How to do user research without users: the synthetic user guide. | The truth about ChatGPT Enterprise adoption.

What 1,000 ChatGPT enterprise customers tell us & A simple framework to name your startup.

👋 Hey, Sahil here — Welcome back to Venture Curator, where we explore how top investors think, how real founders build, and the strategies shaping tomorrow’s companies.

• How to do user research without users: the synthetic user guide.

• How to actually name your startup: A simple framework.

• What 1,000 ChatGPT enterprise customers tell us about enterprise AI adoption in 2025.

---

FROM OUR PARTNER - RING AI

🤝 Meet Rings: AI-powered deal flow for modern VCs

Rings is the proactive extended CRM (XRM) built for investors, by a former founder of a 9-figure VC.

The AI-native platform combines a powerful sourcing engine with relationship analytics and a comprehensive CRM.

Leading VCs are using Rings to get access to more deals, track deal flow and simplify fundraising.

Consolidate your tech stack with Rings →

---

VENTURE CURATORS’ FINDING

📬 My favourite finds

Media Posts:

• How to write a fundraising email: First email, follow-ups approved by leading VCs? (Link)

• How to get warm intros to VCs without a network. (Link)

• AI’s trillion-dollar opportunity: Context graphs. (Link)

• Troy Kirwin on writing a memo for investors after your first meeting (Link)

• Guillaume Moubeche shares startup lies (Link)

Reports/Articles:

• US VC female founders dashboard. (Link)

• Investor data room superpack: everything you need to build a VC-ready data room (guide/templates) (Link)

• Stripe Atlas startups in 2025: Year in review. (Link)

• Where European VC funding grew in 2025. (Link)

• Tactical advice for founders on mastering “personality-message fit”. (Link)

• Finding the balance with AI and idea development. (Link)

---

PARTNERSHIP WITH US

🤝 Get your startup in front of over 103,000+ audience.

Our newsletter is read by 103,000+ tech professionals, founders, executives, investors (VCs / Angel Investors) and managers around the world. Get in touch today.

---

📜 DEEP DIVE

How to do user research without users: the synthetic user guide.

When you’re building an MVP or raising your first round, you usually don’t have the luxury of long research cycles or access to dozens of users. But you still need clarity: Who exactly are you building for? What do they care about? How do they behave?

This is where AI-generated personas become incredibly useful.

Instead of spending weeks interviewing users, tools like ChatGPT can create realistic, behaviour-driven personas based on your ideal customer profile. These aren’t just demographic sketches; they simulate a user’s motivations, fears, habits, workflows, and buying triggers.

Researchers even point out that generative AI can speed up persona creation without sacrificing user-centred thinking, helping teams align faster and make sharper product decisions early on.

Of course, AI personas don’t replace real users. But for founders moving fast, they’re a powerful way to:

• pressure-test product ideas

• refine messaging

• uncover blind spots

• and build team alignment around one clear customer archetype.

Crafting personas with ChatGPT (Prompt Templates)

ChatGPT can generate rich, realistic personas from just a few inputs, but the quality depends on what you ask for. The goal is not to create a demographic sketch; it’s to capture the user’s goals, motivations, behaviours, frustrations, and decision triggers.

That’s the information that actually guides product and UX decisions.

A simple rule: Demographics tell you who they are. Motivations tell you what they’ll do.

Core Prompt Template (Generate Persona)

You can start with a structure like this:

Create a detailed user persona for [ROLE or user segment] of our
[PRODUCT/SERVICE]. Include their background, goals, motivations, and main pain points or frustrations related to [problem domain]. 

Focus on how and why they would use the product. 

This forces the AI to go beyond surface-level traits and gives you a persona you can actually design for.

For example, if you ask for a “Marketing Manager who struggles with data privacy in finance apps,” ChatGPT will produce a persona with:

• job context

• daily tasks

• goals

• fears and blockers

• Why this problem matters to them

• How do they evaluate tools or switch products

These are the elements that matter, not their age or whether they drink cold brew.

Add more detail when needed

You can refine the persona by adding parameters like:

• industry

• team size

• technical comfort

• typical workflow

• Tools they already use

Or by asking follow-ups like:

• “What daily frustrations does this user face?”

• “Describe their typical workflow from morning to evening.”

• “What alternatives are they currently using and why?”

This iteration turns a basic persona into an actionable one.

Another example prompt

Create a persona for a Sales Business Development Representative using a CRM app. Frame the persona around their goals, motivations, and behaviors. Include their main pain points and how they currently use similar tools. 

Do not list demographic information unless it directly shapes their decisions.

