Rami Salam @RamiFSalam

AI & Digital Transformation How businesses actually use AI process, systems, and what works. Lebanon Joined February 2019

Tweets

696
Followers

635
Following

883
Likes

2K

Rami Salam @RamiFSalam

2 weeks ago

A three-day working week sounds great. But would companies actually give people those extra two days off? There’s a lot of discussion about how AI could increase productivity and change the way we work. If people can get the same amount of work done in less time, it raises an interesting question: who benefits from that extra time? Will employees work fewer hours and have more time for their families? Or will companies simply expect them to produce more in the same amount of time? I think that’s one of the biggest questions about the future of work. Technology can change how we work, but the real question is how companies choose to use the time saved. A shorter working week is possible to imagine. Making it happen is another story.

0 0 0 27 0

View Details

Rami Salam @RamiFSalam

2 weeks ago

🇱🇧 A positive development for Lebanon. The U.S. has revoked four previous presidential determinations related to restrictions on aircraft operating to or from Lebanon, stating that these measures are no longer necessary for the stated security purpose. It’s encouraging to see a change like this. We hope to see more positive developments, and a future with greater opportunities and stability. Thank you President Trump, for this decision. Lebanon deserves a better future. 🇱🇧

0 0 0 245 0

View Details

Rami Salam @RamiFSalam

2 months ago

Before you buy another AI tool, ask: “Where are we wasting time every week?” Not where AI would be cool. Where people are actually losing time. Searching for information. Preparing reports. Answering the same questions. Moving information between systems. Pick one. Test AI on that task. Don’t automate everything. Just see if it can make that one part faster without removing human review. You’ll learn more from that than from testing random AI tools.

0 0 0 35 0

View Details

Rami Salam @RamiFSalam

2 months ago

@sama The interesting shift is that safety is becoming part of the development timeline, not something that happens after the model is built.

0 0 0 42 0

View Details

Rami Salam @RamiFSalam

2 months ago

@OpenAI 14x faster is wild. The more interesting question is what workloads become practical when the wait basically disappears…

0 0 0 64 0

View Details

Rami Salam @RamiFSalam

2 months ago

There's a difference between using AI at work and actually changing how the work gets done. Adding ChatGPT to a process isn't transformation. Changing the process because AI can now handle part of it is.

0 0 0 30 0

View Details

Rami Salam @RamiFSalam

2 months ago

@adamhfry quiz feature might be my favorite one here. Simple idea, but genuinely useful.

0 0 0 44 0

View Details

Rami Salam @RamiFSalam

2 months ago

@Codie_Sanchez The hardest part is staying in the game long enough to reach that 0 → 10 → 100 moment.

0 0 1 10 0

View Details

Rami Salam @RamiFSalam

2 months ago

The question isn't: “How much of this can AI do?” It's: “What unnecessary work can we remove, and what remains is actually worth automating?” That usually leads to a better AI workflow.

0 0 0 3 0

View Details

Rami Salam @RamiFSalam

2 months ago

Then split the work into two parts: What can AI handle repeatedly? What still needs someone to decide? I wouldn't automate the second part just because I can.

1 0 0 3 0

View Details

Rami Salam @RamiFSalam

2 months ago

Before you automate a process, watch someone do it manually. You might find the process needs fixing before it needs AI. Here's how I'd decide what to automate first: ⤵️

1 0 1 40 0

View Details

Rami Salam @RamiFSalam

2 months ago

@gdb The “saved money” use case is probably one of the easiest ways to prove real AI ROI.

0 0 1 13 0

View Details

Rami Salam @RamiFSalam

2 months ago

@jxnlco AI catching one expensive mistake is sometimes worth more than all the flashy use cases.

0 0 0 2 0

View Details

Rami Salam @RamiFSalam

2 months ago

The summary was accurate. It was also misleading. I ran the same messy meeting notes through two prompts. The summary correctly said someone suggested the 25th rather than treating it as agreed. It also flagged that nobody had been assigned to the training. Then it gave me seven bullet points and seven open items. For a call where exactly one person had actually committed to something. So I tried a different prompt: “Put this into a table with four columns: Decision, Owner, Deadline, Still unresolved. Leave the cell blank if it wasn’t explicitly stated.” One row had an owner. Every deadline cell was empty. Same notes. Same tool. The summary made the meeting look productive. The table made it obvious that almost nothing had actually been decided. That empty column is the useful output.