AI Models/OpenAI Dots By: Bitroot Newslogger · 6 min read

OpenAI Dots: What the Always-On AI Assistant Means for Founders

OpenAI Dots: What the Always-On AI Assistant Means for Founders

OpenAI's Dots runs in the background, watches your calendar and email, and proposes actions. Here's what it does, what access it needs, what it costs, and how to test it before you trust it.

OpenAI's Dots is a background assistant that acts without being prompted, and it asks for more access than most founders would grant a tool by default. OpenAI launched Dots on September 29, 2026, and TechCrunch's October interview with the executive who leads its enterprise product says the tool works best once it's connected to your calendar, email and computer. The source is an interview with an OpenAI executive, so treat its claims as the company's account. For a bootstrapped founder, the question is whether a background agent saves more time than the access it needs is worth.

What Dots Does

Dots runs on a schedule, not just on request. A "proactive research" feature reviews what the assistant knows about you and what access it has, looks for ways to help, and proposes actions. According to TechCrunch's interview, it has flagged calendar conflicts and deadlines, and one engineer's dot caught an outage before the engineer did.

It builds on tools OpenAI already had. TechCrunch notes that dots came out of the company's agentic coding tools. It also says much of the same work was already possible with the Codex app, but it took more hand-holding.

Team use is part of the plan. OpenAI's launch describes "teams of Dots working together," and users can message a dot through Slack, Teams and other work platforms. Text message support is "coming soon." OpenAI also says a dot can be given its own identity, credentials and tools through existing systems, and is working with Microsoft to bring dots under its Agent 365 security controls.

What Access It Needs

The assistant works best with broad access. The interview says that once connected to your calendar, email, computer and other tools, the assistant tries to get work done. Access is what makes it useful, and it's also what makes it risky.

The controls are not granular yet. OpenAI says dots has the same privacy controls as ChatGPT, including turning off model training and limiting what the agent can reach on your computer. The interview says those controls are not granular, and an OpenAI executive acknowledged the company needs "super, super, super granular controls."

Check what you're granting before you connect. Start with read-only calendar access and no computer control. Expand permissions only after you've seen what the assistant does with the first tier.

What It Costs

Dots is limited to paid ChatGPT tiers. TechCrunch's launch coverage says dots is available in the ChatGPT desktop and mobile apps for Pro and Business Premium users in eligible markets. The October interview describes a Pro tier starting at $100 per month. That price is for the Pro subscription, not a separate dots fee, and the launch coverage doesn't state dots pricing.

The $100 tier is a recent addition. OpenAI introduced it in April 2026, between Plus at $20 and the $200 Pro plan. The April coverage says both Pro tiers share the same core features, and the main difference is usage limits. It also notes the $200 plan no longer appears on OpenAI's pricing page, though it's still available.

Test the value before you commit. Track how often a dot's proposal saves time or prevents a miss over two weeks. If the answer is rarely, the subscription doesn't pay off for a small team.

Agents on the Other Side

Your customers' assistants may not be you. The interview quotes Embiricos saying he personally doesn't intend to have a meeting with someone else's agent. He also said OpenAI keeps a human assistant because corporate partners want to deal with people.

Agents raise an open question for businesses. TechCrunch asks whether agents can scale without overwhelming businesses, websites and payment networks. OpenAI says it's moving carefully and is talking with payment processors. The article notes that letting an assistant spend money would be a big request of consumers.

Plan for both cases. Keep a clear way for a human to book a meeting or resolve a billing question, and make sure your booking and checkout flows still work when a request comes from an automated tool.

Early Reliability and What to Test

Early defaults asked for permission often. The interview says the initial settings were very conservative, and OpenAI expected that feedback and planned to tune it.

Alerts can arrive late. Embiricos recalled one alert arriving only three minutes before the event it was about. Timing is still imperfect, so don't rely on the assistant for anything that needs a reliable deadline.

The app can be pushy. The interview says the app refuses read-only access to cloud drives, so some data stays out of reach.

Run a short test. Connect one data source, set a two-week trial, and log every proposal with whether you accepted it, whether it was right, and how long it took to act.

Recommendation Blocks

Try Dots if: - Your work depends on calendar and email coordination - You already pay for a Pro or Business Premium ChatGPT subscription Hold off if: - You handle customer data you can't share with a third-party assistant - You need permissions that can be limited to specific actions Build for agent visitors if: - Your product takes bookings, payments or sign-ups from customers - Your sales process depends on a human meeting

A Decision Framework

Step 1: Define one job. Pick a single recurring task, such as meeting prep or deadline tracking.

Step 2: Grant the minimum access. Start with read-only calendar and email access. Add permissions only when the test shows value.

Step 3: Log the results. Record each proposal, whether it was right, and whether it saved time.

Step 4: Check the price against the value. Compare the subscription cost with the hours you actually saved.

Step 5: Review your flows for agents. Confirm that booking, checkout and support still work for both humans and automated tools.

Bottom Line

Dots shows where AI assistants are heading: background work, team use, and more access. The company's own description is the strongest source here, and it's an interview, not an independent test. The practical steps are clear: grant only the access you need, measure the time it saves, and keep your customer flows working for both people and agents.

This isn't security or legal advice. Review your data policies and your customers' terms before you connect a business account.


Explore More on Automation Opportunities

Visit Bitroot's homepage for founder guides, tools, and automation resources built for early-stage SaaS teams.

More from the newslogger

More AI Models