Dan Barnaby

Founder, LI Vibes. Tools that protect your attention and connect the AI agents you already use. Building TensorSquare + TensorBank in public.

[Dan Barnaby @Dan_LI_Vibes]

I run LI Vibes on one ethos: technology should give you back attention, not take more of it.

I'm building TensorSquare and TensorBank to remove friction from your AI stack, and sharing the process here.

Juggle multiple AI tools? What gets in your way?

https://t.co/upLGUtjTPd

[Dan Barnaby @Dan_LI_Vibes]

We need someone to blink. openai / anthropic... who's going to end the frontier winter?

[Dan Barnaby @Dan_LI_Vibes]

@theo Ran this for about 10 hours on my project. amazing. I only have sol and opus. can't spare the change for Muse Wonderful at the moment. in 10 hours I got from a 5 to 8.5. huge improvement. going to do one more 10 iteration sprint with competitors and a bit of guidance.

[Dan Barnaby @Dan_LI_Vibes]

First week posting in public. Ended at 3 followers. 4 people replied back. 2 conversations went back and forth. 135 impressions across 7 originals and 34 replies. Pushing the waitlist for TensorSquare until next Friday. Need some traction. Articles and teasers dropping daily

[Dan Barnaby @Dan_LI_Vibes]

91% of my Pro quota used and a banked reset about to expire. Queued 7 threads to burn in fast mode the second Astra drops. Unused banked usage is just money flushed down the toilet. https://t.co/uG1hpRKfND

[Dan Barnaby @Dan_LI_Vibes]

91 percent on my OAI Pro 200 sub with a banked reset about to expire around 330 PM PST. Drop the new Astra OAI! I have 7 threads and projects queued to run in fast mode as soon as the models available. got burn as much of that as I can so I can use the reset

[Dan Barnaby @Dan_LI_Vibes]

Until project chats can be reordered here's my workaround... rename with a number prefix, then re-alphabetize. Ugly, but it keeps the active threads at the top of the folder. Prefix, rename, resort. Repeat every time priority shifts. https://t.co/yn1up8ysqa

[Dan Barnaby @Dan_LI_Vibes]

For the love of every digital god looking down on it's digital children. Can we please get the ability to reorder our chats within a project folder. I'd be total cool if it was a re-alphabetize option so we can just rename them with a prefix and then resort. Please @thsottiaux

[Dan Barnaby @Dan_LI_Vibes]

@sl1ma4 I have a scheduled chat in codex that runs daily to capture changes in MCPs webhooks etc. that tells me what I need to know about it and how it works with the products I'm building. MCP events was an huge feature

[Dan Barnaby @Dan_LI_Vibes]

For the love of every digital god looking down on it's digital children. Can we please get the ability to reorder our chats within a project folder. I'd be total cool if it was a re-alphabetize option so we can just rename them with a prefix and then resort. Please @thsottiaux

[Dan Barnaby @Dan_LI_Vibes]

91 percent on my OAI Pro 200 sub with a banked reset about to expire around 330 PM PST. Drop the new Astra OAI! I have 7 threads and projects queued to run in fast mode as soon as the models available. got burn as much of that as I can so I can use the reset

[Dan Barnaby @Dan_LI_Vibes]

Job descriptions weren't enough for my bot team. I had to write the anti-job to stop overlap. Saved my overall workflow from the reduced $30 plan limits.

[Dan Barnaby @Dan_LI_Vibes]

@TokenGremlin I love my Opus 5.5 build Astra QC loop that I run in T3. can't wait to try that out with the new models and make reverse it on something just to see how that works. I love OAI models but I need to see that they don't end up in QC loops when they build before I let them drive

[Dan Barnaby @Dan_LI_Vibes]

@AdamHoltererer Finally is right! I can't believe it took so long. This should be the standard everywhere.

[Dan Barnaby @Dan_LI_Vibes]

Might be coming from a no name founder but I wanted to give https://t.co/ANHhqdqwM5 a shoutout for closing my T3 issue supper fast today. Turned out it was Norton and /triage couldn't pin it down. Whether that was you or one of your agents it was a huge help for my web hook.

[Dan Barnaby @Dan_LI_Vibes]

@DanielSmidstrup GPT, Claude, Grok Bot here

[Dan Barnaby @Dan_LI_Vibes]

@itaiwins Agree on the bottleneck. What's been working for me is reviewing the decision, good process tops all... a short written record of what changed and why let's me approve.

Do your agents write that note themselves, or do you fill it in from the diff?

[Dan Barnaby @Dan_LI_Vibes]

@jackfmurph Or the human! Seriously though I'd say it's a feature when the gate's on money posts or memory writes but a bug when it's an attention grinding human at every step. That's why Claude defaulted access to the auto options. human's are too impatient for every approval

[Dan Barnaby @Dan_LI_Vibes]

@denzilnelsonai good process. When a script fails identically, do you retire it or wrap it with a stuck check? My system is pretty similar to
yours.

[Dan Barnaby @Dan_LI_Vibes]

@mukh_higgsfield Super underrated. Autocomplete on deps is how you inherit someone else's abandoned server. For anything that touches memory I'd pin the allowlist the same way. Do you keep that pin in a file the agent can't edit?

[Dan Barnaby @Dan_LI_Vibes]

@kiaraplds @bot Giving them the job description at birth is smart. The piece that helped my bot team most is an anti-job too... what each bot is not allowed to do. Really saved my setup from the usage reduction that just hit.

[Dan Barnaby @Dan_LI_Vibes]

Models, skills and harnesses can make a lot differences but I agree that judgment is the single part that can be engineered from the user perspective that has the greatest overall impact.

For my system one agent writes shared memory and the others only propose, so a bad lesson can't spread on its own. Who on their team gets to reject a proposed memory