This approach mirrors how modern UX teams work: personas built around roles, tasks, context, and decision drivers, not superficial details.

The goal: an actionable persona

A good persona should help you answer:

• What are they trying to get done?

• What stops them from doing it?

• What do they value in tools like yours?

• When and how will they use your product?

If your persona clarifies these four things, you’ve got something you can design, build, and pitch around.

Simulate a user interview

Once you’ve built a persona, the next step is to talk to them or at least simulate the conversation. ChatGPT can role-play as your target user and answer questions the way a real person in that role might. This is one of the fastest ways to stress-test assumptions before you speak to actual users.

Set the stage with a role-play prompt

Start by telling ChatGPT exactly who it should become. Give it the persona’s name, context, and behaviours, then frame the interview scenario.

Example setup:

You are [NAME], a [brief persona description] who regularly uses [PRODUCT/CATEGORY]. I am a product manager interviewing you to understand your needs, experiences, and frustrations. Stay fully in character and respond as [NAME] would.

Adding a couple of guidelines helps ChatGPT respond more realistically:

• “You should role-play as a real user with clear goals, preferences, and frustrations.”

• “You have experience with [competitor/product type] and will give honest feedback.”

This primes the model to think and speak like a real user rather than a generic chatbot.

Ask open-ended, conversational questions

From here, treat it like a real interview. Ask questions that encourage depth, not yes/no answers. For example:

• “Can you walk me through how you currently handle [task]?”

• “What’s the most frustrating part of that process?”

• “What tools have you tried before, and why did you switch?”

• “What would an ideal solution look like to you?”

Because ChatGPT stays in character, its responses feel like genuine user feedback, complete with emotions, reasoning, and tradeoffs.

Use ChatGPT to analyse the conversation

After the role-play, you can shift modes and ask:

• “Summarise the key pain points from this interview.”

• “List the top jobs-to-be-done for this persona.”

• “Highlight the strongest buying triggers or objections.”

This gives you a quick synthesis that mirrors what you would produce after a real user interview.

Why this works

Simulated interviews won’t replace real conversations, but they help founders:

• Catch flawed assumptions early

• discover angles they hadn’t considered

• refine messaging and feature priorities

• prepare better questions for real users

In a fast MVP cycle, these simulations act like a rehearsal: low cost, fast feedback, and surprisingly insightful.

Get feedback on ideas & messaging

Once your persona is defined, you can use it as a filter to evaluate your product ideas, landing page copy, feature descriptions, or even early UX flows. This gives you fast, persona-aligned feedback before you test with real users.

Ask the persona to react to your concept

You can prompt ChatGPT to “think” like the persona and tell you how they would respond to your idea or messaging.

Example prompt:

Imagine you are [Persona Name]. Here is our product concept/messaging: [insert idea, copy, or landing page]. How would you react to this? What would you like, dislike, or find confusing? Stay in character and respond as [Persona Name] would.

This usually surfaces:

• unclear value props

• jargon that doesn’t fit the persona

• missing benefits

• objections or concerns

It’s a quick way to see how your message lands with the user you’re designing for.

Ask for targeted improvements

You can also ask for focused, persona-driven suggestions:

• “As a [Persona], what would you improve about this feature description?”

• “What would make this onboarding flow easier for you?”

• “What benefits matter most to you here?”

ChatGPT will respond from within the persona’s mindset, giving you feedback grounded in their motivations, frustrations, and decision-making style.

If something stands out, go deeper:

• “Why would that bother you?”

• “Can you explain what feels unclear?”

• “What would you expect instead?”

This turns a basic reaction into rich insight, like a real user interview.

Role-play design or flow evaluations

You can even test specific moments in the experience:

• “Describe how you would feel going through this signup process on your phone.”

• “As this persona, what questions would you ask after hearing this feature pitch?”

• “What part of this dashboard would be most confusing to you and why?”

These role-plays often reveal usability issues, emotional reactions, and missing clarity that founders typically catch only after shipping.

Your only rule: keep ChatGPT in character

Always remind the model:

“Please answer as [Persona Name] would.”

This keeps the feedback consistent, grounded, and aligned with the persona’s worldview.

Stress-test features or pricing

Personas aren’t just for interviews; you can use them to pressure-test your roadmap and pricing decisions. By putting the persona in buying mode, you can quickly see how they might evaluate tradeoffs, question pricing, or prioritise features.

Test reactions to a specific feature or price

Use a simple role-play setup to understand perceived value:

You are [Persona] (a [brief description]). We’re considering releasing Feature X at price $Y. What questions or concerns would you have? Would this feel worth the cost? Please answer as [Persona] would.

