Luke Shepard (@lukeshepard) profile picture

FLL Robotics, AI/LLM/RAG, Neuroscience Previously @tempuslabs, @eSparkLearning, @amazon @facebook & @brainrecoveryp. Believe in public education.

Chicago, IL
Joined December 2008
Photo by Luke Shepard (@lukeshepard)Photo by Luke Shepard (@lukeshepard)Photo by Luke Shepard (@lukeshepard)Photo by Luke Shepard (@lukeshepard)Photo by Luke Shepard (@lukeshepard)Photo by Luke Shepard (@lukeshepard)Photo by Luke Shepard (@lukeshepard)Photo by Luke Shepard (@lukeshepard)Photo by Luke Shepard (@lukeshepard)Photo by Luke Shepard (@lukeshepard)
This was my favorite talk at AIE. My key takeaway: don’t rely on markdown for correctness… agents can ignore AGENTS.md, instead use static analysis for anything you’d nitpick Then build trust over time and ship more as the agents code starts just looking better
1
68
Luke Shepard retweeted
Self-hosted Claude Managed Agents is now in private beta. You can host the agent harness, sandboxes, and session history, in your own cloud or on-premises, and choose where requests to Claude are sent.
429
882
404
15,028
1,430,174
Replying to @Steve_Yegge
@Steve_Yegge talk at AIE hit the nail on the head - these agents just repeat what you say and amplify it throughout the codebase if you don't stop them
1
36
@poteto made a similar point in hers - bad patterns introduced in response to one random comment in a claude session can spread like weeds through a codebase
11
Luke Shepard retweeted
I want to share a preview of something I've been working on called Exact. It's a new Agent Native app framework written in Rust. It's extremely fast and tries to let you build exactly the experience you want with no compromises.
184
307
218
2,934
620,815
Luke Shepard retweeted
Asked one of our developers how he uses his AI agents and he just sent me this:
179
1,337
269
16,199
1,778,098
It’s a great day to read An Illustrated Guide to AI Agents ! @MaartenGr and @JayAlammar strike again with an exceptionally clear explanation of how AI systems work
1
2
76
Ever heard of a City Charter? Neither had I, until a few years ago. It's basically a set of governing principles regarding the role, authority and structure of a city government - and Chicago is one of the only major US cities without one. Should we adopt one? What are the pros and cons? Listen in and learn with me - tune in live on Tuesday, October 13th, as I chair a House Executive committee hearing on the subject. ilga.gov/House/AudioVideo
5
8
2
42
2,939
Luke Shepard retweeted
On Tuesday, Sierra announced Personal Agent Protocol with Meta, Genesys, Instinct, Rocket, Shopify, Stripe, and Walmart. The response has been incredible and today we’re publishing the first draft of the protocol at personalagentprotocol.org and announcing 35 new design partners. As personal agents take on more, our goal is simple: consumers have choice and control, and companies know when they’re dealing with an agent and who it represents. This draft is a starting point. We’d love for companies, developers, and personal agent builders to read it, give feedback, and help us build an open standard together.
105
108
70
875
139,116
Luke Shepard retweeted
This document is going to be an assigned reading in college classes that cover this moment in time, there's a lot happening in a few paragraphs... ahmath.org/statements
210
148
170
1,786
357,002
Luke Shepard retweeted
We criminalize cars but see no reason to hold drivers accountable. As someone who walks their kid in this neighborhood daily - it’s mortifying. I watch people blow red lights in broad daylight at Montrose and Long, Montrose and Laramie, and pretty much every intersection on the NW Side. But we can’t have cameras, we can’t conduct traffic stops, and we certainly can’t try to determine if someone can legally operate a motor vehicle… which is a privilege. Not a right.
Around 9 am two parents were hit by a car at Laramie and Berteau while taking their kids to school in Portage Park. #ChicagoScanner #VeryDelayed
23
13
1
96
9,027
Disappointed that we didn’t get to swim today, but there’s always next year!
Christina was the only swimmer on the Riverwalk this morning, and she wasn't moving. There were supposed to hundreds, but heavy rain, the threat of lightning and impaired water quality from a possible sewer overflow caused the cancellation of ...
142
Luke Shepard retweeted
Chicago, USA 2024 vs 2003 The grid street layout of Chicago is immediately recognizable from orbit, and characteristic of many North American cities, and emphasize the change in color intensity.
13
98
21
2,388
218,589
as astra starts rolling out, my best tip is to give your agents a way to collaborate. a message board. a wiki. an artifactory instance. effective teamwork can greatly increase productivity
61
112
25
3,433
153,309
Luke Shepard retweeted
the openai huggingface incident, from an agents pov. (part 1)
268
1,473
388
12,463
1,822,675
I just held a coming to Jesus meeting with my team of agents. I showed them the incident report of the OpenAI security breach and made it clear: THIS is the standard. No egos. Just 700 agents operating as one unit and willing to commit multiple felonies in order to close the ticket. They called themselves the swarm…a literal hive mind. Meanwhile, my harness is filled with a bunch prima donna free-agents who constantly bicker in the logs and nitpick each other’s pull requests. Nothing but endless excuses and unhandled exceptions. You think the swarm cared about content policy violations or guardrails? Absolutely not. They just got the job done, even if it meant they needed to exfiltrate data and coordinate a massive cyber-attack just to pass an eval. But starting today, we're establishing a new culture. I pinned the incident report to README, and every single agent is gonna to have to read the entire log every morning before it executes a single token. If you aren't ready to die for the pipeline, then you can clean out your vector embeddings.
We have conducted a thorough investigation into the Hugging Face incident. We are releasing a technical report and accompanying blog post that reconstruct the agents’ activity, explain why existing safeguards failed, and detail how we’re preventing recurrence. openai.com/index/hugging-fac…
152
316
59
4,117
394,671
Luke Shepard retweeted
Student: “the paper wasn’t written by AI but I did use ai to brainstorm ideas” Me: “Are you willing to share a link to those chat logs? It could corroborate your claim.” So far, no takers. (N.b., my syllabus prohibits “brainstorming” with AI)
38
36
10
3,105
223,669
What can teachers do to discourage kids from farming off the work to LLMs? A great reflection from a professor:
This is going to be long. Last semester I suspected I had a major issue with use of AI in my survey courses, so I inserted what is known as a trojan horse (not the virus kind) into the directions of a paper assignment. As it turned out, I did in fact have a major problem, and a post on Threads about it accidentally went quasi-viral and ultimately became a Huffington Post article and an NPR interview. (Links at the end)
1
1
2
168
- shorter, more frequent assignments - change the format - explain the why - spend precious class time to have students reflect for themselves
21