Why Gamma’s CEO Feels They’re “Still Crawling” in AI

“We’re still in the crawl phase.”

That's what Grant Lee, CEO of Gamma, told me about the status of his company's internal AI workflows.

If you've listened to tech podcasts lately, this might sound shocking. Leaders are sharing impressive stories about their AI initiatives—autonomous agents handling entire workflows and custom GPTs replacing entire functions. It's enough to make any company feel behind.

But after hearing how Gamma uses AI to run customer support for 50 million users with just two support employees, I realized something: Grant's "crawling" comment wasn't an admission of being behind. It was a revelation about how bottom-up AI plays out at companies.

Crawling by Design

Gamma uses the classic Form → Storm → Norm → Perform model to frame their AI maturity. Grant says the company is intentionally in the storming phase.

“We don’t mandate tools—we let people figure out what works. Eventually, we’ll consolidate. But right now, every team is learning.”

That means there is:

• No formal AI operations lead.

• No internal prompt library.

• No enforced tooling standards.

Gamma is optimizing for experimentation by allowing ideas, tools, and behaviors to emerge from the bottom up.

Every individual and team is responsible for their own exploration. Show-and-tells act as the connective tissue: someone demos a workflow that saves time or reveals signal, and others adopt it.

This model can look fragmented. But that’s the point. It trades standardization for rapid, spiky advancement.

Gamma’s AI Workflows (A Comprehensive Overview)

Let’s see how this “storming” plays out across the company.

Customer Support: Gamma’s Most Impactful AI Workflow

This is the clearest example of AI delivering operational leverage.

Gamma manages inbound support for its 50 million users with two full-time support employees. The system runs on:

• Intercom for conversation routing and customer-facing interaction

• Fin, powered by OpenAI, to resolve tickets

• Outsourced BPO agents to supplement the team.

“That’s probably our most impactful use of AI. We’ve been able to keep up with growth without needing to build out a big team.”

This isn’t just automation for efficiency. It’s a system that enhances itself over time:

• Fin learns from support threads and resolutions.

• It handles more complexity over time.

• The team spends less time on recurring issues.

This is what AI infrastructure looks like when it’s working: invisible, durable, and compounding.

Design & Prototyping: Powered by a “Unicorn” Operator

Zach, Gamma’s Head of Design, is their internal AI MVP. He has strong UI/UX and visual skill, coding abilities, and thrives in AI-native environments.

His stack includes:

• Midjourney for art direction exploration

• Cursor, for shipping production-level code

• Claude, ChatGPT, and Figma, user research and UI/UX

Zach builds fast. More importantly, he shares what he creates.

Engineering & Technical PMs: Adaptable, Proficient, and Quick

Gamma’s engineers aren’t required to use standard tooling. Some prefer Claude Code, others default to Cursor.

Notable is the emergence of PMs who ship code, especially technical ones using Cursor to move directly from idea to implementation.

This tightens feedback loops and reduces handoffs. PMs can prototype, validate, and iterate in real time.

They’ve also dabbled with Bolt and Lovable but found Cursor most helpful for rapid prototyping.

Growth & Analytics: Tools Before Teams

Gamma’s Growth PM is focused on making data accessible seamlessly.

To do that, they are building internal tools for:

• Self-serve dashboards for common metrics

• Queryable interfaces for non-SQL users.

• Standardized reporting through Sundial

This avoids the trap many companies fall into: hiring data analysts to mediate every request. At Gamma, product managers, marketers, and support leaders can answer their own questions without hindering progress.

The philosophy is clear: build internal leverage through tools, not staffing.

Customer Research: Synthesizing Insights with AI

User feedback is everywhere, especially at fast-growing startups. The challenge is turning it into something usable.

Gamma does that with a smart, organized loop:

• Power users are funneled into a Slack-based community.

• Feature ideas flow into a public Canny board.

• The data is dumped into NotebookLM.

• Internal tagging tools map requests to personas and themes.

This gives product and marketing teams real signals—not just anecdotes. It also creates a foundation for better roadmap planning and segmentation over time.

Knowledge Management & Internal Search

This is still a work in progress, but experimentation is ongoing.

Teams are testing:

• Glean for enterprise search

• Slite’s spin-off product Super as a Glean alternative

• Notion AI is primarily used by CX and marketing teams.

There’s no unified system yet, but teams are sharing notes and allowing usage patterns to guide their next steps.

Marketing & Creative: Lightweight, but Active

Gamma’s marketing team is experimenting with:

• Midjourney and Runway for AI-generated visuals

• An internal Slack bot that suggests daily content ideas.

• AI tools for campaign ideation and creative brainstorming

These tools aren’t fully operationalized, but that’s okay. The experiments are low-risk and high-yield, reflecting Gamma’s approach: explore widely, keep what is effective.

Crawling Isn’t Slowness.

Calling their AI progress “crawling” might sound modest, but at Gamma, it’s intentional.

They’re pacing themselves, exploring broadly, and building institutional intuition before committing to a structure.

The customer support team has achieved world-class leverage. Design and PMs are moving quickly. Analytics is accessible to the whole org.

Gamma has built a foundation for sustainable AI adoption, and I can’t wait to see what their “Perform” phase looks like.