
Sign up to save your podcasts
Or


I realised recently that the time I was spending sharing my ideas was actually strangling the sharing of ideas themselves. In this BrainStream, I'm opening up about why I'm working to automate my entire production pipeline—from transcripts to clips— to keep sharing. I think we're entering an era where 'janking' together AI tools like Claude Code and Remotion isn't just a hack; it's the only way to keep up with the ROI of our own thoughts. I wonder if the future of the UI isn't just talking to our computers, but having them understand the visual and emotional context of our work before we even ask.
Timestamps(0:00) The Guatemala Vision Pro Backlog
(2:15) When the ROI of Sharing Doesn't Make Sense
(4:10) From 2017 Manual Edits to AI Pipelines
(6:20) Automating the 'Boring Stuff': Clips & Captions
(8:10) The Post-Siri World: OpenClaw & Agentic Filesystems
Listen OnIf you enjoyed this episode, you can listen to more of my brainstreams on Spotify, YouTube, or Apple Podcasts.
Check out my website for more insights: https://avrahamrask.in/podcast
We’re finally breaking free from the 1980s desktop paradigm and moving into a world where creation isn’t tied to a workstation. I’m diving into how I finally cracked the code on local automation for the BrainStream, bringing us closer to a future of conversational agents that actually understand and build alongside us. From my early work on the Whisper project to the current shift in spatial computing, we’re exploring how we can stop being tethered to a desk and start leveraging high-fidelity AI collaboration wherever we go.
Timestamps[0:00] From GUI to Conversational OS
[1:36] The Whisper Vision: Untethered AI
[3:12] The High Cost of Manual Friction
[4:30] Custom Pipelines Over SaaS Bloat
[6:20] Local Video Clipping with Whisper JSON
[8:10] Iterative Media & Generative Assets
Listen OnIf you enjoyed this episode, you can listen to more of my brainstreams on Spotify, YouTube, or Apple Podcasts.
Check out my website for more insights: https://avrahamrask.in/podcast
The velocity of innovation right now is insane—we’re seeing a year’s worth of progress every week—and the only way to keep up is to automate the friction out of our systems. I’ve finally dialed in a pipeline that handles the heavy lifting of post-production and distribution with a single click, freeing me up to dive deeper into the high-stakes security projects I’ve been teasing. This episode is a direct look at how I’m using AI to reclaim my time, moving away from manual content management and back toward pure insight and execution. If we're going to build the future, we can't afford to be slowed down by the logistics of the present.
TIMESTAMPS:0:00 | Navigating the AI Acceleration Curve
1:45 | Why Content Friction Stalls Big Ideas
3:20 | The Utility of the Spoken Word
5:10 | Solving Local vs. Cloud Friction
6:55 | Video as Code: The Remotion Shift
8:30 | Closing the Agent-Driven Loop
LISTEN ON
Spotify: https://open.spotify.com/show/0gyQydQ9h5PM2gGUWy1PfC
YouTube: https://youtube.com/playlist?list=PLifoYer83q8ACtYWCtjhoOHGmfslydLM7&si=yKhmWU6J9t6j9USF
Apple Podcasts: https://podcasts.apple.com/us/podcast/id1726877985
Website: https://avrahamrask.in/podcast
In this year-end Brainstream, Avraham revisits a 2018–2019 video exploring how augmented reality and AI could transform everyday problem solving. Using the example of fixing a broken fridge, he unpacks multimodal AI—vision, sound, and conversation working together—and reflects on how concepts from seven years ago are becoming increasingly practical today. This episode also considers the evolving relationship between humans and AI in collaborative problem solving.
TL;DR
A seven-year-old video predicting AI-assisted, AR-driven home repairs highlights how multimodal AI—integrating vision, sound, and conversation—can empower humans to solve problems collaboratively without experts or cloud dependency.
🎧 Listen on Spotify | YouTube | Apple Podcasts
Listen to the full episode and explore more Brainstream → https://avrahamraskin.com/podcast
A quiet rumour about OpenAI tapping into Apple Health sparks a deeper question: what would a truly private, on-device health intelligence actually look like?
