Ryan Carson (ex-Treehouse, Intel; now Builder-in-Residence at Sourcegraph’s AMP) shares his origin story and a practical playbook for shipping software with AI agents. We cover why “tokens aren’t cheap,” how AMP made pro-level coding free via developer ads, a concrete workflow (PRD → atomic dev tasks → agent execution with self-tests), and why managers should spend time as ICs “managing AI.” We close with advice for raising AI-native kids and a perspective on this moment in tech (think integrated circuit–level shift).Timestamps

00:00 – The beginning of intelligence: how LLMs changed Ryan’s view of computing

00:23 – Apple IIe → Turbo Pascal → Computer Science: the maker bug bites

03:20 – DropSend: early SaaS, Dropbox name clash, first acquisition

04:30 – Treehouse: teaching coding without a CS degree; $20M raised, acquired in 2021

05:02 – The “bigger than a computer” moment: discovering LLMs

06:15 – Joining Intel: learning GPUs and the scale of silicon (“my adult internship”)

07:09 – Building an AI divorce assistant → joining AMP as Builder-in-Residence

09:38 – AMP vs ChatGPT/Claude/Cursor: agentic coding with contextual developer ads

11:09 – Token economics: why AI isn’t really cheap

17:27 – Frontier vs Flash models (Sonnet 4.5 vs Gemini 2.5) — how costs scale

21:31 – Private startup: vertical AI for specialized domains

22:36 – The new wave of small, vertical AI businesses

23:01 – Live demo: building a news app end-to-end with AMP

28:18 – How to plan like a pro: write the PRD before you build

30:02 – “Outsource the work, not your thinking.”

32:28 – Turning PRDs into atomic tasks (1.0, 1.1…)

35:50 – Competing in an AI world = planning well

36:28 – Managers should schedule IC time to “manage AI”

37:14 – Designing feedback loops so agents can test themselves

39:47 – “AI lied to me”: why verifiable tests matter

41:11 – Raising AI-native kids: build trust, context, and agency

43:59 – “We’re living in the integrated circuit moment of intelligence.”Tools & Technologies MentionedAMP (Sourcegraph) – Agentic coding tool/IDE copilot that plans, edits, and ships code. Now offers a high-end, ad-supported free tier; ads are contextual for developers and don’t influence code outputs.Sourcegraph (Code Search) – Parent company; enterprise code intelligence/search.ChatGPT / Claude – General-purpose LLM assistants commonly used alongside coding agents.Cursor / Windsurf – AI-first code editors that integrate LLMs for completion and refactors.Bolt / Lovable – Text-to-app builders for rapid prototyping from prompts.WhisperFlow / SuperWhisper – Voice-to-text tools for fast prompting and dictation.Anthropic Sonnet 4.5 – Frontier-grade reasoning/coding model; powerful but pricier per token.Google Gemini 2.5 Flash – Fast, lower-cost model; “good enough” for many workloads.Auth0 (example) – Authentication-as-a-service mentioned as a contextual ad use case.GPUs / TPUs – Compute for training/inference; token cost drivers behind AI pricing.PRD + Atomic Tasks Workflow – Ryan’s method: record spec → generate PRD → expand to dot-notated tasks → let the agent implement.Self-testing Scripts – Ask agents to generate runnable tests/health checks and loop until passing to reduce back-and-forth and prevent “it passed” hallucinations.Family ChatGPT Accounts – Tip for raising AI-native kids; teach sourcing, context, and trust calibration.Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

Aydin sits down with Filip Skrzesinski, co-founder of Subframe, to unpack how AI and code-native design tools are collapsing the classic PM → design → engineering handoffs. Filip explains why “pictures to code” is an unfair ask of engineers, shows how Subframe lets teams design directly in the same material as production code, and demos building a Fellow feature—from screenshot → design system match → working prototype—without access to Fellow’s codebase. They close on what’s next: organizations training their own “house models” to reflect product taste, patterns, and constraints so more people across the company can truly build.Key takeawaysDesign in the same material as code: Subframe treats UI work as editable code, eliminating fidelity loss from design handoffs.Fewer stages, faster loops: PMs, designers, and engineers collaborate in one artifact; prototypes look and behave like the real app.AI as a trained teammate, not a slot machine: Teams will shape models with system prompts, snippets, and feedback—like mentoring a junior designer.Front-end ownership shifts: Designers can own front-end structure and components; engineers wire up backends and complex logic.Prototype to PRD: High-fidelity prototypes beat docs for alignment, user testing, and speed.Timestamps00:00 - Introduction 01:00 Fil's path: audio engineering → CS → design → startup co-founder03:48 Builders everywhere: from Dreamweaver → Webflow → Shopify → now “apps”04:01 What Subframe is: a design tool rooted in code05:48 Bridging LLMs (great at code) with visual design context08:09 The architect vs. printer analogy for product design12:23 Back to the show: “The new way” is collapsing steps and handoffs14:07 “Five-year” vision (sooner than you think): design → code with agents in the loop16:31 Training models on your org’s taste: like raising a puppy—examples & theory19:15 Today’s demo plan: build a Fellow feature in Subframe without codebase access21:04 Recreating Fellow’s UI: import colors/typography; screenshot → layout23:07 Don’t fight the AI: let it rough-in, then designers perfect in visual mode24:11 Why prototypes should look native (not “off-brand” sandboxes)26:07 Syncing components to codebases; where Subframe stops (front-end) and engineers continue (backend)28:33 Programmatic (deterministic) UI code & generative for visuals30:00 PMs in the tool: prompt to add a Share dialog with transcript and video context35:08 Exploring multiple design variations; mix-and-match patterns (“snippets”)37:57 From design to interactive prototype via annotations (“do this on click…”)45:22 First build runs: working Share flow; alert updates after sending47:02 Export code → Cursor/GitHub; hand off real components48:08 The next 12 months: more ideas shipped, more makers, less gatekeepingTools & technologies MentionedSubframe — Code-native design tool for building UI/UX; designs directly edit the underlying code; syncs components to your repo.Fellow.ai — AI meeting assistant with privacy controls; accurate summaries, actions, decisions; broad SaaS integrations.Cursor — AI-assisted code editor; good for continuing from exported Subframe code to production.GitHub — Repo hosting and collaboration for shipping the generated/edited UI code.AI code agents — Used by engineers to wire front-end to backend services and data.Squarespace / Webflow / Dreamweaver — Prior waves that democratized web creation; backdrop for today’s “apps layer.”Shopify — Example of no-code/low-code e-commerce; analogy for app building’s democratization.Lovable / Bolts / V0 — AI code/prototyping tools referenced as peers for generating working app scaffolds.Slack / Asana / HubSpot / Salesforce / Linear / Jira / Confluence — Systems Fellow integrates with to push notes, actions, and records.Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

