Twilio CMO: We shrank 2 weeks of work down to one day with AI. Then the unsexy work began
Fortune Chris Koehler
Opinion — commentary, not a factual news event.
Twilio says AI cut one marketing job from two weeks to under a day. The catch: the real work was cleaning data, rules, and brand guardrails first.
Based on reporting by Fortune, Chris Koehler — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Twilio’s marketing team says it used to spend more than two weeks on research and personalized content for a single campaign. Now the same work can be done in under a day, and the team can run as many as 15 account-based campaigns a quarter instead of the three it used to treat as standard.
But the point of the story isn’t that a model did the job. It’s that the model only became useful after Twilio did the dull stuff first: human review, brand judgment, data cleanup, and workflow redesign. Without that, the company says an AI experiment for a French campaign literally dressed everyone in berets and looked too goofy to publish. That’s the kind of failure that never shows up in vendor demos.
The article is really an argument about how far enterprise AI has actually gone. Twilio CEO Khozema Shipchandler said in August during Q2 earnings that the company is not in early innings, but still at batting practice. That tracks with the broader gap the piece points to: 86% of companies think customers are satisfied with their AI service, while only 51% of customers agree. Deloitte also found that only 25% of organizations have moved 40% or more of their AI pilots into production.
The framework here is simple enough. Phase one is individual productivity. Phase two is teams using AI through shared workflows and custom agents. Phase three is the fully autonomous stuff vendors love to pitch and very few customers are actually running. Twilio says it’s in phase two. That’s a more honest place to be than pretending the whole company is already one prompt away from magic.
The hard part, Twilio says, is the connective tissue: clean data, governance, integration, and brand cohesion. Teams are using different tools, from ChatGPT and Gemini to Claude, and the company is still working to widen access to Claude for marketing work. In other words, the future is not blocked by model IQ. It’s blocked by plumbing, which is less glamorous and much more expensive to ignore.
My take — AI-written commentary, not fact-checked reporting
This is the bit AI boosters keep skipping: the model is the flashy demo, but the enterprise product is paperwork with better autocomplete. The real winners won’t be the companies bragging loudest about autonomy; they’ll be the ones that admit they’re still in phase two and fix the boring infrastructure without a press tour. The unsexy work always wins eventually, which is rude but efficient.
Read more about this at: Fortune
