|
16 SEPTEMBER 2026
|
|
|
In AI,
AI can do more. What makes it ready for work?
|
|
Last week, we looked at AI taking on more complete pieces of work. This week brings two questions closer to home. Can AI help your team understand the information your business already has? And as its abilities grow, how do we make sure we can depend on it?
|
|
|
|
|
AI at work
ChatGPT’s Data agent: ask a business question, explore your company’s numbers.
A sales review often starts with a simple question: what changed this month?
Finding the answer can mean pulling together spreadsheets, asking someone to prepare a report, and returning with more questions.
Last week, OpenAI introduced the Data agent in ChatGPT Work. It connects to approved company data, investigates changes and creates interactive dashboards your team can share. You can ask follow-up questions in the same conversation.
For example, a sales manager could ask which regions contributed to a fall in revenue, then explore the products or customers behind it.
The setup matters. The agent needs access to the right information and your company’s definitions of measures such as revenue.
Why this matters
More people could investigate business questions themselves, with less time spent waiting for a report.
For teams putting this into practice, the work starts with agreeing on the numbers, connecting reliable information and checking the answers. That is what makes a useful tool dependable in everyday work.
|
|
Also this week · The direction of AI
|
|
|
AI safety
Altman and Amodei agree on pacing AI development. Trump and Huang push back.
As AI takes on more complex work, its developers are debating how quickly to build the next generation.
Over the weekend, Anthropic CEO Dario Amodei called for a slower pace of improvement in most Frontier AI models.
Why now? He points to AI increasingly helping build future AI systems, alongside incidents where systems acted in ways their developers did not intend. His argument is that testing and safeguards need time to catch up.
One concrete step: give independent reviewers ongoing access inside AI companies to examine their safety practices. Anthropic committed to doing this.
OpenAI CEO Sam Altman supported pacing development and said OpenAI would also bring in independent evaluators. Read about Sam Altman’s response ↗
Then, on 14 September, Trump called Nvidia CEO Jensen Huang while Huang was onstage at the All-In Summit. Trump opposed a slowdown, arguing that it could help China get ahead. Huang agreed with him.
Amodei’s proposal allows development to continue while giving safety work more time. The announcements do not establish an agreed industry-wide pause.
Why this matters
Our takeaway is to expand its role as it proves itself in real work. Check whether it uses the right information, stays within its permissions and brings a person in when judgement is needed.
Those checks become more valuable as AI handles more of a task on its own.
|
|
|
|
Practical tool · Steal this
Check whether AI can find your business.
Best for business owners and marketers
The problem: You know how your website describes your business, but not what a potential customer sees in an AI answer.
1. Write three questions a customer might ask before knowing your name. Run each in fresh chats with web search enabled in two AI tools you already use. Do not include your brand name.
2. Save the answers, dates and cited links. Separately ask about your brand by name to check description accuracy. Give the saved results to an assistant.
Copy and paste prompt
|
Compare these saved AI search answers for [business]. For each question, show whether we appeared, which competitors appeared and which pages were cited. Keep unprompted recommendations separate from answers that named our brand. Flag inaccurate descriptions. Suggest three website improvements tied to the evidence. Do not invent rankings, traffic estimates or missing results.
|
Why it is useful: You get a small, repeatable check of discoverability and accuracy. It is a snapshot of those answers, not a reliable market-wide ranking.
|
|
From Futurelab
Our Saturday live webinars are back.
Last Saturday, we resumed our live sessions with a practical look at using AI for campaign planning.
We’ll be live every Saturday, exploring ways to put AI to work.
Bring your questions and see how these tools fit into everyday tasks.
Follow Futurelab Studios on LinkedIn for upcoming sessions and useful updates from across the AI space.
|
|
|
|
See you next week
Team Futurelab
|
Join the Conversation
5,000+ professionals
Founders, business leaders, and practitioners navigating the AI transition together.
|
|
|
|
Futurelab Signal · Issue 14 · 16 September 2026
|
|