This helps surface things like:

• expectations around value

• price sensitivity

• missing benefits

• objections before purchase

• reasons they might hesitate or churn

It’s not about predicting real budgets; it’s about understanding how your ideal user thinks about value.

Compare priorities between features

You can also push the persona to evaluate tradeoffs:

Act as [Persona]. Which of these two feature sets matters more to you, and why: [Feature A] vs. [Feature B]?

This forces ChatGPT to reason like the persona, revealing:

• what they care about most

• What feels “nice-to-have” vs. “must-have”

• outcomes they prioritise

• Features they would drop if pricing increases

What you learn

Since the model stays in character, you get insight into how this type of user might:

• justify paying more

• Reject pricing that feels misaligned

• Request additional value

• push back on complexity

• choose between competing features

For example, the persona might say things like:

• “I’d expect more automation before paying extra.”

• “Feature Y feels unnecessary — I’d remove it if it increases the price.”

Use these reactions to validate or challenge your assumptions before locking in scope or pricing.

When to use synthetic personas vs. real interviews

AI personas are most helpful in the early stages, when you’re ideating, aligning the team, or quickly pressure-testing assumptions. They’re great for brainstorming (“How would our target user react to this?”), and they help teams rally around a shared picture of the user. They can also sharpen your real research: a synthetic interview often reveals what questions you should ask human users next.

But they’re not a replacement for real conversations. AI can summarise motivations, goals, and frustrations, but it often misses nuance and emotional depth. ChatGPT can predict reactions, it can’t replicate lived experience. Think of its output as hypotheses, not conclusions.

Put simply: Use AI personas as a practice run. Use real users to make real decisions.

As soon as you have a product concept or MVP, switch to actual interviews to validate what matters. The strongest approach is a blend: use ChatGPT to draft personas, explore ideas, rehearse interviews, then go out and talk to users and compare the insights.

Blending AI insights with real feedback

The strongest approach is a mix of AI-generated insights and real user data. For example, you can take interview transcripts or survey quotes and feed them into ChatGPT for fast synthesis.

Builders like Shavin from Buildspace have shown that giving ChatGPT raw customer quotes can instantly generate themes, job stories, and problem statements that normally take hours.

It’s a great way to accelerate analysis, as long as you double-check everything.

ChatGPT sometimes makes confident but incorrect conclusions, so always validate AI output against the source data. If the model suggests a new pain point, check whether it truly appeared in interviews.

A simple loop works well:

1. Draft a persona or interview script in ChatGPT.

2. Talk to a few real users and see what aligns or diverges.

3. Feed the real answers back into ChatGPT to extract themes or refine personas.

4. Update your product or messaging using both sources.

This lets you move fast while staying grounded. AI helps you explore and structure ideas; real users ensure accuracy and nuance.

Red flags: bias, hallucinations, overconfidence

Be cautious about where AI can go wrong. ChatGPT can hallucinate, inventing details, competitors, or statistics that sound real but aren’t. Research has documented this pattern, and it’s one of the top concerns with AI personas. It can also amplify bias, creating personas that feel overly polished or missing uncomfortable truths.

Avoid confirmation bias by using neutral prompts.

Don’t ask “Does this idea make sense?” Ask “What might be confusing or risky about this idea?” When accuracy matters, keep the temperature low to reduce randomness.

Most importantly, never act on AI-generated insights without evidence. If ChatGPT says, “Users would love Feature X,” ask: Do I have any proof? Cross-check with mentors, data, or real interviews. AI isn’t thinking, it’s predicting patterns, so you must stay critical.

Case examples from founders and teams

Many teams use AI personas to speed up early product work:

Fast persona creation. At Traders Eco, a PM used ChatGPT to generate detailed personas from a simple target audience description, helping her team quickly align on “who our user is.”

Rapid mockups and idea testing. Hootsuite’s GM, Partho Ghosh, shared that his team uses ChatGPT plus image tools to prototype landing pages and concepts quickly, allowing them to explore ideas without engineering time.

Accelerated feedback analysis. Canny’s co-founder Sarah uses ChatGPT to scan support conversations, surface feature requests, and group similar user messages — saving hours.

Another PM built a pipeline to cluster raw reviews and then used GPT to summarise each cluster’s pain points.

In all cases, founders emphasise the same point: AI made them faster, not smarter than real users. They still validate everything with actual conversations before committing to decisions.

So - AI personas and simulated interviews are powerful for early-stage founders: they help you ideate faster, stress-test ideas, and align your team before talking to real users. With clear prompts and careful iteration, ChatGPT becomes a valuable brainstorming partner not a replacement for user research.

