The first marketing hire at an AI startup should spend the first 90 days on research before production: days 1 to 30 running customer interviews, days 31 to 60 narrowing to one or two testable use cases and recruiting design partners, and days 61 to 90 building the website and journey for those use cases and running the first trial.
The first marketing hire at an AI startup often walks into a model with no product. No website, no customer journey, no category playbook to copy. The brief is not to promote a feature list. It is to find credible problems the technology can solve, find the people close enough to those problems to teach you something, and turn that into product and market direction quickly.
That was the job at a frontier speech-to-speech AI lab, as first marketing hire and GTM lead from April to June 2026. In about two months the company went from a model looking for a market to two defined use cases and two design partners before the wider trial launch.
The sequence: model to design partners in two months
- Model, no product. The starting point was the underlying technology, without a finished product, website, or established use case.
- 24 customer interviews, sourced via Clay and Apollo. Clay and Apollo were used to find and research potential consumer and enterprise participants. Before reaching out, each person was researched for niche, role, company, likely workflow, and reason to care.
- One framework, adapted per person. A shared research framework kept answers comparable. The questions were adapted for each participant so the conversation surfaced actual behaviour, constraints, and language rather than generic opinions about AI.
- Two testable use cases. The interview findings were combined with competitor and product research to narrow the model into two use cases specific enough to test.
- Two design partners before the trial launch. The use cases gave the company something concrete to put in front of the right people, and two design partners were in place before the wider trial launch.
Alongside the research, the initial website and marketing foundation were built while the product direction was still taking shape. Outbound was used as a research and GTM channel, not as a volume exercise.
The first 90 days for a first marketing hire at an AI startup
That sequence maps onto a 90-day plan that works for most model-first startups.
Days 1-30: research before production
- Write the research question first: what decision should these interviews change?
- Build a target list by role, company, and workflow, not by job title alone. Clay and Apollo are useful for both sourcing and pre-call research.
- Start interviews in the first two weeks. Ask about the last real instance of the problem: what they tried, what they used, where it failed, what it cost, who else was involved.
Days 31-60: narrow to testable use cases and recruit design partners
- Cluster the interview findings with competitor and product research.
- Cut to one or two use cases specific enough to test. More than two spreads a small team too thin.
- Go back to the participants who described the problem most precisely. They are your design partner candidates.
Days 61-90: build the journey and run the first trial
- Translate the capability into the buyer’s language from the interviews.
- Build the website and journey for the chosen use cases, from first explanation to a useful conversation.
- Instrument it, so the trial produces evidence rather than anecdotes.
How to find design partners before launch
Design partners come out of discovery, not out of a cold “join our beta” campaign.
- Source by workflow, not by title. Find people whose day-to-day work puts them next to the problem. Research them before contacting them.
- Interview first, pitch later. A real interview about their last attempt at the job earns the right to ask for more time.
- Watch for precision. The participants who describe the problem in concrete detail, and want it solved, are the candidates.
- Give them something specific. “Try our model” is weak. “Here is the use case you described, can we build it with you” is a design partnership.
The step-by-step method, from problem list to signed design partners, is in how to find design partners.
GTM engineer vs growth consultant vs first marketing hire
| GTM engineer | Growth consultant | First marketing hire / growth operator | |
|---|---|---|---|
| Best for | Automating a motion that already works | Fixing a specific broken system | Finding the use case, positioning, and first customers |
| Needs to exist first | A repeatable process to automate | A diagnosable problem | A model or early product |
| Typical output | Enrichment, routing, outbound workflows | Diagnosis, fix, handoff | Interviews, use cases, design partners, website |
| Stage | Post-PMF, scaling | Any stage with a defined problem | Before PMF, before launch |
GTM engineer is one of the fastest-rising roles in 2026, and it is easy to hire one too early. Automation multiplies a motion. If the motion does not exist yet, there is nothing to multiply. For a closer look at the role, see growth engineer vs marketing engineer vs GTM engineer.
Why AI GTM before product is a research problem
Marketing a model before it becomes a product is a research and positioning problem first. There is no precedent to copy and no existing demand to capture. At the lab, the useful ideas came from specific interviews and observed needs, not from attaching fashionable claims to the technology.
When to make this hire
Make this hire, or bring in a growth operator for a focused engagement, when you have a model but not a product, use case, or motion. If the motion already works and you need to scale it, you need a different role. This is GTM from zero, and the deciding factor is who will own the research. See also when to hire a growth consultant vs a full-time hire.
Have a model and need to find the market for it? Book a diagnostic call or try the free growth tools.