%title%
Applied AI
When a group of tech companies announce an agreement to support an AI backend-software protocol, we don’t blame corporate IT executives for not paying attention. It sounds quite boring. But one such announcement on Wednesday from Google, Microsoft, Salesforce, Snowflake, ServiceNow and others is worth paying attention to. It demonstrates one way established enterprise software providers may parlay their large customer base into an AI business that won’t be decimated by Anthropic and OpenAI.
Jun 18, 2026

Applied AI

Amir Efrati headshot
Supported by Sponsor Logo

Thank you for reading the Applied AI newsletter! I’d love your feedback, ideas and tips: aaron@theinformation.com.

If you think someone else might enjoy this newsletter, please pass it forward or they can sign up here.

Hi! If you’re finding value in our Applied AI newsletter, I encourage you to consider subscribing to The Information. It contains exclusive reporting on the most important stories in tech, like our exclusive scoop about Google's star AI researcher Noam Shazeer going to OpenAI. Save up to $250 on your first year of access.


Welcome back!

When a group of tech companies announce an agreement to support an AI backend-software protocol, we don’t blame corporate IT executives for not paying attention. It sounds quite boring.

But one such announcement on Wednesday from Google, Microsoft, Salesforce, Snowflake, ServiceNow and others is worth paying attention to. It demonstrates one way established enterprise software providers may parlay their large customer base into an AI business that won’t be decimated by Anthropic and OpenAI.

These traditional tech firms, along with open source model repository Hugging Face and database firm Databricks, said a new software standard would make it easier for workers to use one application to access AI from any apps their company pays for.

In other words, if an employee asks GitHub Copilot or Google Gemini or Salesforce’s customer relationship management app to complete a workplace task using AI, the app would theoretically identify and access any AI features or apps that could help with the task—so long as the makers of those features and apps also adopt the new backend protocol, known as agentic resource discovery.

In pursuing this standard, traditional software providers like Microsoft want to convince customers that they should use their apps as a gateway to access any AI they want, a kind of one-stop shop. 

“ARD helps AI clients discover capabilities, but it doesn’t replace authentication, authorization, governance, or organizational trust decisions,” Microsoft said in a blog post about the new protocol. 

To us, ARD sounds like an extension of the idea behind Model Context Protocol, an open standard Anthropic invented last year to help AI agents be able to access data from other applications, similar to an application programming interface. Other companies including Microsoft, OpenAI and Google also embraced the MCP idea. 

The idea behind ARD seems to be that because every enterprise app under the sun now has its own AI or built-in agents and automations, a new standard is needed to access all those features from one place.

However, Anthropic and OpenAI want to earn the same place in enterprise workers’ minds by making Claude and ChatGPT the main app people use to access all their AI and other applications or tools for work. And the AI firms and the traditional software firms are already in a race to develop so-called superagents, AI that can access any app. Again, the idea is to create a one-stop shop of sorts. This threat is not lost on Microsoft, as we have reported.

As insurgents in the enterprise app game, Anthropic and OpenAI might choose to ignore the ARD effort; they aren’t on the list of initial supporters for the standard.

While it’s too soon to know whether ARD will catch on, Microsoft said the new protocol would make using GitHub Copilot a lot cheaper. When customers today ask the coding app to do research or use other applications to complete a task, GitHub Copilot spends a lot of processing power to figure out what kind of apps or agents it should use; when it has access to features and apps that are ARD-ready, Copilot won’t need to look so hard, the company said in a post. 

That might be enticing to cost-weary IT execs we’ve been speaking to lately!

Cursor Confirms GitHub’s Worst Fear

Speaking of GitHub Copilot, it’s no secret that the pioneering coding product has been losing market share over the past year to competitors like Claude Code and Cursor. And we reported last month that senior Microsoft executives have worried that, in addition to losing ground in the AI coding market, those competitors could also pose a threat to GitHub’s core business of repositories where developers store and collaborate on code.

On Wednesday, Cursor confirmed that those concerns were well-founded. The startup announced that it plans to launch a new code storage and git hosting product called Origin, which will become available later this fall. (The announcement also confirmed The Information’s recent reporting that Cursor was plotting such a move.)

If Origin catches on, Microsoft would have good reason to be worried: currently, most users of AI coding tools other than Copilot still rely on GitHub to store code, especially enterprise customers that need a single unified repository to stash their company’s source code. But Cursor could now aim to peel away more of those users who may find it convenient to simply store that code in Cursor’s products instead.

Cursor isn’t the first AI company to consider such a move. OpenAI has also been privately toying with the idea of building a similar product geared towards users of its Codex coding agent.

It also doesn’t help that GitHub has seen a sharp uptick in outages in recent months that have frustrated customers large and small. Several frustrated customers previously told me that they’d like to move their code out of GitHub due to the bugs but couldn’t find a feasible alternative. Perhaps now they have one. —Aaron Holmes

AWS’ AI Pricing Strategy

Amazon Web Services is trying to make its AI tools more appealing to business customers—by making pricing easier for companies to understand.

In line with that approach, AWS charges for its workplace productivity tool Quick by the number of hours the AI spends completing a task, not tokens, according to Jigar Thakkar, vice president of agentic AI for business. 

“If you [have] a price for everything, it becomes complicated. What is the cost of creating a PowerPoint? What is the cost of asking a basic question and so on and so forth? Instead, we say how much time it took the agent to take care of the question asked,” Thakkar said. “Everything will not be measured closely by tokens. We care about how much time the system spent on it.”

Quick charges $20 or $40 per worker per month, and gives each worker four or eight “agent hours” as part of their plan, depending on the subscription tier. After workers use up that agent time, AWS charges them $3 per agent hour, though it isn't clear how many or what kind of tasks an agent can do in an hour.

Thakkar said Quick also has guardrails to keep customers from spending more than they’d want. —Catherine Perloff

A message from Google Cloud

How is AI strengthening enterprise security?

The ROI of AI in security report explores how AI agents are helping deliver the next wave of proactive enterprise security, and shows where gen AI is delivering value in this space.

Find out where AI is delivering the most value.

Get the report.

Upcoming Events

Wednesday, July 29 — Intimate Dinner with The Information and EY in New York City

More details


Wednesday, September 23 — AI Agenda Live SF 2026

Tickets are on sale now for The Information’s AI Agenda Live in San Francisco. Join top AI leaders for conversations on the future of AI, and secure early bird pricing while it lasts!

More details


Tuesday, October 27 – Wednesday, October 28 — The Information’s 2026 WTF Summit

Save the Date: The Information returns to Napa Valley October 27-28 to convene senior women across tech, media, and finance. The event will feature two days of intimate, candid conversations with the leaders navigating today’s global shifts.

More details

Opportunities

Group subscriptions

Empower your teams to stay ahead of market trends with the most trusted tech journalism.

Learn more


Brand partnerships

Reach The Information’s influential audience with your message.

Connect with our team

About Applied AI

Laura Bratton takes you inside how businesses are using AI to automate work.

Read the archives

Follow us
X LinkedIn Facebook Threads Instagram
Read anywhere
Apple Store Google Play