JD Fiscus (nerding.io) shares how a late-night hack connecting MCP to n8n exploded to ~1M downloads, then demos practical MCP workflows: indexing YouTube channels for Q&A, and auto-building n8n flows from natural language. We dig into the Agentic Commerce Protocol, real security pitfalls (like destructive commands), and how to turn MCPs into products with OAuth and Stripe for authentication and metered billing. He closes with how he teaches this hands-on at the Vibe Coding Retreat.Timestamps1:00 Why build it: “MCP shouldn’t be Claude-only”—bridging MCP into n8n early (Dec/Jan)2:09 Shipping under the pseudonym nerding.io; surprise seeing creators use it2:25 n8n later ships its own MCP server/client; they nod to nerding.io & Simon3:59 “N8n is useful, but so much more useful with MCP”5:12 What MCP means for software: every smart company is exposing an MCP; new login/usage patterns6:27 Agentic Commerce Protocol (ACP): Stripe + OpenAI; agents checkout across the web8:02 Marketing to agents not humans? SEO shifts as agents comparison-shop9:10 Early “agent mode” attempts vs protocol-based purchases (less hacky)10:58 Likely adopters: platforms (Shopify) & big retailers; echoes of early MCP evolution14:11 Security realities: token passing evolved to OAuth; hallucination + destructive actions risk16:04 Personal mishap: agent ran supabase reset on a dev DB—imagine prod! Guardrails matter17:03 Designing MCP servers: don’t just “wrap your API”; use resources/prompts for agentic UX19:04 Demo 1—Influencer MCP: index a YouTube channel, embed transcripts, ask questions in Claude20:54 Storage: embeddings into Postgres; per-channel tables24:46 Keeping it fresh: daily cron to ingest new videos25:18 Demo 2—Build n8n workflows from chat using N8N MCP (by Ramullet); live docs + API27:00 “Create a webhook → send leads to Sheets” built conversationally, with allow/deny prompts31:02 Zapier, Gumloop: agents that build automations via natural-language steps34:00 Next frontier: custom connectors (Claude/Cursor/OpenAI), OAuth auth flows for MCPs39:03 Turning MCPs into products: login with Twitter → Stripe subscription → metered billing41:12 Paid tool call demo: “paid echo” → Stripe usage event logged per user43:41 How to learn this fast: vibecodingretreat.com (small cohorts, hands-on builds)Tools & Technologies Mentioned (quick guide)MCP (Model Context Protocol) — Standard for connecting models to tools/data; supports tools, resources, prompts.n8n — Open-source automation platform; JD wrote an MCP node that went viral; also has native MCP server/client now.Claude / Cursor / OpenAI (custom connectors) — LLM IDEs/chats that can load MCPs; custom connectors enable OAuth + productized access.Agentic Commerce Protocol (ACP) — Early protocol (Stripe + OpenAI) for agent-initiated purchases with confirmations.Web MCP (W3C-oriented idea) — Emerging patterns for agent↔︎website interactions beyond human UI flows.OAuth — Secure, user-consented authentication for MCPs (vs passing raw tokens).Stripe (subscriptions + metered billing) — Attach billing/usage limits to MCP calls; track per-user consumption.YouTube API + Transcripts — Source data for the “Influencer MCP” indexing pipeline.Embeddings + Postgres — Store vectorized transcript chunks in Postgres for retrieval (JD self-hosts).Cron — Schedules daily ingestion of new content.Google Sheets — Target destination in demo for simple lead funnels.Zapier / Gumloop — Natural-language automation builders; early NLA/agent patterns.Git / CLI commands — Cautionary tale: agents running destructive commands (e.g., resets).Do Browser / Comet Browser — Agentic browsing tools referenced for web actions.Fellow.ai — AI meeting assistant with security-first design; generates precise summaries/action items.Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

