other team name a plausible customer and understand the next action?

Use cohort examples as hypotheses. Inhousez or AT Media could investigate a defined business publishing need. ShopTruck could recruit prospective shoppers. Hoofin Around or Parelli could invite relevant learners. Ask MatPort to define its audience before choosing a channel.

Choose one route per team. Lead gathering is not a contest to collect addresses. A publicly visible contact detail does not by itself establish permission for marketing. Keep classroom activity to research and drafts unless the team has an appropriate basis and channel for live outreach.

Fictional example: a studio publishes weekly articles; its page invites business enquiries. That is observable. It needs our AI service is an assumption. Record the source URL and date checked. Reject candidates that have no connection to the customer profile.

Demonstrate with a prepared public source or fictional sample. Have a student check each claim against the source. Avoid promising that an AI tool can discover private contact details or infer budget reliably. Full reusable prompts are in the workbook.

Five is a classroom practice target, not an official assignment quota. Consumer teams can document recruitment routes and use fictional records for CRM practice until volunteers sign up. Keep personal contact details out of shared classroom notes.

Return at minute 100. Set up the fictional CRM sample while the class takes a break. Do not spend the workshop creating multiple tool accounts.

Start with the downloadable CSV in a spreadsheet students already use. A dedicated CRM can come later when needed. The learning objective is a reliable process, not memorising a vendor interface. Explain contact, activity, and opportunity as different concepts.

Use one CRM for the exercise. The other tools serve different steps in acquisition. DistroNow is a repository-backed workflow, not a promise that every live integration is configured. Tool facts were reviewed on 5 October 2026; the tools guide links primary sources.

Our recommendation is based on the small-team free plan and documented CSV import, not a comparative performance benchmark. Teams above three editors can use the spreadsheet exercise or another suitable plan; never share a login. Do not assume all AI features are included without limits.

Use the dedicated fictional companies CSV in the tools guide, not real inbox data. Company name maps to Name and domain to Domains; list attributes hold source, fit evidence, stage, next action and due date. Add an actual teammate as owner after import. Confirm current import permissions. All .example domains are fictional; do not send or enrich them. Documentation-backed steps have not been tested in an authenticated Attio account.

folk lists $24/member/month with annual billing. HubSpot suits students who want a broader familiar CRM ecosystem; folk is a useful agency example. These are teaching judgements. Premium folk features such as API access and sequences should not be assumed in Standard. Follow the source links for current limits.

The pricing FAQ states the annual Cloud Pro price. Self-hosting requires operations work and is not equivalent to a no-cost hosted classroom account. Its published cloud offering includes an MCP server. Keep infrastructure setup outside the core lesson; the learning remains fields, stages and evidence.

Okra CRM exists at okracrm.com; it is distinct from Ora, referenced in LEGIT AGENTIC. Its site advertises trials but several plan/signup links lead to anchors and the pricing blocks are unclear. We have not validated a working account or its integration behaviour. Do not require students to buy it.

DistroNow commit 4c4d7e9 documents a browser-local fictional demo with no discovery or sends. Use the instructor-provided app address; no public URL was verified. Live mode requires storage, provider and mailbox configuration. Current Treg docs differ from the repo contracts, so live discovery is an unverified extension. Simulated replies, meetings and revenue never count as actual portfolio validation.

LEGIT AGENTIC distinguishes deterministic readiness checks from measured AI visibility. Ora is an agent-readiness platform and worked with Vercel on is-agentic. A higher readiness score does not prove recommendations, demand, or sales. Tie the audit to the user journey and include observed test evidence.

Contacted requires a sent message; replied requires an actual reply; won requires a real agreement. For validation projects use test completed and learning recorded instead of commercial closing. Do not advance stages because an AI score looks promising.

Import the blank template. Show the fictional sample row marked DEMO. Dates in a real record must be actual dates. Minimise information collected. Do not put prospect data in the public repository or cohort-wide notes.

Use the fictional sample if research produced no suitable candidates. Demonstrate a declined contact: mark do not contact and remove the follow-up task. This is a workflow exercise, not a new database feature in the course app.

Template: Hi [name], I noticed [verified observation]. We are testing [solution] for [audience]. Is [problem] something you encounter? If so, would a short conversation be useful? Review facts, tone, and contact appropriateness before sending.

Avoid would you buy my amazing product. Ask about actual situations and existing workarounds. Role-play for four minutes, switch roles, then identify one leading question. AI role-play is practice; it is not customer validation.

This is a conservative classroom exercise rule, not a statement of law or a universal sales cadence. Choose appropriate channels and check applicable requirements before live outreach. Automate internal reminders before external messages. Never auto-send AI drafts.

Students can draft a research invitation, a service enquiry invitation, or a response to an inbound signup. The audience and context should determine the message. No requirement to contact real people during class.

Use a fictional form submission for the demonstration. If there is no integration, a documented manual handoff is acceptable for the prototype. Do not claim a CRM integration exists unless students can demonstrate the saved record.

For a manual workflow, test the steps and ownership explicitly. Collect only needed information and explain the stated purpose. A failure should be visible and recoverable. Record observed behaviour rather than ticking success for an untested case.

Fictional example: five invitations, two replies, one booked test. Reply rate is 2/5, not 2/10 researched candidates. A booked test is not a completed test or a paying customer. Record zero honestly and investigate the cause.

Suggest ten reviewed candidates or an opt-in route and five appropriate invitations, adaptable to the project. These are practice targets, not grading requirements. Define learning signals before acting; do not guarantee replies or sales. Let teams test their strongest remaining assumption.

The advertisement still needs its creative link and copy, and the prototype still needs live and repository links. A list of contacts alone is not testing evidence. Show the existing portfolio fields; no separate CRM upload is mandatory.

Example portfolio paragraph: I observed two of three volunteer testers miss the enquiry action. I moved it beside the product details and added a confirmation message. My next test will check task completion and expectations about a reply. The retest is planned, not completed. Never copy these invented numbers into a real submission.

Give a minute of silent writing. Strong reflection can include a failed test, an incorrect assumption, or a tool limitation. It should connect personal action to evidence and judgement. Shared project results do not replace an individual account.

Refer to the official hub bands: 0.0-5.4 insufficient, 5.5-6.9 sufficient, 7.0-8.4 good, 8.5-10 excellent. These summaries are explanations, not a replacement rubric. More leads, more screenshots, or a sale alone do not establish a higher grade.

Pairs review the eight-item checklist. Each student records one concrete improvement and its owner. Make sure private evidence is shared appropriately with the assessor rather than posted publicly. Allow time for questions about the actual assignment.

Also clarify: no customer interviews yet means the team needs a test plan and actual validation evidence before claiming completion. AI simulations cannot stand in for real users. A CRM screenshot alone is insufficient. Deadlines and exam chances remain in Google Classroom; do not invent dates.

Exit ticket: each student states the next action, who owns it, and when it will happen. Collect unresolved assignment questions for the instructor. Remind students to save their portfolio evidence in the existing hub.