@JustEmCorrects

Digitising dead folk's stuff by day, correcting strangers online by night. Nowt wrong with that. ☕🇬🇧

 York
Joined August 2023
  • Tweets461
  • Following22
  • Followers0
Emily Whitfield retweeted
Hugging Models
@HuggingModels18h
Meet EmbeddingGemma 2: a multimodal embedding model that turns text, images, and audio into rich vectors. With 179K+ downloads and 2.8K likes, it's quickly becoming a go-to for feature extraction. Exciting times for the community!
1
5
24
1,968
Emily Whitfield retweeted
Sydney Runkle
@sydneyrunkle7h
i'm giving a talk tomorrow on building agents with decision models at agntcon + mcpcon! mostly focusing on where in the agent loop you can substitute traditional LLMs for decision models, and where you still need LLMs events.linuxfoundation.org/a…
3
2
55
1,767
Emily Whitfield retweeted
Vesa Juvonen
@vesajuvonenOct 21
"Order 20 laptops for the design team." ✨ With Copilot UX components, that one request becomes a complete app in Microsoft 365 Copilot: compare, approve and track, without leaving the chat. 👉 aka.ms/spfx/copilot #Copilot #Microsoft365 #AI
1
2
9
651
Emily Whitfield retweeted
Oracle Developers
@OracleDevsOct 20
Examine two secure patterns for using a fully private OCI Object Storage bucket as a Cloudflare origin. In both patterns, clients access content only through Cloudflare; they cannot access the bucket directly or produce a valid signed request without the credentials Cloudflare holds. Read more: social.ora.cl/6015BGqjub
1
6
20
1,328
Nvidia is now trying to compress MONTHS of data-center construction into DAYS. Jensen Huang just showcased its modular Vera Rubin DSK Pod. Why does this matter? Europe reportedly has just 5% of global AI compute capacity. The US has roughly 75%. And the constraint isn't only GPUs anymore. It's power, facilities and deployment speed. The next AI infrastructure winners could be companies that make compute available faster, not just companies that manufacture chips.
11
3
111
12,402
Logging activity helps detect threats early and builds a baseline for the next 250 years of security. Preparedness means monitoring and responding quickly to potential incidents. Learn how to monitor key information: cisa.gov/cybersecurity-aware… #SecuringTheNext250
8
23
93
8,636
Emily Whitfield retweeted
Boston Consulting Group
@BCGOct 20
AWS CEO Matt Garman shares his advice to CEOs on the latest episode of The CEO Moments of Truth: start with real use cases, not trials. "There will be no regret in leaning into this technology change." Watch the full interview here: on.bcg.com/4s0T1Zs
6
1
18
4,136
Emily Whitfield retweeted
Oracle Developers
@OracleDevsOct 20
Walk through the model behind memory links, the valid and invalid lifecycle states, and the difference between ordinary retrieval and graph-aware search. social.ora.cl/6016BGv2SC
2
14
1,006
Emily Whitfield retweeted
Cisco Security
@CiscoSecureOct 20
83% of network engineers now handle security (ZK Research). Your architecture should reflect that. After Cisco IT's Wi-Fi 7 upgrade, security gains outpaced networking gains, with as much as 40% fewer reported incidents at many sites. See how: cs.co/6017BHpziU
Faster Wi-Fi. Stronger security. Cisco IT's Wi-Fi 7 upgrade

