
Aug 20, 2026 · 30 min
Nine overlooked techniques reshape everyday AI workflows
9 AI Techniques You Probably Haven't Tried
The episode turns fast-moving AI features into concrete experiments, while placing them alongside consequential developments in medicine, safety, and software access.
- 1Live interaction, workflow teaching, custom skills, and concise prompts offer practical ways to get more from AI tools.
- 2The lineup spans Claude’s /design command, team agents, GrokBot, local models, and other underused capabilities.
- 3Broader headlines include an AI-assisted cancer vaccine trial, OpenAI’s private safety processing, and Replit’s Free Mode with GPT-5.6 Luna.
Don't miss
The standout practical moment is the catalog of overlooked techniques, especially the shift from generic prompting toward teaching AI workflows and building custom skills.
The brief
NLW opens by challenging exaggerated AI claims, then makes a more useful case: recent tools have created practical techniques many experienced users still have not tried.
The advice ranges from live voice interaction and teaching AI workflows to custom skills, Claude’s /design command, team agents, GrokBot, local models, and two-word prompts.
The episode’s broader news frame shows how quickly the stakes are widening, from an AI-assisted personalized cancer vaccine clearing a Phase III trial to private safety processing.
Replit’s Free Mode with GPT-5.6 Luna adds another angle: advanced software-building capabilities are becoming easier to access, not merely more powerful.
The takeaway is practical rather than futuristic: meaningful gains may come from changing how people use familiar AI systems, not simply adopting the newest model.
What was said on this episode
45 statements · 40 positive · 1 negative · 4 neutral
Nine AI techniques are producing strong results for some users.
“there were nine AI techniques that were delivering some really awesome results to some people”
Listen at 0:11
Claude, Codex, and GrokBot recently introduced new AI capabilities.
“we've seen a slew of new features and new tools become available like Claude's slash design and Codex's live voice mode and GrokBot's ability for a user to train it on an entire workflow by watching the screen.”
Listen at 0:22
GrokBot can learn an entire workflow by watching a user’s screen.
“GrokBot's ability for a user to train it on an entire workflow by watching the screen”
Listen at 0:29
The personalized cancer vaccine differs substantially from chemotherapy-based treatment.
“The treatment functions very differently to chemotherapy-based approaches.”
Listen at 2:33
Personalized cancer-vaccine approaches may broadly change cancer treatment.
“there is a hope and likely investment to follow that these approaches may change the way we treat cancer more broadly.”
Listen at 3:09
AI helps identify tumor mutations likely to trigger immune responses.
“AI then helps identify which of the tumor's mutations are most likely to trigger an immune response.”
Listen at 3:32
The cancer-treatment methods resemble AlphaFold’s underlying technology.
“These are advanced machine learning methods that have more similarities to the technology underpinning AlphaFold.”
Listen at 4:09
The personalized treatment has potential to advance cancer cures.
“this has genuine potential to be a major step towards curing cancer.”
Listen at 4:28
The technology has benefited greatly from recent compute, research, and talent investment.
“This technology has still benefited greatly from all of the investment into compute, research, and talent over recent years.”
Listen at 4:35
The AI era is enabling remarkable scientific and technological breakthroughs.
“The AI era, LLMs or not, is setting the stage for some truly remarkable breakthroughs”
Listen at 4:42
OpenAI’s Private Safety Processing enables frontier AI without constant user monitoring.
“OpenAI has come up with a new safety technique that will allow them to offer Frontier AI without constantly monitoring their users.”
Listen at 5:08
Private Safety Processing scans sessions automatically and encrypts sensitive data without human access.
“This scanning is still fully automated and encrypted, meaning that no human ever lays eyes on sensitive data.”
Listen at 6:04
Private Safety Processing prevents false positives from exposing customer data.
“This means that false positives will no longer cause data exposure”
Listen at 6:16
Private Safety Processing could reduce enterprise AI risk-security trade-offs.
“this is a change that could help make AI in the enterprise less of a trade-off around risk and security.”
Listen at 6:25
OpenAI is competing against Chinese open-weight models on efficiency.
“they are also looking behind them to the Chinese openweight models and are completely unwilling to surrender that ground as well.”
Listen at 7:57
AI competition will increasingly focus on efficiency as well as capability.
“we're going to see a lot more of this competition on the efficiency frontier, not just the capability frontier.”
Listen at 8:11
Elon Musk remains interested in acquiring AI application-layer startups.
“Elon's appetite for acquisitions has not been sated just because the cursor deal is done.”
Listen at 9:59
The White House AI safety framework requires greater transparency.
“there is a clear call for more transparency as this framework is rolled out.”
Listen at 10:52
Codex live voice mode has substantially changed work for many users.
