25 years making computer systems meet regulatory requirements in clinical trials. Data. AI. Chess & Music ❤️ Building. Inferring. Living. Field notes 📓
Joined January 2009
- Tweets1.1K
- Following381
- Followers111
- Likes1.3K
Kayode retweeted
Worried AI is coming for your job? Then it's probably not the time to ask for a raise.
My literal advice to friends and family: become the best AI person at your company. You'll be the last one fired. newsletter.platypuseconomics…
Kayode retweeted
I am seeing a lot of how to's and how grokbot is amazing as a personal assistant ai. But it's not fit for use. They need to call it out. The usage credits don't last its utterly ridiculous and no one is responding. Loathe him or like him at least @thsottiaux did something about poor service at Openai
Kayode retweeted
wtf is an agent harness.
An agent is a model plus a harness.
Model reads the context and picks the next action. The harness is all the code that turns that choice into work.
Claude Code, Codex, Antigravity, OpenClaw and Hermes Agent are all harnesses.
8 common parts every agent harness shares:
1. The loop
• Sends the context to the model and runs whatever tools it asks for
• Feeds the results back in for the next step
• Decides when the task is finished
2. Tools and sandbox
• Give the model a terminal, files, a browser and your connected apps
• Run every task inside its own isolated computer
• Keep the damage from a bad command inside the sandbox
3. Subagents
• Take side tasks off the main agent and run them in parallel
• Start each one with a fresh context window
• Send back only a short summary, so the main context stays clean
4. Context management
• Decides what the model sees on every turn
• Loads skills and files only when the task needs them
• Summarizes old history before the context window fills up
5. Memory
• Stores progress notes, git history and skills on disk
• Lets a new session read what the last one did
• Picks the work up where it stopped
6. Guardrails
• Hold risky actions like emails, payments and deletes for your approval
• Keep real credentials outside the sandbox
• Give a prompt injection no credentials to steal
7. Verification
• Stops the agent from handing you a summary that only sounds finished
• Runs tests, browser checks or a separate grader first
• Lets the agent call the task done only after those
checks pass
8. Observability
• Traces every model call and tool call
• Spots the agent repeating the same failed edit
• Pushes it to try a different approach
follow @Saboo_Shubham_ for more on Agent Harness.
Kayode retweeted
If Claude wrote your website's privacy policy, it may have handed 𝘆𝗼𝘂𝗿 𝘄𝗵𝗼𝗹𝗲 𝘀𝘁𝗮𝗰𝗸 to attackers.
Someone found that when Claude writes a privacy page, it lists every third-party tool you've plugged in and adds promises you never meant to keep. Press it and it says it writes from your customers' side by default.
A page that spells out which services you use hands attackers a checklist for known exploits, phishing, and credential stuffing. Send this to Claude Code or Codex to pull the traps out of your policy pages first:
"Read my privacy policy and terms pages. Quote every promise, guarantee, named person, and vendor, and list them for me to confirm. Swap tool names for categories (hosting, payments, analytics). Change nothing until I approve."
Kayode retweeted
ChatGPT's Dots do almost everything apart from actual work. The following is a guide to the words, and what I feel like they mean, that it uses. But I just needed to get on with the work, come back to me saying it's done, rather than all the following.
“Checking”
Looking at something specific. Tell me what you looked at and what you found.
“Verifying”
Comparing a claim with evidence. Which source, which version, and did it match?
“Validating”
Seeing whether the result meets the requirements. Something can run perfectly and still do the wrong thing.
“Testing”
Running it under stated conditions. What passed, what failed, and what haven't you tested?
“Reviewing”
Reading or inspecting the work and making a judgement. Give me the findings, rather than just saying a review happened.
“Auditing”
Working through a defined scope. If you looked at one folder, say one folder. Don't quietly turn that into “the whole system”.
“Reconciling”
Comparing records and explaining the differences. Which items match, which don't, and what still needs fixing?
“Triaging”
Deciding what needs attention first. Useful, but I'd also like the important work to get done.
“Investigating”
Trying to establish the cause. Tell me what the evidence supports and where you're still guessing.
“Researching”
Gathering information to answer a question. What did you learn, which sources support it, and how does it help finish the task?
“Delegating”
Giving the work to another agent or person. Has someone actually picked it up, and who is following through?
“Monitoring”
Watching for something to change. What is being watched, how often, and what happens if it fails?
“Blocked”
Unable to take a particular next step. Tell me exactly what is missing and carry on with the work that doesn't depend on it.
“Implemented”
The change has been made. Has it reached the place I actually use, and does it work there?
“Done”
The thing I asked for is finished, in the right place, and I can use it. If I asked for a draft in Typefully, show me the draft in Typefully.
The last two are what I want to hear and way less of the others. I'm sure this is related to the different types of sessions it has to go through.
If we are to use the dot sessions, they need to be more powerful and be able to do a lot more. I'd also argue that they need to have higher numbers of parallel work and a queuing system that puts 30 things in a queue of 10 lanes.
Kayode retweeted
How can healthcare workers focus more of their time on patients and less on paperwork?
In our latest episode of Dialogues on Technology and Society, @GoogleLabs VP @joshwoodward sits down with @EricTopol to talk about how technology can help alleviate healthcare worker burnout and accelerate scientific discovery by returning the "gift of time" to frontline care.
Watch the conversation → goo.gle/4dLZNfI
Kayode retweeted
You can now run Perplexity Computer fully local on your Mac with no cloud based inference, without WiFi too. Enjoy!
Kayode retweeted
‼️ BREAKING: Revolut's app and cards have stopped working across Europe.
Google Cloud's London region is affected by network failures. Revolut, and many other services, runs on Google Cloud.
Revolut says money and data are safe but has not named a cause.
Kayode retweeted
Dang, seems like @opencode got hacked. Change your keys!
Kayode retweeted
We’re exploring how AI can accelerate health science research.
While AI has made significant breakthroughs in formal sciences like mathematics, health data science depends on imperfect real-world data and consequential choices about cohorts, variables, and causal assumptions.
To explore how AI can assist health researchers, we built a human-led research scaffold powered by Muse Spark 1.3 and Muse Wonder with access to 5M retrospective de-identified electronic health records and publicly available data. A cohort of medical experts used the system to conduct original research, accelerate hypothesis generation, code execution, and manuscript drafting while maintaining full authority over scientific decisions.
Today, we’re sharing 10 research manuscripts produced through this exploration, spanning cardiology, oncology, infectious disease, emergency medicine, and more.
Read more in our research blog: go.meta.me/3U06wvD
Kayode retweeted
We’re getting closer to a world where calling customer support might mean never speaking to another human.
Are people actually ready for that?
Kayode retweeted
An AI health assistant in China reportedly surpassed 100 million users, offering services including free medical-report interpretation. While not a replacement for doctors, it shows how quickly access to health information can scale. The opportunity is enormous.
Kayode retweeted
Cursor is now part of The X Subscription.
Current Cursor subscribers can keep their plan or switch to The X Subscription.
Paid plans share one usage pool across Cursor, Grok Build, and other Grok products, with twice the coding usage on Grok models as the equivalent Cursor plan. Premium, Super, and Ultra also include Grok Bot.
To make room for more Grok usage, third-party models in Cursor move to API pricing on the new plans. We'll keep working to make your usage go further, with more compute, better inference, and more token-efficient models.
The X Subscription is available at cursor.com/pricing.