Aydin and Kieran Klaassen (Cora) unpack Compound Engineering—treating every task as an investment so the next time is faster. Kieran shares his path from film composer to startup CTO and live-demos how he plans → prototypes → ships a feature using AI agents (Claude Code), then runs multi-agent reviews. They discuss why managers are primed to orchestrate agents, how to capture your own feedback patterns, and why there’s “no excuse not to have a prototype” anymore.Timestamps0:07 — “Every piece of work should be an investment.”2:15 — What Cora is: an AI Gmail layer that auto-archives ~80% and briefs you twice daily.3:32 — Launch notes & early user reactions.5:21 — The Claude Code pricing saga and “finding the limits.”8:06 — Compound Engineering defined (codify how you work so AI does it next time).15:01 — From “automation” to pattern-capturing systems; natural-language rules over brittle workflows. 22:03 — Demo kickoff: planning the “Invite friends” improvement inside Cora.26:11 — Rapid mockups from a screenshot + voice description; iterate in seconds.33:06 — Multi-agent planning: repo research, best-practices scout, framework researcher.41:01 — Human judgment on plans; simplify when encryption/perf add hidden complexity.50:00 — Feature running end-to-end; agentic PR + test flow; sub-agent code reviews.Tools & Technologies MentionedCora — AI inbox copilot for Gmail that prioritizes, summarizes, and drafts replies; batches the rest into twice-daily briefs.Claude Code (Anthropic) — Agentic coding/terminal assistant used for planning, building, and reviews.Monologue — Voice-to-text for quickly describing UI and generating mockups.Every.to — Partner/design/content hub Kieran collaborates with; also publishes his writing on Compound Engineering.GitHub + GitHub CLI — Issues, branches, PRs automated by agents from plan → code → review.VS Code (with Claude Code extension) — IDE setup for hands-on edits when needed.Anthropic Console Prompt Generator — Used to scaffold robust prompts/agents, then refined manually.Model mix for reviews (e.g., “GPT-5 Codecs,” “Claude Opus”) — Alternative model passes for plan/code critique.Fellow.ai — Aydin’s AI meeting assistant for accurate notes, actions, and privacy-aware summaries.Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

Content marketer and video lead Emily Kensley (Fellow) walks through a near-zero-friction workflow for creating polished product videos fast. She records clean, auto-animated screen demos with Screen Studio, fixes (or replaces) audio with Podcastle (Magic Dust + AI voices), and drafts scripts by riffing into a Fellow meeting then refining the transcript in ChatGPT. The result: 11-minute, brand-consistent tutorials produced in hours instead of days—repeatable by any team (marketing, CS, product, sales).Timestamps01:19 — Daily use of AI; from occasional to constant over last 6 months01:53 — What you’ll learn: a minimal-human, video-centric content workflow03:41 — Tool #1 intro: Screen Studio for screen recordings05:27 — Live capture of an AI meeting recap demo (click-through highlights, actions, decisions)06:23 — Raw → instant output: auto-smoothing cursor paths & smart zooms (no manual keyframes)07:23 — Host example: using Screen Studio for a Zapier + Fellow automation video07:44 — “Done is better than perfect”: quick crop fixes, branded backgrounds, cursor presets08:24 — Team presets = consistent brand across departments09:44 — Tool #2 intro & story: Podcastle rescues a day of bad mic audio10:59 — Podcastle audio editor: noise reduction, levelling, silence removal12:10 — Magic Dust AI demo: echoey room → studio-quality voice13:38 — AI Voices in Podcastle: when to clone vs. pick a preset (e.g., “Abigail”)16:12 — Long-form scripts → generated narration in minutes; edit/regenerate on typos17:54 — Brand consistency: shared voice so any team can ship VO18:29 — Putting it together: Screen Studio video + Podcastle narration19:24 — Finished example: Fellow YouTube settings walkthrough (11-minute tutorial)21:06 — Syncing visuals to VO: record screen while listening to the generated narration22:59 — Script creation workflow: Fellow call → transcript → ChatGPT → clean script23:34 — Full recap of the end-to-end pipeline25:01 — Repurposing: scripts → blogs, help center, CS clips; scale breadth of tutorials26:28 — Looking ahead: excitement about fast-evolving AI agentsTools & Technologies Mentioned (with quick notes)Screen Studio — Smart screen recorder that auto-smooths mouse movement, adds tasteful zoom/pan animations, and supports brand presets for consistent output.Podcastle — Audio suite used here to edit audio clipsMagic Dust AI: one-click studio-quality enhancement (denoise, de-reverb, leveling).AI Voices & Voice Cloning: generate narration from text; keep brand-consistent VO.Fellow — AI Meeting assistant used to host a solo “idea dump,” generate transcripts, AI recaps, chapters, and action items; doubles as the seed for scripts.ChatGPT — Refines raw Fellow transcript into a clean, concise voiceover script.YouTube — Publishing destination for finished tutorials.Zapier — Example in host’s Screen Studio demo (automation with Fellow).Google Meet / Zoom — Where the solo Fellow “recording” session can happen.Adobe (Premiere/After Effects) — Old manual workflow stand-ins (contrast to auto animations).Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