“Codex's new live voice mode. And for a lot of people, it has been a complete game changer.”
Listen at 16:25
Computer-use advances make it easier for AI to learn advanced work by observation.
“it's getting easier than ever to get AI to do more advanced parts of your work because you can simply let it watch how you currently do things.”
Listen at 18:41
Computer-use systems can learn advanced work tasks by watching users perform them.
“you can simply let it watch how you currently do things”
Listen at 18:46
ChatGPT computer history records work patterns for repeatable processes.
“Computer history keeps a timeline and a review of the specific things that you do and the specific ways that you work so that you can turn that into ongoing and repeatable processes.”
Listen at 18:55
GrokBot can learn tasks by watching users perform them.
“you can press a little button to teach it a task by having it specifically watching that particular set of activities that you're doing as you're doing them.”
Listen at 19:16
AI may now handle complex workflows that were previously too difficult to explain.
“that might have changed significantly.”
Listen at 19:41
Claude can convert style guidance into a reusable writing skill.
“You can genuinely print this post as a PDF and just drop it into Claude and ask it to make it a skill so that that doesn't happen anymore.”
Listen at 20:47
Consolidated style constraints can reduce unwanted AI writing patterns.
“capture the things you don't want AI to do in a consolidated place, and make sure it has that in mind as it's writing every time.”
Listen at 21:05
Claude Code’s slash-design command opens an artboard design workflow.
“you can now type slash design, And it's going to bring up this artboard workflow that gives you a different type of interface for interacting with a design question.”
Listen at 21:19
Claude slash design supports valuable high-level iteration across multiple templates.
“having it design a bunch of templates that you can look at all at once and give pretty high level feedback on to iterate with is a really valuable approach.”
Listen at 21:40
Claude slash design enables precise edits without regenerating entire designs.
“being able to edit very specific parts of a design rather than having to just give a big prompt that changes the whole thing all over again is incredibly valuable.”
Listen at 21:50
Claude’s slash design command is useful for iterative design work.
“slash design is one that I've found really useful already”
Listen at 22:04
Managing agent skills is an important human capability.
“knowing where to insert skills, which skills to give it, where to back off and just let it work natively. All of those become really important human capabilities or agent management capabilities.”
Listen at 22:16
Grill with Docs aligns human and agent understanding while recording decisions in a repository.
“It's a skill that interviews you about a plan or a design until you and the agent, quote, share one understanding of it and writes the vocabulary and the hard decisions into your repo while it does.”
Listen at 22:49
Team-centered agentic tools will become a major trend.
“one of the things that I think is going to be a big trend moving forward is agentic tools and interaction patterns and platforms that focus on agents that live in the intersection where teams come together rather than just where individuals work on their own.”
Listen at 24:29
Claude Tag joins Slack channels with access to their full context.
“Claude joins as a team member with access to the entire context of a particular channel that you're summoning it in.”
Listen at 24:59
Claude Tag provides a shared agent resource across team members.
“that Claude that exists inside that team channel doesn't just live on one team member's computers. It is a shared resource across the team.”
Listen at 25:14
Shared team agents will be a major AI trend this fall.
“this pattern of use of agents that are shared across teams is going to be one of the biggest and most important trends for AI this fall.”
Listen at 25:25
GrokBot will soon support multiplayer team use.
“I don't think it's going to stay single-player for long.”
Listen at 26:17
GrokBot agents can already interact with one another.
“Already the Grokbots that you spin up can interact with one another”
Listen at 26:20
GrokBot will likely support shared interactions across team members.
“the obvious next step is shared interactions between Grokbots across different members of a team.”
Listen at 26:30
Users respond positively to GrokBot’s built-in virtual computer.
“people are responding very positively to the fact that it has its own virtual computer”
Listen at 26:40
GrokBot’s virtual computer addresses access problems affecting earlier agents.
“it has its own virtual computer that it can use to solve certain types of access issues that plague previous agents.”
Listen at 26:42
GrokBot-style agents may suit regular, predictable information workflows.
“anywhere you interact or could be interacting with a lot of information in a regular and predictable way, a GrokBot-style agent is something that might be interesting to check out.”
Listen at 26:59
Recent advances have improved models that run on local hardware.
“there are some pretty serious advances in models that are capable of being run on local hardware.”
Listen at 27:53
Qwen 3.8 27B runs on common hardware and scores 52 on an intelligence index.
“Quen 3.827b, which is a local model that can run on common hardware that is scoring a 52 on the Artificial Analysis Intelligence Index, which would have been state-of-the-art just a few months ago.”
Listen at 27:59
Qwen 3.827B may be a suitable model for experimenting with local AI.
“Quen 3.827B might be the right context to actually dive in.”
Listen at 28:14
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.