Use it to explore possibilities, reveal blind spots, and draft hypotheses. Then go validate those hypotheses with real conversations. That combination is where AI actually shines: giving you speed without losing the human insight your product depends on.

---

TODAY’S RESOURCE PICK

🔥 Equity dilution decisions guide: ownership, control, and negotiation

Equity decisions are permanent, and most founders get them wrong not because they’re careless, but because they’re rushed and negotiating without clarity.

This guide helps you understand how dilution actually works, how investors think about ownership and control, and how to make equity decisions without panic or regret.

Includes practical tools: Equity dilution simulator, ESOP planner, raise-now vs raise-later comparison, and cap table + term sheet guidance.

If you want to raise capital without waking up later wishing you’d handled equity differently, this guide is for you.

Get access here →

---

📃 QUICK DIVES

How to actually name your startup: A simple framework.

Most founders think naming is a quick brainstorm or a “let’s run it through ChatGPT” exercise. Arielle Jackson (from First Round) has helped hundreds of startups name companies and products, and her view is simple: naming is emotional, subjective, and slow, yet still one of the most important early brand decisions you make.

The biggest mistake founders make? They want the perfect name on day one, even though the best names often become great over time (Disney, Nike, Square). Your job is to pick a name that fits the strategy today and doesn’t limit you tomorrow.

Start with a placeholder

Your incorporation name is not your final name. Most good companies don’t start with the name they end up with. Some founders even intentionally pick ridiculous incorporation names just to avoid attachment. It’s better to incorporate as something unusable and rename it once you’re clear about your product.

When should you pick the real name?

• When you’ve hit clarity on positioning

• When you’re about to launch and need a public-facing brand

• When your old name stops fitting your product

• When legal risk forces a change

Founders who rush naming end up boxed in. Founders who delay it slightly make far better choices.

Start with positioning, not creativity

Always begins with a simple question: Who is this for? What space do we occupy? What makes us different?

A clear positioning statement makes naming dramatically easier because you know what your name should signal: sophistication, speed, trust, creativity, community, or something else entirely.

Build a naming brief

Before brainstorming, outline the constraints:

• Who is the target user?

• What tone and personality should the name convey?

• Do you prefer descriptive, suggestive or abstract names?

• Any languages, meanings or connotations to avoid?

• Any domain or trademark constraints?

• Five names you love, five you hate, and why

This reduces randomness and helps others understand your taste.

How to namestorm

This is where founders usually go wrong. They brainstorm alone. Arielle suggests the opposite: widen the surface area. Bring in employees, friends, even linguists. And use AI not to “name your startup” but to explore:

• synonyms

• technical words

• metaphors

• foreign-language variations

• historical references

Keep the prompts specific. AI is better at generating raw material than finished names.

What themes to explore

Use both direct and indirect angles:

• Your core value (expertise → Maven)

• Your product’s motion (gather, connect, accelerate)

• Patterns, metaphors, symbols

• Names from mythology or history

• Simple phrases (PayPal, First Round)

A good brainstorm should produce hundreds of words, not ten.

Narrow the list with structured criteria - Evaluate your list against:

Distinctiveness, Memorability, Trademark risk, Domain feasibility, How it sounds when spoken, How it looks visually, & Whether it will age well.

Then shrink to a shortlist of 10–25, then a top 3.

Signs you’ve found a good name

• It’s polarising; some people love it, some hate it

• It grows on you over a few days

• It works on two levels (simple now, deeper meaning later)

• It avoids the obvious, generic and forgettable

Signs it’s a bad name

• Too many companies already use it

• It’s easy to parody

• Locks you into a narrow future use case

• Sounds like a competitor

User feedback helps, but lightly

Ask small groups of users what each name communicates, not which one they “like.” People tend to choose descriptive names because they’re familiar. You want clarity, not consensus.

A simple memorability test works well: share three names today and see which one people recall tomorrow.

Legal + domain basics

Arielle’s rule: Get the name right first. Then solve the domain. You can use:

• prefixes (usemaven.com, tryfigma.com)

• suffixes (squareup.com, awaytravel.com)

• alternative TLDs (.ai, .co, .io)

• lease-to-own domain deals

Trademark clearance should be done before spending money on identity work. A quick USPTO search rules out obvious conflicts. A trademark lawyer helps determine if you can register the name.

Product naming vs company naming

If you have one product, keep the same name as the company. Once you have multiple products, use a simple structure like: Company + Descriptor

Example: Google Maps, Google Drive, Square Reader.

Use unique product names only when they help establish a separate identity (Cash App, Android).