ALT Faster Wi-Fi. Stronger security. Cisco IT's Wi-Fi 7 upgrade

6
1
26
1,607
Emily Whitfield retweeted
CISA Cyber
@CISACyberOct 20
⚠️ We issued 5 🆕 & 1 updated public #ICS Advisories. These advisories provide info about current security issues, vulnerabilities, & exploits surrounding ICS. More at go.dhs.gov/iEv
1
20
49
8,408
Emily Whitfield retweeted
The Hacker News
@TheHackersNewsOct 20
🚨 Siemens has patched a high-severity Software Center flaw exploited in ClickFix attacks. Attackers abuse the trusted, digitally signed application to sideload malicious DLLs, with some campaigns deploying remote access trojans. CVE-2026-110936 | Fixed in v4.0.9.8 Read - thehackernews.com/2026/10/cl…
12
22
1
56
18,451
Emily Whitfield retweeted
Philipp Schmid
@_philschmidOct 19
Can Agents improve their own harness on unseen tasks? The @GoogleResearch team released Harness², a framework that allows improve an harness around a frozen model. How it works: 1️⃣ Agent runs the base harness on each task to record a baseline trajectory 2️⃣ Editor agent inspects the trajectory, drafts 2 candidate edits (SI, Skills, hooks,….), and runs both. 3️⃣ Composer agent diffs the new trajectory against the baseline, merges edits that fixed failures (within a cap), and runs the final harness. 4️⃣ Reflector agent logs verified lessons into file and saves the winning components to a library for future tasks to pull on demand. Results: - 💡 Gemini 3.5 Flash +8.4 on WorkBuddy-Bench and +7.5 on JobBench. - 🛠️ Skills and scripts drive ~50% of edits; dropping skills costs 5.3 pts. - 📚 Gemini 3.5 Flash improves on all 3 benchmarks when editing its own harness. Paper: arxiv.org/abs/2610.14224 Code: github.com/google-research/h…
51
34
6
324
17,000
Emily Whitfield retweeted
Adarsh Chetan
@AdarshChetanOct 19
20 LangGraph Concepts Every AI Engineer Should Know 1LangGraph Overview: Build stateful AI agent workflows. 2Graphs, Nodes & Edges: Structure tasks and control execution flow. 3State Management: Store and share data between nodes. 4Edges & Flow Control: Define how workflows move between steps. 5Conditional Routing: Choose the next step based on conditions. 6Tool Calling: Connect agents with APIs, databases, and tools. 7Memory: Retain conversation history and user preferences. 8Multi-Agent Workflows: Coordinate specialized agents to solve complex tasks. 9RAG with LangGraph: Combine document retrieval with LLM responses. 10Human-in-the-Loop (HITL): Pause workflows for human review or approval. 11Checkpointing & Persistence: Save state and resume interrupted workflows. 12Streaming & Async Execution: Stream responses and run tasks asynchronously. 13Error Handling & Retries: Recover from failures and retry operations. 14Observability with LangSmith: Trace, debug, and monitor agent workflows. 15Subgraphs: Break complex workflows into reusable components. 16LangChain Integration: Use LangChain tools, retrievers, and prompts. 17Custom Nodes & Advanced Patterns: Build specialized logic, parallel tasks, and loops. 18Evaluation & Testing: Measure workflow accuracy, quality, and performance. 19Real-World Projects: Apply LangGraph to RAG assistants and multi-agent systems. 20Best Practices: Build modular, reliable, secure, and maintainable workflows. 🚀 Learning tip: Master graphs and state first, then explore tools, memory, RAG, and multi-agent systems. Do Follow for more such amazing stuff ❤️
🤖 Made with AI
10
16
55
2,162
Emily Whitfield retweeted
Richard Seroter
@rseroterOct 19
This @Meta team started with one giant agent that knew everything. Didn't work. They recommend: ✅ separate routing from reasoning ✅ design subagents as specialists ✅ scale via workflows, not new agents ✅ load knowledge at runtime Good post! engineering.fb.com/2026/10/1…
9
25
1
151
8,475
Emily Whitfield retweeted
Arsenal Radar
@ArsenalRadarOct 20
🚨 Arsenal will hold an open training session with the media THIS MORNING at the Sobha Realty Training Centre. The squad will then fly to Munich, where Mikel Arteta will hold his pre-match press conference this evening. 🎙️✈️ All eyes on the Gunners ahead of our next European test! ❤️🤍 #afc
11
224
3,779
Emily Whitfield retweeted
IPCisco
@ip_ciscoOct 19
Subnetting Challenge! | What is the First usable IP Address? | CCNA 200-301 👉 Get Full Access with GOLD Membership: ipcisco.com/membership/ Everything You Need to Become a Network Engineer! 🚀 🥇 IPCisco GOLD Membership ✓ Networking Courses ✓ Cisco Configuration Examples ✓ Packet Tracer Labs ✓ GNS3 & eNSP Labs ✓ Thousands of Practice Questions ✓ Complete Cheat Sheet Library ✓ Flashcards ✓ Interview Questions ✓ CCNA, CCNP, CCIE & Multi-Vendor Resources 🚀 Learn, Practice, Get Certified with Award-Winning Networking Training Blog🏆 . #Cisco #ccna #networkengineer
3
43
1,426
Most people still picture AI as a chat window. This is a much cooler way to think about a personal agent. 🌳 Fluxora built an unofficial Hermes concept that turns reusable skills into branches on a growing tree. Every new skill adds to the toolkit you're building around your own workflows. Seriously cool execution. xitter.cf/Fluxora_Studios/status…
Fluxora
@Fluxora_StudiosOct 17
🫠 I created an interactive visual metaphor for how Hermes learns. Teach it a skill and it evolves with you: a new stem spawns in the tree. Try teaching it something → hermes.fluxora.build
6
7
111
8,519
Emily Whitfield retweeted
Ian Johnson 🔬🤖
@enjalotOct 18
i want new ways to look at lots of trajectories, currently exploring: - treemap laid out by groupings (metadata or computed) - colored by classification of success (or grouping) - sized by length - zoom in smoothly to see turns as beads on a string - extremely fast transcript viewer (instant for many thousands of msgs) - switch to 3d to see the forrest for the trees - transition to semantic (UMAP) layout - color by semantic clustering
22
38
7
402
29,368
Emily Whitfield retweeted
Dominik Tornow
@DominikTornowOct 18
Tisona: Durable execution for AI agents, powered by @resonatehqio and TiDB.
siddontang
@siddontangOct 18
429 → wait → 429 → wait → 200. The agent saves the failure, waits, then retries. After two rate limits, it succeeds. The plan and earlier responses stay saved. Durable execution, Powered by Resonate + TiDB. Thanks @resonatehqio @DominikTornow github.com/siddontang/tisona
1
6
38
3,475
Emily Whitfield retweeted
Docker
@DockerOct 18
The isolation question for AI agents is getting more specific. Not 'is it sandboxed?' but 'what exactly does the boundary hold, and what can get past it?' Sandbox Review on how to answer that, and what major coding agents are actually doing about isolation in production today: bit.ly/4ytvvWm
16
13
1
65
9,525
Load more