AI is rewriting the GTM playbook - but craft still wins
For years, startups pulled from the same sales and marketing templates. AI changed the tempo. You can do more with fewer people and smaller budgets. The catch: results still depend on domain know-how and seasoned judgment.
Max Altschuler of GTMfund cautioned against throwing engineers at GTM and expecting magic. AI helps, but you still need domain experts and advisors who know the proven plays. The art is knowing where AI amplifies and where human judgment sets the course.
Craft still matters
Alison Wagonfeld of Google Cloud noted that the craft hasn't vanished. Curiosity, research, customer insight, and sharp creative still drive outcomes. Pair that with technical fluency and you've got range.
Teams using AI publish more messages faster and tie efforts to clear metrics. Speed without signal is noise; AI lets you test at a pace humans can't match.
What AI changes in GTM
- Lead discovery: prompts can express a hyper-specific ICP and find lookalikes across structured and unstructured data.
- Prioritization: inbound is qualified and scored with more precision using AI-driven signals.
- Messaging: dynamic personalization at scale based on firmographics, intent, and behavior.
- Feedback loops: instant analysis of replies, calls, and site events to refine outreach.
Marc Manara from OpenAI summed it up: do more with less, and do it with focus. The edge now comes from personalization and signal tracking that wasn't feasible before.
Build an AI-augmented GTM in 30 days
- Clarify ICP and promise: write a one-line outcome and three must-have traits for target accounts.
- Collect data: CRM hygiene, intent sources, product analytics. Map fields you'll feed into prompts.
- Create prompt systems: templates for lead search, enrichment, and first-touch copy. Version them and store in a shared repo.
- Test three channels: cold email, LinkedIn, and partner co-marketing. Ship small tests daily for two weeks.
- Score and route: auto-grade inbound on fit and intent. Route A leads to sales, B to nurture, C to product-led paths.
- Metrics review: track reply rate, meetings booked, CAC, sales cycle, and win rate. Cut what underperforms.
Prompts that actually ship pipeline
- Lead finder: "Find 50 companies that 1) use [tool X], 2) have 50-250 employees, 3) hired 2+ data roles in 90 days, 4) sell to fintech."
- Contact picker: "Within those, choose 2-3 titles closest to [economic buyer] and 1 influencer per account."
- Email draft: "Write a 90-word email that ties [pain] to [measurable outcome]. No fluff. Include one question."
- Scoring: "Label inbound as A/B/C using rules: A = ICP match + high intent, B = ICP match + medium intent, C = research lead."
Keep outputs short. Ask the model to cite the signal it used for each choice. That makes review fast and training easier.
Guardrails that save you from cleanup work
- Quality: review the first 200 AI-generated messages before scaling.
- Compliance: scrub for claims, privacy, and opt-out language. Log prompts and outputs.
- Brand voice: feed a style guide and 10 top-performing samples. Reject anything that drifts.
- Data freshness: auto-refresh firmographics and tech stack weekly.
Hiring: new traits beat narrow titles
Past hiring favored micro-specialists. Now, bias for curiosity, range, and the ability to learn tools fast. Pair that with a few deep experts you can call on as advisors.
- Marketing generalist with AI chops: testing, copy, analytics.
- RevOps with ML exposure: data pipelines, scoring, routing.
- Content strategist who can prompt and edit: short, frequent, on-message.
- SDR who runs AI-assisted research and personalization loops.
Tools and resources
- OpenAI prompt practices: Prompt engineering guide
- AI in marketing from Google Cloud: AI for marketing
Next step
Pick one slice of your funnel and automate it end to end this week. Then expand. Momentum compounds.
If you want structured learning for your team, see these resources:
Your membership also unlocks:
