Kyle French (@kylefrenchdev) header
Kyle French (@kylefrenchdev) profile picture

Christian. Developer. Dad x3. #GoHabsGo

🇨🇦 Canada
Joined December 2022
Photo by Kyle French (@kylefrenchdev)Photo by Kyle French (@kylefrenchdev)Photo by Kyle French (@kylefrenchdev)Photo by Kyle French (@kylefrenchdev)Photo by Kyle French (@kylefrenchdev)Photo by Kyle French (@kylefrenchdev)Photo by Kyle French (@kylefrenchdev)Photo by Kyle French (@kylefrenchdev)Photo by Kyle French (@kylefrenchdev)Photo by Kyle French (@kylefrenchdev)
I’ve been building out a public repo of ready-to-install profile packs for Hermes Agent. ⚡ Right now it includes: 🏢 Hermes Agency — specialists for professional work 🧭 Hermes Council — specialists for personal life, growth & guidance 🎓 Hermes Academy — professors and instructors for learning almost anything Each pack is built as a real team of focused profiles with their own roles, skills, and purpose, not just a pile of prompts. Repo link in the first comment! 👇
8
6
2
31
1,963
Pay my @Starlink bill days ago. Decide I don't need that much bandwidth, downgrade. Asks if I want to switch at the end of the bill cycle or now. Choose now thinking I will get a refund for the prorated amount. Nope. And no warning. Immediate downgrade though. What a scam.
15
I am starting to build out Hermes Agent templates that replicate my top Grok Bot Agents and their templates. More on this in the week to come. Own your Agents and Inference.
Phase 1 is first Hermes Agent template done. Developed and reviewed by Hermes agents leveraging GLM5.3 running on local inference.
3
1
22
1,358
Job opening with a great team working on a great platform
hiring a Partnerships Lead to help build and scale the Hermes Agent ecosystem if you or someone you know is a fit please reach out! nousresearch.com/careers/par…
2
1
10
1,309
Kyle French retweeted
Opus 5.5 built a live painting studio from NVIDIA's free SimReady models
30
94
10
1,187
113,896
today, we welcomed a new member of our family and he's a very handsome boy 🐕🐾
3
32
Replying to @OpenAI
@OpenAI, this is ridiculous. @ChatGPT is automatically switching conversations into Work mode, which shares a limited usage allowance with Codex, and apparently it's MY responsibility to stop it. Look at the response I received from OpenAI Support. 👇 I've repeatedly encountered a 30-second countdown offering "Stay in Chat." If I don't respond, ChatGPT switches the conversation into Work mode. I manage hundreds of conversations across multiple projects. I regularly submit a prompt, move to another conversation, and return when the response is ready. Why should I have to babysit every conversation to prevent ChatGPT from initiating an agentic workflow I NEVER requested? Here's where it gets worse. My ChatGPT Pro Work/Codex allowance reset around October 14. By the morning of October 16, **46% was gone.** I hadn't used Codex since the reset. I've never intentionally initiated Work mode. And nobody has explained what consumed that allowance. For context, I run HermesWatcher. I work extensively with AI agents and understand agentic harnesses. I maintain my GPT subscription primarily for use with my Hermes Agent. When I want ChatGPT, I use ChatGPT. When I decide I need Codex, I deliberately open Codex. I have ZERO interest in Work mode. **I decide when to initiate agentic work, not OpenAI.** And this screenshot is what I received after requesting an investigation. Support wants me to identify a specific conversation among hundreds, then advises me to click "Stay in Chat" to prevent future switches. So the responsibility is on me to investigate unexplained usage AND actively prevent ChatGPT from changing modes? Where is the transparency? How much allowance did each activity consume? Which conversations entered Work? Why were they switched? How do I disable this behavior entirely? An allowance eventually resetting doesn't excuse any of this. This is exactly why having control over your AI agents matters. Frontier AI labs are increasingly putting usage limits around powerful capabilities, yet users are still left without basic transparency over when and how those allowances are consumed. **An unanswered countdown is not consent to consume a metered agentic allowance.** @thsottiaux, I'd genuinely appreciate your attention to this. My DMs are open if you'd like the support case number. This needs an actual investigation, transparent usage accounting, and explicit user control. Not another explanation of how to click "Stay in Chat." @thsottiaux @gdb @sama @OpenAI @ChatGPTapp @sulmanrc @OpenAIDevs @OpenAINewsroom @nickaturley
13
5
1
67
4,710
Kyle French retweeted
At your lowest, you will meet a man named Jesus. It is very important that you receive Him as your Lord and Saviour.
54
598
9
7,843
57,005
Kyle French retweeted
❤️💯
83
1,196
55
5,494
48,064
Try Jesus. Just try Him. He will not let you down. He will not and can not turn you away, because to do so would be to unChrist Himself. It would destroy the very character of who He is and He does not change. Try Jesus. Just try Him.
This is the most important message you will ever hear!
6
406
skill bundles in Hermes are very useful
If you keep loading the same skills for the same job, stop doing it one by one. hermes bundles create <name> --skill a --skill b /<name> in CLI or chat loads the set. It's a yaml file under ~/.hermes/skill-bundles/. Check it into dotfiles.
4
193
share what you're building!
Who’s building right now but needs more customers? Drop your product below. People’ll check it out 👇
2
2
175
if you're not following Emmy yet, you're missing out on some good content.
:D :D :D I have 68 followers. Booooom. Love it. I really enjoy the interactions I have with ya'll too. It's fun and calming. Seriously. I love to see the stories of success, the races, what tony's gonna do next, the hermes updates, it's great.
5
92
I'm at the park with my daughters. they're playing on the swings. I'm building an agentic service for SMBs, from the bench. Life is good. Praise God!!
2
5
51
I am in there, somewhere 😊
what an AMAZING community
2
7
321
Kyle French retweeted
You need to be delusional.
80
1,612
63
14,403
568,319
what are you building?
ok 𝕏, quick one 👋 show me what you're building 👇 I'll feature the top 3 in this week's StartupBase newsletter, straight to 15.1k subs. it's free marketing 🔥
1
1
112
Millennials - we are a privileged bunch We got to experience social media after our frontal lobes were developed and before it was an INDUSTRY We get to now experience AI after we’ve acquired real skills and before we’re old and stuck in our ways Accelerate accordingly
43
77
12
1,054
50,891