I use Seef every day. By now it has taken over every AI use case I have: chat, a stack of agents I run constantly, meeting and interview audio, custom apps like the one that writes my invoices. I used to have all of that spread over separate tools.

Which AI subscriptions I replaced with Seef

  • ChatGPT, for day-to-day chat and custom agents.
  • Copilot, for finding things in our company knowledge base.
  • ElevenLabs, for speech to text on interviews and meetings.
  • n8n, where I had built a few AI flows for processing email and lead generation.

I have since cancelled nearly all of it.

What I do with a meeting recording

I have an agent for meetings. I give it my own notes and the recording, nothing else. I get back a summary with all the action points in it.

Those action points become my to-do list. They are also what I bring to the next meeting, so I can hold people to what they said they would do.

I save that summary to the knowledge base, so the agent can read it back later. It is not stuck with the meeting in front of it. It knows what was said in the earlier ones, which is how it flags issues and spots patterns across them.

It saves me around 15 minutes per meeting or interview. There is no back and forth. I upload the recording and press the button. What comes out is what I wanted. My meetings got much more efficient.

Before this I needed two platforms: one to turn the audio into a transcript, another to do something useful with that transcript. I never wanted the transcript.

Is the chat as good as ChatGPT?

I think chat and agents are on par with what I had.

Two of the agents I lean on most:

An email drafter. I give it the thread and tell it what I want out of the exchange. It drafts the reply in my tone, and it can pull from our whole knowledge base for anything else it needs. I told it what I look out for and how I want an email structured. It has my writing style too.

An agent that makes agents. Its job is writing the system instructions for my other ones. I describe the problem and hand over the context. It writes me a new agent, or fixes one that is not working well.

With a private ChatGPT account the agent stays mine. Here I share it with anyone in the company. They use it, they improve it. Our internal processes get better as a group, not one chat window at a time.

Is it better than Microsoft Copilot on SharePoint?

This is where I saw the biggest gap. Against Copilot connected to our SharePoint, Seef is massively better. It feels more intelligent. I ask for something and I get it.

Part of that is which models it runs on. In Seef I pick the models myself, and I keep it on the best one available.

The rest of it was the unnecessary limitations I kept hitting in Copilot. Some of those are written down. Microsoft’s own advice for building a Copilot agent on your SharePoint documents: keep a file under roughly 15 to 20 pages, split anything longer into shorter ones, take the tables out first, because Copilot cannot read them.

I am not going to reformat our documents so the AI can cope with them.

When a process is too big for a chat agent

Some of what I do, an agent cannot handle on its own. Those used to be flows in n8n, simple ones, because I built them myself. My complicated internal processes run as custom apps instead. The Seef team builds those.

The one I use most makes my invoices. I give it the project and the price. I get back the invoice on my own template, already filed to my bookkeeping platform. It takes the rest from the context, usually meeting notes or an email thread. I do not type anything.

Where I still use another tool

Image and video generation. I go to Gemini for that, because I find it better.

Was it worth switching

The biggest win for me is having it all in one place. I cancelled nearly every other AI subscription. The custom apps matter almost as much. Processes too complex for a simple workflow now run on something the Seef team built properly.