Insights

How Executives Can Put ChatGPT’s dot to Work

Meeting briefs, follow-through and feedback checks: where an always-on agent helps an executive, and which decisions should stay with you.

By Angelica Sirotin

AI at Work

5 min read

How Executives Can Put ChatGPT’s dot to Work — Sirotin Ventures Insights

Before a partner meeting, you may spend more time finding the latest proposal than deciding what you want to discuss. The previous conversation is in one thread, a revised scope is in another, and you still need to check whether the question you sent last week has been answered.

That kind of preparation is where we’d consider giving ChatGPT’s dot a defined responsibility. OpenAI describes dot as an always-on agent that can work across connected tools and projects, with access depending on the account, available connections and permissions. The useful starting point is a recurring piece of work you can check, such as preparing a brief from the material you already use.

At Sirotin Ventures, we’ve used dot to help bring research into a client brief, review feedback against draft content and prepare partner follow-ups. Those assignments work best when we explain what the finished result needs to help us do, then review it against the source material before relying on it.

Make the meeting brief answer the questions you have

A useful brief should help you join the call knowing what has changed and what needs a decision. With the relevant correspondence and documents available, dot can help identify the commitments from the previous conversation and check what remains outstanding. We’d ask it to link back to the evidence, so a change in scope or an unanswered question can be checked without rereading the whole thread.

You still decide how to approach the conversation. A brief may show that a partner missed a deadline, but the relationship may call for a different opening than a list of overdue tasks. Keeping that judgment with the person handling the relationship lets the agent do the preparation without deciding how the discussion should feel.

Interactive figure: the meeting brief

We’ve also used dot to prepare a partner follow-up with proposed meeting times checked against the calendar and presented across time zones. It brought the availability and wording into a draft for approval before sending. That distinction matters: assembling an accurate message and deciding it should go out are separate parts of the work.

After a meeting, the same sources can help keep the follow-up list current. An item that has been answered should come off it, and an update should explain why something still needs attention rather than repeat every commitment from the original conversation.

Bring scattered information into a usable brief

A new assignment may begin with an agreed scope, interview notes and public research, each contributing a different part of the picture. For one engagement, we used dot to help bring that material together into a brief covering the intended audiences and messaging. Questions the sources could not answer stayed separate, which made it easier to see what still required clarification.

That separation is worth asking for explicitly. Otherwise, a well-written brief can make an inference look like a decision everyone has already made. We’d rather have a clear question to take back to the client than a confident paragraph built around an unresolved assumption.

For an executive preparing to discuss a partnership, the same approach means starting with the decision the brief will support. A general market overview may be interesting, but it will not necessarily explain what is uncertain about this opportunity. Identify those questions first, then have dot gather and organize the material that can help answer them. The final recommendation still needs the judgment of the people responsible for the business.

Check which parts of the work a comment changes

Feedback often arrives in several places: an email adds context to a document comment, while a conversation changes only part of a previously approved direction. Before revising everything, someone needs to work out exactly what has changed.

In one review, we asked dot to compare client email and Word feedback with social calendar copy and the associated graphics. It identified a single line that needed changing and confirmed that an approved graphic could remain. We could then review that correction rather than reopen work the client had already accepted.

Interactive figure: the feedback check

We’ve also used it to help check edits against the original sources and carry approved changes into working documents. For that to be useful, the current files and the approval need to be clear. If several versions are available, we’d establish which one governs before asking the agent to make them agree.

OpenAI’s documentation includes keeping launch materials aligned with product changes as a possible use. In practice, we’d give the reviewer the approved change and the affected documents together, so they can check both whether it was carried through and whether the revised wording has introduced a new promise.

Start with a responsibility you can evaluate

If meeting preparation keeps spilling into the evening, try a brief before expanding the assignment to other work. If outstanding partner commitments are harder to track, start there instead. The first few results should tell you whether the agent has enough information and whether the instructions produce something you can use without rebuilding it.

Before starting, decide where the information lives and whether it is appropriate to give the tool access. Explain what the result should contain, which changes deserve immediate attention and who reviews anything that will be shared. A narrow assignment is easier to adjust when a brief misses an important reply or a task list keeps resurfacing completed work.

For recurring work at fixed times, OpenAI’s guidance calls for a saved schedule with a defined task, timing and destination. Ask dot to confirm the schedule rather than assume a request has been saved. Connecting an app alone does not create a monitoring task, and supported capabilities should be checked in the account you are using.

Agree on access and sending before delegating

Company information should only be connected through accounts and tools appropriate for the organization. Access to read a document does not mean every use of its contents is appropriate, and the output still needs to be checked against its sources.

OpenAI describes built-in safeguards and action review, and its guidance distinguishes asking for a draft from authorizing a send. We would retain human review of consequential messages, commercial commitments and externally shared claims. You can get help preparing the work while remaining responsible for what the business says and promises.

At Sirotin Ventures, we can help you identify a recurring communications task where that support would be useful, then work out what the agent needs and what you want to review yourself. If meeting preparation, follow-ups or scattered feedback are taking more of your week than they should, email angelica@sirotinventures.com. We can start with the part you keep having to pick up again.

Angelica Sirotin

Written by

Angelica Sirotin

CEO, Sirotin Ventures