Jun 3, 2026 · 23m · tbpn

Microsoft Takes on Frontier AI with Project Solara, OpenClaw, and More at Build 2026 | Diet TBPN

Martin Scorsese · 26s spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

This episode of Diet TBPN analyzes major announcements from Microsoft Build 2026, focusing on the agentic computing vision of Project Solara, the MAI model suite, and the open-source OpenClaw framework. The hosts also examine broader industry shifts, including autonomous web agents outnumbering human internet traffic, Ramp's autonomous accounting OS, and creative AI tools.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →

The hosts as informed peer 6.0 Guest teaching 1.3 Guest disagreement 1.1 The hosts pushing back 1.5
05100:0010:0020:000:00–4:15 · The hosts as informed peer 5/10 Overview of Microsoft Build 2026 AI Announcements The host provides a broad overview of Microsoft Build announcements including MAI models, OpenClaw, and Project Solara. When the co-host questions the utility of a dedicated hardware badge over a smartphone app, the host honestly concedes the ambiguity before speculating on delegation workflows.4:16–7:18 · The hosts as informed peer 6/10 Ben Thompson's Analysis on Cloud Hubs and Thin Agent Clients The host synthesizes Ben Thompson's analysis on cloud-centric hub-and-spoke agent architecture versus on-device locked-down mobile ecosystems. The co-host interjects playfully with smart cowboy hat ideas, keeping the tone light and non-combative.7:18–9:45 · The hosts as informed peer 6/10 OpenClaw Integration in Scout and Apple Intelligence Comparison The host and co-hosts review Alex Heath's reporting on Microsoft embracing OpenClaw for Scout within enterprise security boundaries. The host draws contrasts with Apple's ecosystem approach ahead of WWDC.9:45–12:11 · The hosts as informed peer 6/10 Datacenter Pushback and Benchmarking MAI Against Meta The co-host educates the host by shifting the benchmark comparison from frontier labs to Meta, pointing out MAI Thinking 1 is competitive with Meta's new models despite having less public hype. The host readily accepts the reframe and expounds on Meta's talent PR strategy.12:11–14:55 · The hosts as informed peer 7/10 Training Data Hygiene, Distillation Ethics, and Enterprise Fine-Tuning The host demonstrates deep technical domain knowledge by detailing training data hygiene, anti-distillation positioning amid recent lawsuits, and contrasting Microsoft's post-training RL with Amazon's mid-training checkpoints.14:55–18:10 · The hosts as informed peer 6/10 Rise of Chinese Consumer Brands: Li-Ning and DJI A lively debate occurs over Chinese consumer brands where the host rejects the panelist's suggestion of Temu and Shein, arguing they lack aspirational brand admiration compared to DJI. The host and co-host also banter combatively over procedural map generation versus legacy gaming in Rust.18:10–20:53 · The hosts as informed peer 6/10 Anduril Racing Sim and Bots Surpassing Human Web Traffic The co-host shares Cloudflare data on bot web traffic surpassing human traffic, which the host immediately contextualizes through agent reasoning traces and tool calls before transitioning to Ramp Stack's accounting automation.20:53–22:44 · The hosts as informed peer 6/10 Martin Scorsese and Black Forest Labs' Cinematic Intelligence The host corrects Jordy's film trivia confusion regarding Martin Scorsese's filmography and evaluates Black Forest Labs' strategic partnership with Scorsese around 'cinematic intelligence'.0:00–4:15 · Guest teaching 1/10 Overview of Microsoft Build 2026 AI Announcements The host provides a broad overview of Microsoft Build announcements including MAI models, OpenClaw, and Project Solara. When the co-host questions the utility of a dedicated hardware badge over a smartphone app, the host honestly concedes the ambiguity before speculating on