Drowning in news? How to use AI agents as a filter

Good communication has always been about relevance. The challenge facing us today isn’t access to information. It’s the volume of it. Ask yourself, how many newsletters do you receive in your inbox, and how many do you ignore because you don’t have the time to read them all?

What if you could get an AI tool to help you with the constant flow of updates, research, opinion and commentary?

Google’s Gemini can help.

Over the past year, I’ve started using Gemini’s AI agents in a very practical way.

If you’d like to create something similar, you can paste the prompt below into Gemini and adapt it for your own role or industry.

How to create your agent

I’ve set up a Gemini task to scan selected technology and industry publications on a specific day each week. The brief is narrow and focused. ‘Look for leadership changes, enterprise technology developments and trends affecting sectors such as financial services, healthcare, insurance and retail.’

The output is short. A summary is provided in the chat giving me a structured snapshot.

It gives me a starting point. Everything that follows still relies on my experience to ask the right questions.

I read the summary and decide what is useful, what I should share and what might make good material for content. Used properly, the agent becomes a filter.

If you’d like to create something similar, you can paste the prompt below into Gemini and adapt it for your own role or industry. You can search for client news, competitor updates or specific trends in an industry - the possibilities are endless.

Prompt you can adapt

Role: Act as a senior research analyst specialising in [Insert Industry, e.g., South African technology].

Task: Every [Insert Day, e.g., Friday] at [Insert Time, e.g., 10:00 AM], scan [Insert Specific Sites, e.g., TechCentral, ITWeb, MoneyWeb] for stories published in the last seven days.

Focus areas:
– Leadership appointments or C-suite changes
– Significant enterprise technology developments (cloud, AI, infrastructure)
– Regulatory changes or major market movements

Exclude:
– Consumer gadget reviews or product unboxings
– Speculative opinion pieces without cited facts
– Personal commentary

Format:

Headline (with direct URL)

Context: Two-sentence summary explaining why this is relevant to industry leadership

Potential impact: One short bullet point outlining possible implications for the local market

Setting it up as a scheduled task

Gemini allows you to schedule prompts so they run automatically.

When you include a clear time and frequency in your prompt (for example, “Every Friday at 10am”), Gemini will ask whether you’d like to schedule it. Once confirmed, the task runs at the set time.

You’ll receive a notification when the update is ready. On desktop, a small indicator appears next to the chat thread in the sidebar. On mobile, you may receive a push notification.

You can pause, edit or delete scheduled tasks in Settings under Scheduled Actions. This allows you to adjust timing or refine the sources the agent monitors.

The difference between a chatbot and an agent

There’s a practical difference. A chatbot answers a question. An agent works from a brief.

Instead of asking, “Summarise this article,” you define a task. Scan specific sources. Focus on defined sectors. Present the findings in a consistent format.

The more specific the brief, the more useful the output. The agent simply follows instructions.

Where agents help in communications

Used carefully, agents can support communications in very practical ways.

For example, if you’re pitching media in the financial services space, an agent can scan recent coverage and flag which themes journalists are writing about. If three different publications are suddenly writing about AI governance, that will tell you it’s a hot topic.

Before drafting a thought leadership piece, an agent can pull together recent commentary on the topic. Not to copy it, but to see what’s already been said and where there’s repetition. That makes it easier to find a different angle.

If you’ve hosted a webinar or recorded a panel discussion, an agent can extract key themes from the transcript and suggest possible article angles or LinkedIn posts. It won’t replace editing, but it removes the first round of manual sorting.

And when regulatory changes are being discussed across multiple outlets, an agent can surface those references early. That gives you time to prepare a position, brief a spokesperson or update messaging before questions start coming in.

What agents should not replace

AI agents shouldn’t pitch journalists. They shouldn’t decide strategy. They shouldn’t handle sensitive conversations.

They don’t understand nuance or timing in the way we do. They process information. People interpret it.

Relevance has always been the goal. AI just makes the process more manageable.

For me, it’s like having a personal assistant focused on the reading I don’t have time to do.