|
AI Builders Digest
Thursday, September 17, 2026
|
|
OpenAI is getting into advertising. Let that sink in for a second. The company that spent years positioning itself as a research lab building AI for humanity's benefit just launched Sponsored Agents. Meanwhile, Box CEO Aaron Levie is making the case that the real AI opportunity isn't in the models at all. These two stories are more connected than they look.
|
|
---
|
|
01
|
OpenAI launches Sponsored Agents and goes into the ad business
|
|
|
OpenAI announced a new advertising layer for its platform: Sponsored Agents, plus marketing tools with integrations for HubSpot and Shopify. The pitch is that AI agents can now surface branded recommendations and take actions on behalf of advertisers inside ChatGPT conversations.
|
Why it matters: Your next ChatGPT session asking for product recommendations will have a business model behind the answer. That's a different product than the one you signed up for. It's also a direct play for Google's search advertising revenue, except the ad appears inside a conversation instead of above a search result. The company that trained users to trust its outputs as neutral is now selling placement inside those outputs. Worth watching how transparently they label it.
|
|
Source →
|
|
---
|
|
02
|
Box CEO Aaron Levie: the model isn't the hard part
|
|
|
Aaron Levie posted a sharp take on where enterprise AI actually gets stuck. The gap isn't between weak AI and strong AI. It's between capable models and the messy reality of enterprise workflows: process re-engineering, data aggregation, human-in-the-loop checkpoints, domain-specific evaluation, security, governance. His argument is that the "applied AI layer" filling that gap is where most of the real work and real value will be created.
|
Why it matters: If you're a company that bought an AI platform expecting the model to do the heavy lifting, you're about to discover your internal processes were the bottleneck all along. The vendors who win the next three years won't be the ones with the best models. They'll be the ones who figured out how to wire AI into workflows that were designed before AI existed.
|
|
Source →
|
|
---
|
|
03
|
Google adds voice Q&A and lecture recording to Gemini for students
|
|
|
Josh Woodward, who leads Gemini at Google, announced two new features in Gemini Notebook for students: live spoken Q&A with class materials in around 100 languages, and automatic audio note capture from lectures. University students in 140-plus countries can also still access a free Google AI Plan.
|
Why it matters: Google is building Gemini into the study workflow before students graduate and become enterprise buyers. Getting someone comfortable with your AI tools at 20 is a more durable acquisition strategy than any ad campaign.
|
|
Source →
|
|
---
|
|
04
|
Salesforce comes to Claude
|
|
|
The official Claude account announced a Salesforce integration, linking through to details on the Anthropic site. No further context in the post itself.
|
Why it matters: Salesforce is in more enterprise sales and CRM workflows than almost any other software. Claude getting native access there matters for the same reason Levie's post above matters: the model's value depends entirely on whether it can reach the data where work actually happens.
|
|
Source →
|
|
---
|
|
05
|
Quick hit: Perplexity drops swag on door handles
|
|
|
Thibault Sottiaux posted about a Perplexity surprise swag drop delivered to door handles, calling 2026 "the year of efficiency." Cute stunt. The 4,800 likes suggest people found it charming.
|
|
Source →
|
|
Follow builders, not influencers. A daily digest of what matters in AI.
Read online ·
Archive
|