Eddie Yoon, Sr Director, Paid Media at NP Digital, shows how CMOs can spin up a full creative campaign in ~30 minutes using AI. He breaks down a rapid “three-tab” workflow—Meta Ad Library for competitive research, GPT for strategy and prompts, and an image generator (Reeve) for instant mood boards—then extends it into testing (Trial Reels, TikTok hooks), product R&D, and agentic pipelines. We also riff on why the next decade could normalize solo billionaire founders, how Netflix foreshadowed AI-driven content, and what real-time, stylized, monetizable media will look like.

Timestamps

1:07 Meet Eddie Yoon—NP Digital, paid social × creative × AI background.

1:49 “AI is redefining growth”: blistering company speed and scale.

2:16 The solo-founder era & agentic executive teams.

4:39 Enterprise example: HubSpot’s leadership going all-in on AI.

5:29 Founder example: Tyler at Beehive—shipping fast by listening + acting.

6:30 Design & media: Netflix’s early AI play; House of Cards data story.

11:29 The 30-minute campaign challenge—Eddie’s live plan.

12:53 The three tabs: Meta Ad Library → GPT prompts → Reeve mockups.

14:37 Copy/paste every active ad into GPT; ask for strategy synthesis.

16:06 Five “board-level” ideas; forcing a single high-acceptance pitch.

17:56 Image prompt for “Comfort 2.0” (eco-luxury, performance lifestyle).

20:27 Prompting hack: “200+ IQ” to push for originality (avoid clichés).

21:06 Locking on Comfort 2.0—“performance tech meets everyday life.”

23:06 Iterating the mood board; feeding outputs back into GPT.

23:30 If the client has the shoe already: do it all in AI (no photoshoot).

24:39 Rapid tests: ethnicity, angle, color; Instagram Trial Reels.

26:03 Beyond ads: full-funnel → product design & R&D with agents.

27:24 100-page competitor deep dives from public signals.

28:26 Scoring system (cutoff 85; 95+ are “winners”) to prioritize assets.

30:13 Spinning GPT outputs into 10 TikTok hooks for creators/founders.

31:32 Domain-tuned agents that deliver 90%-ready work.

33:13 What’s next: automatic video analysis and creative fixes.

34:13 Next 12 months: IP-driven brands, real-time stylized video, avatars.

35:43 Meta: capturing AI audio; partner via your agent in the future.

36:12 Why solo $1B is realistic (and $100M solos even more so).

Tools & Technologies Mentioned (with quick notes)

Meta Ad Library — Public index of active FB/IG ads; great for competitive creative research.

GPT — Used to analyze competitor ads, generate board-level strategies, image prompts, TikTok hooks, and run scoring frameworks.

Reeve — Static image generator (Midjourney-like) for fast mood boards and spec creative.

Midjourney — Alternative image generation tool for photorealistic concepts.

VO3 — Motion/video generation tool referenced for animated concepts.

Instagram Trial Reels — Organic test surface to gauge hooks/creatives with cold audiences before spend.

TikTok — Distribution + hook testing via short scripts for creators/founders.

Semrush — Search/keyword intel to complement social competitive analysis.

SocialPeta — Creative/spend intelligence (legacy use; less relied upon now).

AI Avatars & Agentic Flows — Persona-based creators and multi-agent pipelines to speed research, ideation, testing, and post-mortems.

Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

In this episode, Alexandra Sunderland (VP of Engineering at Fellow) pulls back the curtain on how she runs engineering with agentic workflows that actually move the needle: background coding agents in Cursor that fix bugs while she’s in meetings, Claude + MCPs to query Linear and auto-generate reports in seconds, and Zapier pipelines that turn meeting transcripts into daily briefs, real-time project risk pings, sales insights, and even 1:1 growth trackers. The theme: make conversations computable, specialize agents narrowly, and wire every tool together so ops happen while you sleep.Timestamps1:11 — Background: 13+ yrs with Aydin; author of Remote Engineering Management.2:13 — What is an “agent”? Alexandra’s practical definition (automation + LLM).3:39 — Why specialized agents beat general ones (Sept 2025 reality check).5:25 — Cursor background agents via Slack VIP notifications—coding while she’s away.8:00 — Hackathon: hand-built dev productivity dashboard vs. Claude + Linear MCP.10:38 — Why use Claude here instead of Cursor: downloadable PDFs & exploratory insights.13:03 — Interface shift: logging into Linear/GitHub less; notify via Slack instead.14:21 — Plan: live workflows that leaders can copy.15:31 — Workflow #1: Daily Brief in Zapier (9:00 a.m. trigger → transcripts → CoS-style digest).18:00 — Slack example of the generated daily brief.20:22 — Workflow #2: Project Meeting Insights—real-time blockers & cross-team risks.22:00 — Prompting style (“best VP of Eng in the world”) and why it helps.26:40 — Idea: an “Alexandra agent” that drafts her responses.27:59 — Workflow #3: Sales call mining → bug/feature requests for Eng.29:14 — Next step: Cursor agents created via API—fixes ready for human review minutes after calls.30:23 — Rolling Cursor to product & success; non-engineers leverage code context.31:16 — Auto-drafting help center docs with Cursor that can browse.32:34 — Future: docs auto-update—or vanish into on-demand LLM answers.34:52 — Workflow #4 (WIP): 1:1 growth tracker—extract coaching, strengths, feedback into a living doc.37:41 — Sales coaching automation: enforce key phrases/objection handling.38:10 — Playbook: start with simple “yesterday’s conversations → insights,” then stack.39:24 — Next 12 months: tools connecting to each other, patterns across datasets.Tools & Technologies Mentioned (with quick notes)Cursor — AI-powered code editor with background agents (cloud-run) and Slack integration for async coding and fixes.Cursor Background Agents API — Programmatically spin up agents to implement bug fixes/features for later human review.Slack (VIP Notifications) — Marking the Cursor app as VIP ensures agent updates punch through Do Not Disturb.Claude — LLM used with MCPs to query data sources (e.g., Linear), generate PDFs, surface trends, and build ad-hoc reports.MCP (Model Context Protocol) — Standard to connect LLMs to tools/data (e.g., Linear) for live, permissioned operations.Linear — Issue/project tracker; source for ticket analytics (resolution rates, triage time, stage durations).Zapier — No-code automations; schedules, filters, formats, makes API calls, and runs AI by Zapier LLM steps.Fellow.ai — AI meeting assistant capturing summaries, actions, decisions; acts as an “AI chief of staff” across meetings.GitHub — Code hosting referenced as a UI Alexandra now visits less thanks to agentic workflows.Google Docs / Notion / Wiki — Destinations for auto-appending 1:1 growth notes and team principles.APIs (custom + vendor) — Zapier “Webhooks by Zapier”/custom API calls used to fetch transcripts and trigger agents.Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

In this episode, Aydin chats with brothers Emil and Cassy—founders behind Hoppier (snack stipends for teams) and Postbeam (an AI-native LinkedIn content engine). They show how transcripts, voice interfaces, and AI browsers can 10× content output and product velocity for small teams. Demos include: turning transcripts into LinkedIn posts, Postbeam’s “Marv” voice interview, Vercel v0 mockups, and Perplexity’s Comet browser agent. The theme: tiny teams, mighty outcomes—when AI is baked into every workflow.

Timeline & Timestamps

01:08 – Hoppier origin: ~1,200 customers, profitable, still founder-run.

03:57 – Why transcripts are gold for creating unlimited content.

05:06 – Demo: pulling a podcast transcript into Claude → strong LinkedIn post hooks.

08:55 – Volume matters: consistency wins; learning from creators like Pablo.

11:15 – Remix vs. original insights: two formulas for content that works.

14:38 – From process to product: Postbeam lands early paying customers.

16:26 – Inside Postbeam: sources, remixing, images, and multi-team member voices.

18:33 – Demo: Marv voice feature interviews you to capture authentic tone.

24:21 – Building with AI: using Vercel v0 for rapid UI mockups and team feedback.

29:16 – Aydin’s day job plug: Fellow.ai meeting assistant.

31:36 – Replit vs. V0 vs. Lovable: pros, cons, and caution for prod-grade apps.

35:58 – Comet browser demo: finding Toyota RAV4s on Marketplace with AI.

42:42 – Tiny but mighty: Postbeam (2 founders + Gen Z cousin) and Hoppier (7-figure biz with 4 ppl).

Tools & Technologies Mentioned

Claude (Anthropic) — Generates LinkedIn posts from transcripts.

ElevenLabs / YouTube Transcript Tools — For pulling transcripts.

Postbeam — AI LinkedIn content engine.

Marv (inside Postbeam) — Voice interview AI to capture tone.

Vercel v0 — Natural language → React UI mockups.

Replit / Lovable / Cursor — AI coding platforms, with tradeoffs.

Perplexity’s Comet Browser — Agentic browser for automated browsing.

Whisper Flow — Voice-first workflow automation.

Fellow.ai — AI meeting assistant.

Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

Host Aydin Mirzaee welcomes Marquis Murray, productivity consultant and YouTuber, to share how AI agents are transforming day-to-day work. Marquis walks through live demos connecting Claude to Asana via MCP to auto-build projects from transcripts, generate executive-ready status PDFs, and create “AI teammates” in Asana that triage requests, draft briefs, write emails/blogs, and route approvals—keeping humans in the loop. Once you see it, you won’t go back to manual setup.

Timestamps

0:06 – Why manual project planning is over; AI compresses weeks into minutes.

0:17 – Introducing Marquis Murray.

0:43 – Starting the YouTube channel during lockdowns; documenting Asana learnings.

2:06 – From corporate to consulting; helping teams adopt Asana, HubSpot, Zoom, Slack.

4:05 – Making companies more productive with AI and integrations.

4:53 – Today’s plan: Claude + Asana + agents.

6:06 – Using Claude as a “central AI” via MCP.

8:17 – Building a Customer Appreciation Event project in Asana directly from Claude.