delegation workflows.4:16–7:18 · Guest teaching 0/10 Ben Thompson's Analysis on Cloud Hubs and Thin Agent Clients The host synthesizes Ben Thompson's analysis on cloud-centric hub-and-spoke agent architecture versus on-device locked-down mobile ecosystems. The co-host interjects playfully with smart cowboy hat ideas, keeping the tone light and non-combative.7:18–9:45 · Guest teaching 1/10 OpenClaw Integration in Scout and Apple Intelligence Comparison The host and co-hosts review Alex Heath's reporting on Microsoft embracing OpenClaw for Scout within enterprise security boundaries. The host draws contrasts with Apple's ecosystem approach ahead of WWDC.9:45–12:11 · Guest teaching 4/10 Datacenter Pushback and Benchmarking MAI Against Meta The co-host educates the host by shifting the benchmark comparison from frontier labs to Meta, pointing out MAI Thinking 1 is competitive with Meta's new models despite having less public hype. The host readily accepts the reframe and expounds on Meta's talent PR strategy.12:11–14:55 · Guest teaching 0/10 Training Data Hygiene, Distillation Ethics, and Enterprise Fine-Tuning The host demonstrates deep technical domain knowledge by detailing training data hygiene, anti-distillation positioning amid recent lawsuits, and contrasting Microsoft's post-training RL with Amazon's mid-training checkpoints.14:55–18:10 · Guest teaching 2/10 Rise of Chinese Consumer Brands: Li-Ning and DJI A lively debate occurs over Chinese consumer brands where the host rejects the panelist's suggestion of Temu and Shein, arguing they lack aspirational brand admiration compared to DJI. The host and co-host also banter combatively over procedural map generation versus legacy gaming in Rust.18:10–20:53 · Guest teaching 2/10 Anduril Racing Sim and Bots Surpassing Human Web Traffic The co-host shares Cloudflare data on bot web traffic surpassing human traffic, which the host immediately contextualizes through agent reasoning traces and tool calls before transitioning to Ramp Stack's accounting automation.20:53–22:44 · Guest teaching 0/10 Martin Scorsese and Black Forest Labs' Cinematic Intelligence The host corrects Jordy's film trivia confusion regarding Martin Scorsese's filmography and evaluates Black Forest Labs' strategic partnership with Scorsese around 'cinematic intelligence'.0:00–4:15 · Guest disagreement 1/10 Overview of Microsoft Build 2026 AI Announcements The host provides a broad overview of Microsoft Build announcements including MAI models, OpenClaw, and Project Solara. When the co-host questions the utility of a dedicated hardware badge over a smartphone app, the host honestly concedes the ambiguity before speculating on delegation workflows.4:16–7:18 · Guest disagreement 0/10 Ben Thompson's Analysis on Cloud Hubs and Thin Agent Clients The host synthesizes Ben Thompson's analysis on cloud-centric hub-and-spoke agent architecture versus on-device locked-down mobile ecosystems. The co-host interjects playfully with smart cowboy hat ideas, keeping the tone light and non-combative.7:18–9:45 · Guest disagreement 1/10 OpenClaw Integration in Scout and Apple Intelligence Comparison The host and co-hosts review Alex Heath's reporting on Microsoft embracing OpenClaw for Scout within enterprise security boundaries. The host draws contrasts with Apple's ecosystem approach ahead of WWDC.9:45–12:11 · Guest disagreement 2/10 Datacenter Pushback and Benchmarking MAI Against Meta The co-host educates the host by shifting the benchmark comparison from frontier labs to Meta, pointing out MAI Thinking 1 is competitive with Meta's new models despite having less public hype. The host readily accepts the reframe and expounds on Meta's talent PR strategy.12:11–14:55 · Guest disagreement 0/10 Training Data Hygiene, Distillation Ethics, and Enterprise Fine-Tuning The host demonstrates deep technical domain knowledge by detailing