Naming is messy. It’s emotional. And it takes time. But with the right framework, it becomes structured enough to produce a name that supports your positioning, doesn’t box you in, is easy to remember and ages well.

What 1,000 ChatGPT enterprise customers tell us about enterprise AI adoption in 2025.

Henley Wing Chiu scraped every detectable ChatGPT Enterprise and Teams customer using Bloomberry’s real-time B2B purchase data. OpenAI claims explosive enterprise growth this year, but Henley’s analysis shows a more grounded picture of how AI is actually being adopted inside organisations.

Most large companies still aren’t using it. Only about 5 per cent of the Fortune 500 have adopted ChatGPT Enterprise.

Even companies that banned ChatGPT last year still haven’t reversed course. Apple, Amazon, JPMorgan, and Spotify, all remain absent. And none of the world’s ten largest tech companies appear as paying customers. Privacy fears, regulatory pressure, and internal AI strategies all slow adoption.

But where speed matters, adoption is much higher. Tech companies lead with 8.2 per cent penetration, followed by fast-moving fintechs at 6.6 per cent.

Startups use ChatGPT Enterprise twice as often as public companies because they don’t have procurement bottlenecks, political risk, or complex security reviews. They just move faster.

Here are the clearest patterns Henley found:

• Tech + fintech industries lead adoption

• Industrials, utilities, and energy barely use it

• Startups are twice as likely to purchase as public companies

• US companies adopt faster than the EU and Canada

• Government adoption is nearly zero due to compliance and data laws

One unexpected insight: ChatGPT Enterprise usage is not reducing hiring. In fact, companies using it posted roughly the same number of new job openings as last year.

Non-customers saw a decline of about 7 per cent. If AI is replacing workers, it’s not showing up in hiring data yet.

Another counterintuitive trend: companies not using ChatGPT Enterprise are hiring AI talent much faster. They increased AI-related job postings by 120 percent. ChatGPT Enterprise customers only grew AI hiring by 27 per cent.

The cleaner explanation is that non-customers are building internal AI tools and need engineers. Customers are effectively outsourcing that layer to OpenAI instead of staffing those teams.

There are also signs that even slow, conservative industries aren’t ignoring AI entirely. Three of the Big Four accounting firms — PwC, KPMG, and EY — are already active ChatGPT Enterprise customers. Audit prep, tax analysis, policy reviews, and research-heavy workflows map perfectly to LLMs.

What it means for operators and founders:

• ChatGPT Enterprise isn’t mainstream yet, but early adoption is accelerating

• Startups and tech-first organisations are leading the curve

• Slow-moving enterprises won’t switch until privacy, compliance, and data retention guardrails feel airtight

• Companies choosing not to buy ChatGPT Enterprise often compensate by building internal AI stacks

• AI isn’t eliminating jobs yet — but it’s reshaping where companies invest talent

---

TODAY’S JOB OPPORTUNITIES

💼 Venture Capital & Startup Jobs

All-In-One VC Interview Preparation Guide: With a leading investor group, we have created an all-in-one VC interview preparation guide for aspiring VCs. Don’t miss this. (Access Here)

• Investment Fellowship - H7 Bio Capital | Remote - Apply Here

• Analyst - BDC Capital | Canada - Apply Here

• Program Manager - gener8tor | USA - Apply Here

• Associate - Health Velocity Capital | USA - Apply Here

• Associate - Mana Venture | USA - Apply Here

• Summer Fellowship 2026 - Coastanova Venture | USA - Apply Here

• Portfolio Management Intern - LVL1 Accelerator | India - Apply Here

• Associate, Growth Equity - Adam Street Partner | USA - Apply Here

• Venture Relations Associate - Angellist | USA - Apply Here

• VC Fellowship - 1752 VC | USA - Apply Here

• Analyst - Griffin Gaming Partner | USA - Apply Here

• VC Associate - Proglobal Search | UAE - Apply Here

• Intern - Prologis Venture | USA - Apply Here

• Investment Summer Associate - B. Next | USA - Apply Here

• Founder’s Associate - Fertile Soil Venture | UK - Apply Here

---

WHEN YOU’RE READY, HERE’S HOW I CAN HELP:

💁 You might need our help!

• Reach 103,000+ Founders & Investors: Partner with our newsletter to reach a highly engaged audience.

• Build Your Pitch Deck: We write, design and model your pitch deck into a storyteller book within 4-5 days.

---

🔴 Share Venture Curator

You currently have 0 referrals, only 5 away from receiving a 🎁 gift that includes 20 different investors’ contact database lists - Venture Curator