12:20 – Custom fields/sections: what connectors can and can’t create.

13:06 – Finished example: phases, tasks, owners, dates.

14:05 – Feeding transcripts and docs to generate realistic demo projects.

19:05 – “If you’re not doing this yet, start today.”

19:42 – Pulling Asana status into Claude and exporting a polished PDF.

23:34 – Exec-friendly reports: progress bars, metrics, priorities.

24:50 – Asana AI Studio: agents as virtual teammates.

27:23 – Auto-correcting human errors: naming, missing info, duplicates.

29:02 – Agents rename tasks, create briefs, draft assets.

35:42 – Agents gatekeep incomplete requests; ask for specifics.

37:13 – AI-generated campaign brief, email, and blog drafts.

39:08 – Human-in-the-loop approvals before going live.

43:01 – Triage demo: vague video request → structured follow-ups.

45:25 – Auto-created subtasks to collect missing details.

46:33 – “Easy mode” for building agents with natural language.

47:03 – Marquis’s wish: a true AI chief of staff that restructures your day.

48:56 – Where to find Marquis’s tutorials; wrap-up.

Tools & Technologies Mentioned

Asana — Project management platform; AI Studio builds rule/LLM agents (“teammates”).

Claude (Anthropic) — AI assistant used for brainstorming, MCP connections, summaries.

Perplexity — AI search and research assistant.

HubSpot / Salesforce / Jira — CRM/dev tools commonly integrated with Asana workflows.

Zoom & Slack — Core collaboration stack surfaced during remote shift.

MCP (Model Context Protocol) — Lets LLMs securely interact with external tools like Asana.

Fellow.ai — AI meeting assistant for accurate summaries, action items, and insights.

Google Drive, Gmail, Calendar, Canva — Connected apps Claude can use to orchestrate work.

Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

In this episode of This New Way, Aydin sits down with Ali Pourshahid, Chief Engineering Officer at Solace, to explore how he’s woven AI into his daily routines as a technology executive. Ali shares real demos of how he uses Solace’s Agent Mesh and other AI tools to:

Generate security-focused slide decks from Confluence in minutes

Schedule a 12-person leadership offsite without lifting a finger

Transform messy Word reports into polished heat maps for government updates

Automate customer support workflows across Jira, CRM, and internal systems

Ali also breaks down how Solace is productizing their internal AI system, why agent-to-agent communication (A2A) is critical, and how to build a culture of experimentation with “AI champions” inside your company.

This is a masterclass in how executives can stop just “keeping up” with AI—and instead lead the charge.

Timestamps

0:58 – What is Solace and Ali’s role as Head of Engineering

2:09 – Ali’s daily “AI deep hour” and why he treats it like a workout

3:19 – Prepping for a customer security call with AI + Confluence

5:14 – Auto-generated sequence diagrams and value slides in minutes

9:48 – Using Microsoft Copilot to instantly format professional slides

13:35 – AI as an executive assistant: scheduling a 12-person workshop

17:01 – Turning unstructured Word reports into project heat maps

20:07 – Building an AI champions group and lightning talks at Solace

25:05 – Solace Agent Mesh: event-driven architecture for agents

29:01 – Live demo: Automating Jira support tickets with agent workflows

33:33 – Scaling digital employees with orchestrator agents

37:08 – Why evals are critical for testing and deploying AI agents

39:05 – Ali’s habit: always ask “How can I do this better with AI?”

40:08 – What excites Ali most about AI in the next year

Tools & Technologies Mentioned:

Solace Agent Mesh – Multi-agent orchestration platform built by Solace

Confluence – Wiki where AI pulls technical details for slide prep

Claude & ChatGPT – LLMs used for connecting to internal tools

MCP (Model Context Protocol) – Framework for securely connecting AI to enterprise data

Microsoft Copilot – AI inside PowerPoint, Excel, and other Office tools

Mermaid – Visualization tool for generating diagrams and heat maps

A2A Protocol (Agent-to-Agent) – Open standard for agent communication donated by Google

Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

In this episode of This New Way, Aydin sits down with Maddie Engelmeier, AI leader at Motive, to dive deep into how her team is transforming employee productivity with AI. Motive has set an ambitious company-wide goal: boost productivity by 50%. Maddie shares the three-tier strategy behind this initiative, showcases real AI agents in action—from self-assessment tools to executive account summaries—and explains how Motive fosters an AI-native culture across 100,000 customers and 1.3 million drivers.

From using Glean to power performance reviews, to leveraging Notebook LM for instant enablement videos, Maddie gives a behind-the-scenes look at how AI is not just saving time but elevating effectiveness across the company.