training data hygiene, anti-distillation positioning amid recent lawsuits, and contrasting Microsoft's post-training RL with Amazon's mid-training checkpoints.14:55–18:10 · Guest disagreement 3/10 Rise of Chinese Consumer Brands: Li-Ning and DJI A lively debate occurs over Chinese consumer brands where the host rejects the panelist's suggestion of Temu and Shein, arguing they lack aspirational brand admiration compared to DJI. The host and co-host also banter combatively over procedural map generation versus legacy gaming in Rust.18:10–20:53 · Guest disagreement 1/10 Anduril Racing Sim and Bots Surpassing Human Web Traffic The co-host shares Cloudflare data on bot web traffic surpassing human traffic, which the host immediately contextualizes through agent reasoning traces and tool calls before transitioning to Ramp Stack's accounting automation.20:53–22:44 · Guest disagreement 1/10 Martin Scorsese and Black Forest Labs' Cinematic Intelligence The host corrects Jordy's film trivia confusion regarding Martin Scorsese's filmography and evaluates Black Forest Labs' strategic partnership with Scorsese around 'cinematic intelligence'.0:00–4:15 · The hosts pushing back 1/10 Overview of Microsoft Build 2026 AI Announcements The host provides a broad overview of Microsoft Build announcements including MAI models, OpenClaw, and Project Solara. When the co-host questions the utility of a dedicated hardware badge over a smartphone app, the host honestly concedes the ambiguity before speculating on delegation workflows.4:16–7:18 · The hosts pushing back 1/10 Ben Thompson's Analysis on Cloud Hubs and Thin Agent Clients The host synthesizes Ben Thompson's analysis on cloud-centric hub-and-spoke agent architecture versus on-device locked-down mobile ecosystems. The co-host interjects playfully with smart cowboy hat ideas, keeping the tone light and non-combative.7:18–9:45 · The hosts pushing back 1/10 OpenClaw Integration in Scout and Apple Intelligence Comparison The host and co-hosts review Alex Heath's reporting on Microsoft embracing OpenClaw for Scout within enterprise security boundaries. The host draws contrasts with Apple's ecosystem approach ahead of WWDC.9:45–12:11 · The hosts pushing back 2/10 Datacenter Pushback and Benchmarking MAI Against Meta The co-host educates the host by shifting the benchmark comparison from frontier labs to Meta, pointing out MAI Thinking 1 is competitive with Meta's new models despite having less public hype. The host readily accepts the reframe and expounds on Meta's talent PR strategy.12:11–14:55 · The hosts pushing back 0/10 Training Data Hygiene, Distillation Ethics, and Enterprise Fine-Tuning The host demonstrates deep technical domain knowledge by detailing training data hygiene, anti-distillation positioning amid recent lawsuits, and contrasting Microsoft's post-training RL with Amazon's mid-training checkpoints.14:55–18:10 · The hosts pushing back 4/10 Rise of Chinese Consumer Brands: Li-Ning and DJI A lively debate occurs over Chinese consumer brands where the host rejects the panelist's suggestion of Temu and Shein, arguing they lack aspirational brand admiration compared to DJI. The host and co-host also banter combatively over procedural map generation versus legacy gaming in Rust.18:10–20:53 · The hosts pushing back 1/10 Anduril Racing Sim and Bots Surpassing Human Web Traffic The co-host shares Cloudflare data on bot web traffic surpassing human traffic, which the host immediately contextualizes through agent reasoning traces and tool calls before transitioning to Ramp Stack's accounting automation.20:53–22:44 · The hosts pushing back 2/10 Martin Scorsese and Black Forest Labs' Cinematic Intelligence The host corrects Jordy's film trivia confusion regarding Martin Scorsese's filmography and evaluates Black Forest Labs' strategic partnership with Scorsese around 'cinematic intelligence'.

