Futureproofing your Brand for AI Agents

That's the conversation we set out to have at our latest Futureproof Project session. Twenty-six leaders joined us, ranging from CMOs at large enterprises to consultants embedded in financial services, tech, and creative orgs. Some have been building agents for two years. Some are just getting their feet wet (building “baby agents!”). All of them are trying to figure out how to move with purpose in a space that's changing faster than anyone can keep up with.

The meeting began with a conversation with Amit Shah, founder of InstaLILY and former president of 1-800-Flowers, where he helped grow the company from $500M to $2B in revenue. Before AI was a boardroom word, Amit was operating at the intersection of code and culture, which is the lens he brought to this conversation.

Here's what stood out from our session on AI Agents.

The Arc: Code became culture.

Amit traces 70 years of marketing through one lens: the evolving relationship between code and culture.

Now we're in the agentic era. Code is teaching us autonomy by being autonomous. As Amit put it, "Marketing no longer moves with human mediation. The code is moving on its own."

You're not just managing new technology. You're navigating a new kind of workforce.

The Context Moat.

Every conversation I have about agents orbits one word: context.

The tools are table stakes. You cannot build a moat around code. Software can now be replicated easily with AI. The differentiator is what you bring to them.

As Amit put it, context is not equal to storage.

"Context is not equal to storage. There is a huge desire to treat it like Google Drive — load everything up and the system gets smarter. Actually, it causes what we call context rot."

Amit Shah, InstaLILY

Three practical implications:

Explicit vs. implicit signals.

The difference between a tool you use and a platform that gets smarter is how it learns. For example.

Explicit: InstaLILY monitors recorded calls — when a pattern appears repeatedly (say, a company intro running over 90 seconds), the system flags it as a suggested SOP. It becomes a rule: "Keep it to 90 seconds. We've seen this."

Implicit: If enough people on the team use an initial capital "Insta" and always capitalize "LILY," the system learns the brand standard “InstaLILY” and applies it automatically.

Something often overlooked is what not to include matters as much as what to include. Context isn't just additive. The subtractive form — what's irrelevant to this person, this role, this task — also compounds.

Three criteria for evaluating an agentic initiative.

Sumantro Das, Amit’s Co-Founder and COO at InstaLILY AI offered these three criteria.

Fail faster.

Someone asked whether speed or agility matters more right now.

Amit's answer: Speed, every time.

"The marginal cost of rebuilding yourself is almost very close to zero. So you should just fail faster, because then you can start rebuilding quicker."

Amit Shah, InstaLILY

Grant McDougal, CEO of AI marketing platform, BlueOcean.ai, told the audience rebuilt a competitor's LLM share-of-voice measurement tool from scratch — customer request to enterprise deployment — in a single day.

According to Grant, "There's no competitive advantage in software anymore. The competitive advantage is derived from unique data access."

The software is just the container. The data is the secret sauce.

As a leader, that means creating a culture where failing fast isn't just tolerated — it's expected.

Build fast. Learn fast. Rebuild fast.

The trust factor.

One truth to reckon with is that people are building tools that nobody uses. The reason is trust. Many humans still aren’t confident in the results they get from AI.

Sometimes distrust isn't a problem — sometimes it's even healthy. One call participant pointed out, "You're accountable for the quality of the work. I wouldn't trust a junior writer 100% either."

The problem is when distrust becomes paralysis. It takes much less time to edit than to create. If an agent gets you 80% of the way there, that's still a massive time save. The human-in-the-loop isn't a workaround. It's the architecture.

For example, Traction has built an AI research workflow that flips traditional research on its head using synthetic audiences. When we tested in on a real client situation and then validated the outcome by uploading 30 transcripts of real stakeholder and customer interviews, we used Gemini how close it had gotten. It told us 70-90%. We then asked the strategy director who actually led the interviews. His response? “I’d say about 80%” We were then able to plug the holes in the process by asking what gets us to that last mile of confidence?

So the leadership challenge isn't just building the tools — it's building the habits around them. Adoption doesn't happen at launch. It happens when people see their peers actually using it, and getting better results because of it.

The "Baby Agent" reality.

Most organizations are still in baby agent territory, and that's a legitimate place to start.

Practical advice: do a repeatable-task audit. Every role, every recurring task, no matter how small. Even automating a 20-minute monthly report starts to compound — and it gets your team believing before you ask them to tackle the bigger stuff.

Start where you can move fast. Show impact. Build confidence. THEN expand.

Your next customer is an agent.

If you're a marketer, your next customer is an AI agent. Your brand needs to be machine readable.

Agents will be making purchasing recommendations, evaluating vendors, surfacing choices to humans. If your brand isn't legible to AI systems — positioning unclear, differentiation unstructured — you're invisible in an agent-mediated buying process.

Brand may be the most defensible moat in an agentic world. The things that make brands human-resonant matter more as agents become intermediaries.

Key Takeaways

Requests from the community.

The Futureproof Project is Traction's community for marketing leaders navigating the AI transformation.

Email us if:

Check out our 2026 AI Marketing Guide.