Timestamps

0:00 – Setting the stage: Motive’s ambitious 50% productivity goal

1:04 – Maddie introduces Motive and her role leading AI initiatives

3:12 – The three-tier AI adoption framework (democratization, automation, transformation)

6:51 – Why Motive adopted Glean and how it evolved from search to an agentic platform

8:08 – Demo: Self-assessment agent for performance reviews

13:06 – How Glean pulls from Slack, Drive, Gmail & more to save recall time

15:19 – Probing reflection questions vs. copy-paste AI output

19:02 – Over 1,000 unique runs: thousands of hours saved in one cycle

19:27 – Stakeholder feedback agent explained

21:21 – Shifting from recall to reflection and effectiveness

23:50 – Demo: Executive account summary agent for customer insights

27:55 – Scaling AI internally: AI labs, Genius Bars & Slack communities

31:00 – Why AI is enhancing—not killing—creativity

32:07 – Notebook LM demo: from docs to enablement videos in seconds

35:27 – How weekly “snippets” create accountability and unblock teams

36:00 – Agents growing faster than employees—future of work projections

38:05 – Scaling adoption with big, relevant use cases

39:09 – Maddie’s outlook: building comfort with experimentation and collaboration

Tools & Technologies Mentioned

Glean – AI-powered enterprise search and agentic workflow builder, securely connected to company data sources.

Notebook LM – Google’s AI notebook that now generates enablement videos instantly from documents.

Salesforce, Slack, Google Drive, Gmail, Confluence – Data sources integrated into Motive’s AI agents for recall and analysis.

Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

Ethan Hochberg and his partner have built three AI-powered X personas that rack up 200 million views annually—all while running a lean creative agency for AI and tech brands. These aren’t just vanity accounts; they’re distribution engines that feed into their core business.

In this episode, Ethan walks us through:

- How they blend AI content generation with a human touch to keep posts authentic and engaging.

- The Grok Deep Search workflow they use to find and validate influencers in minutes (a task that used to take days).

- A multi-step outreach strategy that personalizes every message for higher response rates.

- How to brainstorm and produce short-form video content with V GPT for TikTok, Reels, and Shorts—without a production studio.

- This is a masterclass in using AI tools not just to save time, but to elevate the quality of your marketing while running a “tiny team.”

Timestamps

02:45 – From AI Marketing Directory to running a creative agency

04:48 – Tiny teams, big results with AI workflows

06:25 – Demo 1: Grok Deep Search for influencer discovery

08:15 – Why influencer marketing is more relevant than ever

12:29 – Grok’s validation process: avoiding fake engagement

15:25 – Applying the workflow beyond influencers (newsletters, blogs, PR)

17:51 – Using AI to think strategically, not just automate

22:09 – The output: influencer tables in minutes

24:01 – Demo 2: Personalized outreach angle generator

28:01 – Applying this to PR and content partnerships

32:00 – Demo 3: V GPT for short-form AI video creation

37:40 – How much of your feed is already AI-generated?

40:00 – Matching viral templates to your niche for higher watch time

41:50 – Using AI video tools mainly for brainstorming content ideas

42:56 – How to connect with Ethan

Tools & Technologies Mentioned

Grok – AI chatbot from X with live access to the Twitter feed; ideal for influencer discovery and current-event research.

Grok Deep Search – Paid Grok feature ($8/mo with X Premium) that crawls hundreds of live sources for validated results.

ChatGPT + GPT Store – OpenAI’s platform for custom GPTs, used here to integrate with video creation tools.

V GPT – AI video generator with realistic avatars, captions, and voiceover.

CapCut (mentioned) – Popular editing tool; V GPT is positioned as a more advanced alternative for avatar-driven content.

Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

Ethan's website: https://marketingguys.co/

Alex Lee joins the show to walk us through an end-to-end automated newsletter generator built using N8N, Airtable, and generative AI. From aggregating news and generating summaries to crafting branded HTML and distributing it via email, Alex shows how businesses can reduce newsletter production from 6 hours to 30 minutes. He also shares how Model Context Protocol (MCP) is enabling real-time access to company data, explains his decision-making process between using workflow automation tools vs. vibe coding, and previews what's next in AI-powered business automation.

Timestamps:

00:23 – Welcome Alex Lee: Career journey from SAP to Google to AI consulting

01:31 – How ChatGPT changed his mind about NLP

02:38 – Why Alex is focused on AI enablement for businesses

03:36 – Use case #1: AI-powered newsletter generator

04:49 – The manual pain of newsletter creation

06:12 – Why email is the best owned marketing channel

07:25 – Step-by-step demo: Aggregating articles, adding context, and generating drafts

09:09 – Human-in-the-loop editing and brand tone tuning

10:01 – HTML generation and branded email output

11:03 – Use cases beyond marketing: Internal custom newsletters

15:26 – Why Airtable powers the backend of the workflow

17:27 – Behind the scenes: N8N automation workflow overview

20:26 – Tool selection: When to use N8N vs. Zapier vs. Make

21:49 – Hosting your own N8N instance for cost efficiency

24:04 – How clients send the generated newsletter (Mailchimp, HubSpot, EasyMail)

27:12 – Vibe coding vs. workflow automation: which path to choose?

