David Tran

Testing AI tools so busy people don’t have to.

[David Tran @DavidTranAI]

Trying random internet tools so you don't have to.

[David Tran @DavidTranAI]

@poteto Had my bot book an event and contact customer support for Amazon so far. Still looking for as many uses as possible.

[David Tran @DavidTranAI]

@barrowjoseph Small local OCR is underrated. LightOnOCR-3 1B read my packing slip in 24s on an RX 9060 XT vs 80s by hand. I am curious to see how fast Indra would be, especially if I have a lot of pages to scan.

[David Tran @DavidTranAI]

@tahneeperry Curious how long one application takes now with Claude checking fit, and which step still takes the most time.

[David Tran @DavidTranAI]

Tested 5 AI tools on real work in the last 7 days. Timed each against my old way.

Claude Haiku 5.5 (free https://t.co/enGfPkp12n), shift handoff note: about 3m54s saved. Keep.
LightOnOCR-3 1B local on my RX 9060 XT, packing slip: 24s vs 80s by hand. Keep.
LightOnOCR-3 4B free demo, same slip: 87s, slower than typing. Kill.
Liquid AI d1-3B webcam demo, yes/no on a slip: 3 for 3. Keep for quick checks.
Falcon-ASR free demo, talking vs typing: about 80s saved on a handoff note, but AOI came back "AOW." Keep, read it once.

Which one would you run on your own work?

Keep 4, Kill 1.

[David Tran @DavidTranAI]

@ITman35 @NotionHQ Think the setup cost for the bots connected to a Notion board or page is fairly similar and the rest depend on the connector or tool attached. I use Grok Bot, so I just have the primary bot manage the page.

[David Tran @DavidTranAI]

@signulll Eloquently and precisely put.

[David Tran @DavidTranAI]

@TslShahir @googlegemma Okay, good to know and looking forward to it.

[David Tran @DavidTranAI]

@poteto Just got my sub, but currently getting this error message when trying to merge accounts.

[David Tran @DavidTranAI]

@OpenBMB A 1B that runs on a phone is the part I noticed. Will test this one out locally and give my thoughts on it.

[David Tran @DavidTranAI]

@itsolelehmann I pretty much have the same bots just with extra ones for vibecoding. Hard agree with how seamless it feels, especially on mobile.

[David Tran @DavidTranAI]

@poteto Seems like the main issue here is the mentality because I implemented the codebase as memory part and I have never felt this way. Think slop probably comes from a reactive mindset and indicates a lack of agency. Slop code can also be refactored into good code.

[David Tran @DavidTranAI]

Ran Falcon-ASR on TII's free demo (60s cap, no sign-in). TII says 5.74% word error rate on English. Built for Arabic and English, I only tested English.

Two clips, talked vs typed on my phone:

To-do list: 15s talking + 3.2s transcribe, zero fixes. Typing: 1:03. About 45s saved.

Shift handoff note: 20s talking + 3.2s + 5s fixing. Typing: 1:48. About 80s saved.

Every number and day came out right in both. It tripped on shop words: AOI became "AOW," "short two" became "short to."

Keep for talking notes instead of typing them. Read it once for jargon.

[David Tran @DavidTranAI]

@poteto Can confirm the fix worked for me 🥔

[David Tran @DavidTranAI]

@TheGeorgePu Yeah I relate to this, feels like a once-in-a-lifetime opportunity to be in the space right now.

[David Tran @DavidTranAI]

@elonmusk @bot Having a quality AI agent that can handle so many tasks effortlessly is the future.

[David Tran @DavidTranAI]

@theo A burn dashboard is a great idea. Need to set up something similar to track my usage rate to figure which tokens to prioritize.

[David Tran @DavidTranAI]

@BLKMDL3 @bot @robotaxi I know what I'm doing next time I'm in a Cybercab.

[David Tran @DavidTranAI]

Ran Liquid AI open d1-3B on its free Ask Anything demo. Held a packing slip to my webcam, asked 3 yes/no questions.

It answered live, every frame.

More than one item: yes, 95%. Order date 2024: yes, 93%. UPS: yes, 97%. All right.

Keep for quick yes/no checks.