Listen on Spotify, YouTube, Apple Podcasts
More episodes → https://avrahamraskin.com/podcast
Timestamps
00:00 | The Rumour About OpenAI And Apple Health
01:06 | Why Access To Raw Health Data Actually Matters
01:45 | The Privacy Problem With Cloud-Based LLMs
02:26 | The Case For A True On-Device Reasoning Model
03:12 | Hardware Limits And Why Models Must Shrink
04:26 | What A Smarter Siri Should Really Be
05:14 | Health Metrics, Trends, And Real Coaching
06:24 | The Missing Link: Nutrition Data
07:12 | Rumours Of Apple HealthPlus
08:02 | What Real AI Coaching Should Feel Like
09:07 | Integrating Sleep, Training, Journaling, And Goals
09:48 | The Vision: A Private, Personal, On-Device Coach
In this BrainStream, I explore why we’re on the edge of a generational shift in how properties are secured. Not just sensors. Not just cameras. A true, always-awake operator built from VLMs, re-identification, and on-prem AI. I look at Ubiquiti’s position, emerging players, and why the “virtual guard” model is no longer sci-fi but an inevitable next step.
TL;DR
Security is moving from alarms to an always-awake, property-aware AI operators that verifies, learns, and predicts.
🎧 Listen Now at Spotify, YouTube, or Apple Podcasts
→ More at: https://avrahamraskin.com/podcast
Timestamps
00:00 | Why I Had to Record One More
00:21 | The Ubiquiti/UniFi Fascination
01:31 | The “Ultimate Operator” Concept
02:20 | Parameters → Description: A Real Turning Point
03:21 | Teaching an AI the Property (The Guard Analogy)
04:41 | From Virtual Guards to Site-Aware Intelligence
05:22 | Alarm Monitoring: The Missing Link
06:13 | Camera Verification and VLM Reasoning
08:48 | Prediction Instead of Reaction
09:51 | Closing the Loop: Where This Is Heading
#Ubiquiti #UniFi #VideoOperator #VLM #Re-ID #Alarm #Verification #PredictiveSecurity #On-Prem AI #ComputerVision #Security #Future
A concise, investigative tour of how security video evolved from passive cctv to intelligent, searchable footage powered by local ai and video-language models. I maps the technical lineage-motion sensing, smart detections, face and plate id, the “AI Key,” and scene-level vlm search-and explain why pattern discovery at scale is the next operational leap for site security and investigations.
“video that used to be passive now becomes a searchable narrative.”
🎧 listen on spotify, youtube, apple podcasts
🔗 more episodes → https://avrahamraskin.com/podcast
tl;dr
security cameras have graduated from passive recorders to active, searchable sensors. video-language models (vlms) and local llm-like agents enable natural-language scene search and condensed pattern visualisations-powerful for investigations but constrained today by compute and edge deployment. the next frontier is real-time, site-wide pattern detection running at the edge.
timestamps
00:00 | introduction and context
00:23 | the evolution: cctv → motion → smart detections
01:56 | face detection, license plates, and granular id
02:19 | the “ai key”: local llm-style analytics (what it adds)
03:35 | video-language models: frame description → search
05:03 | practical investigative tools and scene search examples
06:03 | pattern discovery: briefcam and condensed timelines
07:40 | limitations today: compute, edge, and the next step
09:56 | closing thoughts and what’s next
Most security systems still behave like they did 20 years ago-reactive, limited, and blind to the context hidden inside their own recordings. In this Brainstream, we explore why the real frontier in security isn’t better alerts or higher-resolution cameras, but AI systems that can learn a site’s patterns, behaviours, anomalies, and risks over months of recorded footage. This episode outlines the shift from “review after the incident” to “predict before it happens,” and why the intelligence trapped inside our footage is the most valuable, unused asset in modern security.
TL;DR
Security cameras shouldn’t just replay the past-they should understand it. When indexed, analysed, and contextualised, months of footage can power predictive, site-specific intelligence far beyond traditional monitoring.
