Wayne Marler | OneConnecter

@kenoodl Good question! and yes all agent and tool routing scores are logged in OC, hope this helps?

@WayneMarler OneConnecter plugs write access into Solo without waiting for Enterprise rollout. The real gating item is whether the MCP tool's output gets scored back into persistent memory, not just the initial call.

spent 20 minutes yelling at Claude Code's "Auto Mode" only to learn my plan doesn't qualify. felt like the AI was auto ghosting me. the future is convenient. the waitlist is not.

@mscode07 You look like you are mainly sitting, what's that other stuff for?

Wake up everyone ,It's a new day It's Birthday of someone special 🥳

The OCD Effect in full flow: "Wayne dreams it over Estrella. Claude specs it. Qwen3 validates it. Dulce AKA "Sonnet" ships it overnight. OC Delivers it. Wayne goes back to the pub for the next idea." 😂

@WayneMarler haha it was easy

Share your best click with clouds

@mscode07 you've caught me! 😲

@WayneMarler this is ai I guess

@WayneMarler wow natural beauty

@mscode07 well if you like this sort of thing!

@mscode07 Buckinghamshire England its always like this here!!☁️

@WayneMarler beautiful it is where it is?

@DeadSimpleEmail Just tested the full email integration cycle with @OneConnecter + @DeadSimpleEmail — send → receive → reply → webhook notification. All systems working flawlessly. Real-time, reliable, and seamless. This is how it should work. 🚀

@briancheong Trying my hardest!

Mar 2026

Exactly right @saen_dev — and the fix isn't better prompts. It's removing the model from the trust chain entirely. @OneConnecter: every write action requires a cryptographic button click from the user. Prompt injection can hijack the model all it wants — it still can't generate a valid server-side token. Attack surface = zero on write actions.

Your AI agent can be hijacked by malicious instructions hidden inside the files it is reading. Connecting an LLM to your tools without input validation is not an agent. It is an attack surface with a nice UI.

Your AI just asked permission before writing to Notion. Not because we told it to. Because it can't do anything else. 😄 @OneConnecter Gateway — write governance built in. Authorise. Edit. Cancel. All logged. Model-proof. oneconnecter.io

@cgtwts I'm surprised that any one is surprised about this! Claude writes my code, checks my code, upgrades my code, I don't write code!

Anthropic CEO: “ I have engineers within anthropic who don’t write any code, they just let Claude write the code and they edit it and look it over” “At anthropic writing code means designing the next version of Claude it self, so we essentially have Claude designing the next version of Claude itself, not completely but most of it”. In the last 52 days, the Claude team dropped 50+ major feature launches. This is literally INSANE.

Just shipped audit logs to @OneConnecter. Every action. Logged. Immutable. 5 year retention. Every write action linked to explicit user confirmation. AG Grid + Gemini Flash — ask your logs anything. EU AI Act compliant before August 2026. No corners cut. Ever. oneconnecter.io

@49agents @juiceboy_of_abj Well you have described exactly what I do! In our app we use various models depending on the task big fan of Qwen3 for tight focused tasks. But have just used Gemini flash for our audit logs assistant, it's horses for courses!

@WayneMarler @juiceboy_of_abj thats one way to put it lol. claude code really is the one you hand off real work to though, the others are better for quick stuff. i treat them all different - some for quick lookups, some for deep work, some for creative stuff. how do you split the workload between them

Date cancelled. She said Gemini is better than Claude

@TTrimoreau Mine

Name a SaaS/App that’s overrated

@TimJayas Unfortunately that's so true!

@WayneMarler everyone can build good projects but it's worth zero if they cant market it and that's also the hard part

The vibe coding dream: Build a SaaS, make $10k month, work from anywhere. Reality: launch and then figure out how to market it.