speaking balance: gold is the hosts, purple is the guest (3 minute bins)

0:00 · the hosts 0% · guest 100%0:00 · the hosts 0% · guest 100%3:00 · the hosts 0% · guest 100%3:00 · the hosts 0% · guest 100%6:00 · the hosts 0% · guest 100%6:00 · the hosts 0% · guest 100%9:00 · the hosts 0% · guest 100%9:00 · the hosts 0% · guest 100%12:00 · the hosts 0% · guest 100%12:00 · the hosts 0% · guest 100%15:00 · the hosts 0% · guest 100%15:00 · the hosts 0% · guest 100%18:00 · the hosts 0% · guest 100%18:00 · the hosts 0% · guest 100%21:00 · the hosts 0% · guest 100%21:00 · the hosts 0% · guest 100%
Sharpest disagreement ▶ 17:48 Pushback on generative map design in favor of Rust

The co-host rejects the host's enthusiasm for procedurally remixed map layouts, holding firm that classic static maps like in Rust are vastly superior.

Hardest push from the hosts ▶ 16:27 Host pushes back on Temu and Shein comparison

The host refuses the co-host's suggestion that Temu and Shein represent admired Chinese brands in the US, distinguishing cheap volume from genuine brand affinity like DJI.

Biggest teaching moment ▶ 10:36 Panelist reframes MAI benchmark perspective to Meta

The co-host informs the host that Microsoft's MAI Thinking 1 is remarkably competitive with Meta's Spark rather than lagging behind older frontier models.

The host holds their own ▶ 13:00 Host delivers detailed technical breakdown of model distillation and fine-tuning

The host commands the conversation by distinguishing Microsoft's clean data pre-training and RL fine-tuning from Amazon's mid-training checkpoints and industry distillation controversies.

the scores for every segment, with the reasoning behind each
ChapterTopicThe hosts as informed peerGuest teachingGuest disagreementThe hosts pushing backWhy
Overview of Microsoft Build 2026 AI Announcements 5111 The host provides a broad overview of Microsoft Build announcements including MAI models, OpenClaw, and Project Solara. When the co-host questions the utility of a dedicated hardware badge over a smartphone app, the host honestly concedes the ambiguity before speculating on delegation workflows.
Ben Thompson's Analysis on Cloud Hubs and Thin Agent Clients 6001 The host synthesizes Ben Thompson's analysis on cloud-centric hub-and-spoke agent architecture versus on-device locked-down mobile ecosystems. The co-host interjects playfully with smart cowboy hat ideas, keeping the tone light and non-combative.
OpenClaw Integration in Scout and Apple Intelligence Comparison 6111 The host and co-hosts review Alex Heath's reporting on Microsoft embracing OpenClaw for Scout within enterprise security boundaries. The host draws contrasts with Apple's ecosystem approach ahead of WWDC.
Datacenter Pushback and Benchmarking MAI Against Meta 6422 The co-host educates the host by shifting the benchmark comparison from frontier labs to Meta, pointing out MAI Thinking 1 is competitive with Meta's new models despite having less public hype. The host readily accepts the reframe and expounds on Meta's talent PR strategy.
Training Data Hygiene, Distillation Ethics, and Enterprise Fine-Tuning 7000 The host demonstrates deep technical domain knowledge by detailing training data hygiene, anti-distillation positioning amid recent lawsuits, and contrasting Microsoft's post-training RL with Amazon's mid-training checkpoints.
Rise of Chinese Consumer Brands: Li-Ning and DJI 6234 A lively debate occurs over Chinese consumer brands where the host rejects the panelist's suggestion of Temu and Shein, arguing they lack aspirational brand admiration compared to DJI. The host and co-host also banter combatively over procedural map generation versus legacy gaming in Rust.
Anduril Racing Sim and Bots Surpassing Human Web Traffic 6211 The co-host shares Cloudflare data on bot web traffic surpassing human traffic, which the host immediately contextualizes through agent reasoning traces and tool calls before transitioning to Ramp Stack's accounting automation.
Martin Scorsese and Black Forest Labs' Cinematic Intelligence 6012 The host corrects Jordy's film trivia confusion regarding Martin Scorsese's filmography and evaluates Black Forest Labs' strategic partnership with Scorsese around 'cinematic intelligence'.

Statements from this episode (1)

Insight
Martin Scorsese says generative AI parallels DeMille's pre-production oil paintings
“DeMille would have his production designers do oil paintings. This is that, in a sense, conveys a cinematic intelligence.”
Martin Scorsese Jun 3, 2026 ▶ 21:58
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 500 episodes transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.