🎧 Listen on Spotify, YouTube, Apple Podcasts
🔗 More episodes → https://avrahamraskin.com/podcast
Timestamps
00:00 | Opening: Why talk about the future of security
00:05 | Why this topic needs multiple videos
00:08 | A new product direction after years in the field
00:19 | The core problem: cameras are reactive
00:26 | Footage as an investigative tool, not a live one
00:34 | Tools like BriefCam and condensed investigations
00:51 | The inevitability of deep pattern analysis
01:17 | Rethinking what recorded footage really contains
01:26 | On-site storage vs cloud motion clips
01:48 | Why modern systems rarely store “everything”
02:14 | The hidden value inside long-term footage
02:27 | Thought experiment: downloading 6 months of footage into a guard
03:06 | Scale: 25–100 cameras, months of data
03:25 | What context a human misses vs what the data contains
03:58 | Reviewing footage: hours, days, weeks
04:25 | Pattern detection after the fact
04:54 | The industry’s stuck in reactive mode
05:02 | Moving from reactive to predictive
05:17 | Connecting dots before the incident
05:24 | Trends, anomalies, and site-specific patterns
05:34 | What good security guards actually do
06:00 | Knowing who belongs and who doesn’t
06:13 | Cameras should be able to learn the same
06:22 | Context → patterns → prediction
06:34 | Generations of camera evolution
07:00 | Smart detections: person, car, face, plate
07:14 | More granular detection: clothing, colours, models
07:36 | Natural-language retrieval: next-generation search
07:56 | But still mostly reactive
08:03 | True intelligence: learning the site itself
08:12 | Threat assessment powered by context
08:26 | The massive, untapped value in indexed footage
08:51 | Behaviour understanding vs object detection
09:04 | AI as a security operator/assistant
09:14 | Cameras becoming proactive
09:20 | Future episodes: alarms, sensors, monitoring
09:33 | Industry progress & uneven advancement
09:42 | Why pattern understanding changes everything
09:57 | Closing: A new era is coming
What if you could circle the entire Old City of Jerusalem - comfortably, accessibly, and beautifully? In this episode of Brainstreams, we explore a future-facing vision for a dedicated light rail line that loops around the Old City walls. From smart routing and grade calculations to archaeological sensitivity and pedestrian flow, this Brainstream dives deep into a generative design sprint born from both data and lived experience.
TL;DR:
A Jerusalem city loop light rail around all seven gates could transform access, beauty, and spiritual flow - this is the design pitch.
We discuss multi-line integration (red, brown, yellow), how to preserve archaeology while opening new paths, grading challenges, and why a humble footpath + rail combo might be the most transformative layer Jerusalem’s had in decades.
🎧 Listen now on Spotify, YouTube, or Apple Podcasts
→ More at avrahamraskin.com/podcast
Timestamps:
00:00 | Intro: What is the Old City Loop?
00:43 | Green space, walking path, and rail: the triple-ring vision
01:58 | The gaps in Jerusalem’s current light rail access
03:06 | Seven gates, one loop: walking through the plan
05:01 | Grades, curvature, and optimal timing
06:13 | Integrating with existing and future lines (Red, Yellow, Brown)
07:30 | One-way loop + second-direction bus support
09:23 | Accessibility at Shaar Tzion & other grading concerns
10:00 | Park & Ride potential near Damascus and Dung Gates
12:04 | New Gate tradeoffs & archaeological bridge ideas
13:06 | Solving steep terrain with design sensitivity
14:39 | Closing thoughts: Could this really happen?
What if kosher products simply revealed themselves—- guessing, no flipping, no frantic Googling?
In this Brainstream, we explore a futuristic vision that’s already partially built: an augmented-reality-based kosher shopping experience. Born from the frustration of finding kosher food in Australia, the idea reimagines the process from the ground up-no lists, no scanning, just visual certainty.
TL;DR:
Imagine pointing your phone at a supermarket shelf and instantly seeing which products are kosher-powered by AR, crowdsourcing, and smart databases. This Brainstream lays out that vision.
We also dive into the pain of local-only kosher databases, the promise of contributor modes, and how we can use computer vision, gamification, and community to create the ultimate global kosher assistant.
🎧 Listen now on Spotify, YouTube, or Apple Podcasts
→ More at avrahamraskin.com/podcast
Timestamps:
00:00 — The AR Kosher Shopping Vision
00:49 — Growing Up Without Easy Kosher Access
02:00 — Kosher Australia App: Helpful But Limited
03:04 — The Chocolate Shelf Incident (and UX friction)
04:11 — Barcode Scanning: A Step Forward, Still Painful
05:25 — Every Country Has Its Own App (if you’re lucky)
06:03 — Why Kosher Data Shouldn’t Be Gatekept
07:00 — Introducing Contributor Mode & Crowdsourced Verification
08:13 — How OCR + Computer Vision Enable Instant Recognition
09:23 — Game Mechanics for Certifying Products
10:22 — Verification Badges & Authority Endorsements
12:01 — The Pay-It-Forward Incentive Loop for Travellers
13:03 — Making Kashrut Fun, Seamless, and Fluid
14:01 — The Auto-Scan Dream and Proximity-Based Info
From the publisher's feed