The alpha, twice a week
That was 3 min.
Get the best new moments in your inbox every Wednesday and Saturday. It's free.
Check your inbox
We sent a link to your address. Click it to finish signing up. Not there in a minute? Look in spam.
or
Upstarts · 3 min · Jun 25, 2026
Writer says workflow scaffolding cuts token use by 30–50%
May Habib
May Habib says Writer’s model efficiency and shared workflows use 30–50% fewer tokens.
In this episode
Full episode- 40:15AI buyers should be held accountable for results4 min
- 35:09Writer's moat grows as enterprise AI gets more capable5 min
- 22:06May Habib describes bias in her VC fundraising5 min
- 30:20Writer kept training models because enterprise adoption lags3 min
- 18:11Writer uses review policies to scale personal outreach3 min
- 9:16Internal politics determine how far enterprise AI projects go3 min
- 11:49Writer says workflow scaffolding cuts token use by 30–50%3 min
- 13:59A pharma briefing can link Outlook, CRM and spreadsheets3 min
“the way that we scaffold and set preferences means that you're using 30, 40, 50% fewer tokens each time.”
May Habibat 12:09 · Upstarts
Takeaways
- Shared workflows reduce repeated setup for AI tasks.
- Habib reports 30–50% fewer tokens used each time.
- Writer is aggregating customer practices for AI FinOps.
About this segment
May Habib says Writer’s model efficiency and shared workflows use 30–50% fewer tokens. Its CFO is gathering customer FinOps practices to share across an advisory board.
May Habib is cofounder and CEO of Writer.
Co-founded and leads Writer, an enterprise agentic AI platform serving Accenture, Mars, Uber and hundreds more.
Alex Konrad, Host · Founder & editor, Upstarts Media
Ran Forbes' Midas List for ten years and co-created its Cloud 100 list; founded Upstarts in 2025.
From “Writer's May Habib: Building The Enterprise AI That Anthropic And OpenAI Sell — Then ‘Leave’ Behind”, Jun 25, 2026. Watch the full episode
Thanks. We'll look at and fix what's off.
Transcript
Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.
- 12:01
and we're starting ones by industry as well because people need to learn from everybody else. We just had an advisory board meeting where tokenomics came up and we have very efficient token economics in our business. You're not setting out a new set of plates every time someone comes over for dinner with Ryder. So not only are the underlying models much more efficient, but the way that we scaffold and set preferences means that you're using 30, 40, 50% fewer tokens each time.
- 12:29
And because we're built for shared and scaled work, all those workflows get set before, right? And yes, you can, you know, augment them, et cetera, but also you're not overusing tokens as a result of model needing. to rethink how to do something. And so everybody had a lot of questions about that. And we said, all right, what we're going to do is Ryder's own CFO is going to interview each of you anonymously.
- 12:52
I mean, you know, that's obviously not an anonymous conversation, but we will aggregate the learnings. And at our September meeting, we're going to share out everything we've learned about what everyone is doing on the FinOps side with regards to AI. And so it is really fun to be a builder in the enterprise right now because I don't think anyone has had this level of intimacy with the customer ever, like not Palantir or not Salesforce and the customer themselves.
- 13:17
I mean, the first time I was called a vendor, I sort of like almost teared up and cried because I was like, vendor, what? But you really work beyond that so quickly because this is, it is existential for folks where they're putting their chips on AI and, you know, where their secret weapon and, you so secret in many pockets, but it's incredible to have that level of co-creation. It's like having design partners across every business, right? Across every customer. Now, not every pocket, right? But you find who are the power users, the innovators, the executives who really are ambitious, right? And can think big about what the technology can do. And you become the tip of the spear for the account together. I don't think anyone's ever built software that way. The labs certainly don't build software that way.
- 14:07
So it's very exciting for us to get to do it. When you guys were sort of first going out