28:41 – Why Lovable stands out among V0, Replit, Cursor

30:19 – Benefits of prototyping and vibe coding for non-technical folks

31:24 – What is MCP and why it matters

33:05 – Example: Using Claude + MCP to search Google Drive and draft an executive summary

38:59 – AI-powered time tracking via calendar and file analysis

40:58 – What’s next: legacy system integration, coding agents, MCP standardization

42:39 – How to contact Alex

Tools and Technologies Mentioned:

N8N – Open-source workflow automation platform used to orchestrate the newsletter process

Airtable – Serves as the data layer and user interface for non-technical users

Claude (Anthropic) – Used for summarization, HTML generation, and MCP interaction

MCP (Model Context Protocol) – Enables AI models to access external systems like Drive and calendars in real time

Zapier, Make – Workflow automation tools considered depending on client preference

Lovable – No-code/low-code app builder that successfully integrates with Supabase and OpenAI

HubSpot, Mailchimp, EasyMail – Email service providers used to distribute the newsletters

Supabase – Backend database often used in vibe-coded apps

Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

In this episode of This New Way, Aydin sits down with Tom Crawshaw, founder of an AI automation agency, to explore how he built an AI SDR (Sales Development Rep) system that books over $200,000 in sales calls per month—completely automated and with no humans in the loop. Tom breaks down the tech stack, the flow of conversations, and why these two-way AI-powered chats sound so natural that they’re almost undetectable as bots. He also shares how this system scales personalized customer conversations at a fraction of the cost, and how similar workflows can be applied to everything from e-commerce abandoned carts to B2B demo follow-ups.

Timestamps:

1:15 – Tom’s background and pivot from email/SMS marketing to AI automation

2:57 – Why AI enables true two-way conversations at scale

4:06 – Building custom AI SDR agents vs. off-the-shelf chatbots

6:09 – Live demo: Booking a sales call through the AI SDR workflow

10:13 – How the system qualifies leads and handles objections

12:04 – Tech stack breakdown: Go High Level, N8N, Twilio, and A2P verification

17:02 – Under the hood: prompts, custom fields, and conversation logic

23:00 – Automating what 1,000 SDRs would do manually

27:04 – Costs: Running conversations at $0.25 each

29:25 – Other use cases: abandoned carts, B2B no-show follow-ups, e-commerce

34:00 – Context files: training AI on viral posts and high-performing copy

38:14 – Prompt Cowboy: turning lazy prompts into viral-ready content

40:29 – Where to follow Tom and learn more about AI SDR systems

Tools & Technologies Mentioned:

Go High Level – CRM platform used for SMS automation and pipeline management

N8N – Workflow automation tool connecting AI agents and custom scripts

Twilio – SMS and WhatsApp messaging infrastructure

A2P Verification – Compliance process required for sending business SMS in the US and Canada

OpenAI / Claude – LLMs powering natural language conversations

Prompt Cowboy – Tool for turning simple prompts into fully structured, optimized ones for better AI output

Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

In this episode, Aydin sits down with Rob Williams, a former Chief Product Officer turned AI consultant, to explore the future of work, apps, and personal development—powered by generative AI. Rob demos Limitless, an AI pendant that helps him become a better human, and Claude Code, an agentic AI development environment that builds apps like a team of tireless developers. Plus, he shares his game-changing discovery-to-deliverable workflow that cuts a week’s worth of consulting into a single day.

Timestamps:

01:00 – Rob’s tech background and founding an AI consultancy

05:01 – Demo 1: Limitless AI pendant – the wearable mentor

08:19 – Rob’s daily AI automations for personal growth

10:28 – The privacy dilemma and how Rob handles it

13:35 – Society’s shifting comfort with constant recording

18:20 – Rewind: screen-tracking AI and quantified work

21:16 – Dystopia or augmentation? Competing views on AI ubiquity

27:02 – Demo 2: Claude Code – a real agentic AI dev experience

33:10 – Claude Code spins up dashboards from Excel in minutes

37:39 – Debugging and security auditing with Claude

40:20 – Rob’s gamified AI-powered habit tracker

41:47 – Claude Code for prototyping with dev teams

44:47 – Implications: Will dynamic apps kill the App Store?

47:00 – AI as the new operating system

50:26 – Future: UIs disappear, apps build themselves

52:00 – Demo 3 (Explained): Deep research AI for consulting workflows

54:00 – Talking for the AI: How Rob narrates calls for context

58:30 – Why you must rethink—not just speed up—your workflows

59:36 – Two more tips (in newsletter only!)

Tools & Technologies Mentioned:

Limitless (limitless.ai) – Wearable AI pendant that records, transcribes, and summarizes your day with daily automations and feedback loops.

Claude Code – Anthropic’s CLI tool for building full applications using agentic AI workflows, including dependency management and debugging.

Rewind – Screen-capturing app that logs your activity with searchable recall capabilities.

Fellow – AI meeting tool that transcribes and summarizes meetings. Used by Rob for work-related action tracking.

Typora – Markdown editor Rob uses to annotate and refine AI outputs.

Deep Research – Rob’s name for his long-context LLM-based analysis prompt stack, used for summarizing 20+ hour discovery projects.

RescueTime – Productivity analytics tool used to track app usage and categorize time spent.

👉 Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

What if your intern had a better executive assistant than your CEO? In this episode of TNW, Aydin sits down with Nick Sonnenberg—Wall Street veteran, bestselling author, and founder of Leverage—to explore how AI agents are radically transforming how work gets done. Nick demos a real AI-powered assistant that can handle email, prep meeting briefs, and even outperform human teammates. You'll hear how he structures agent orchestration, optimizes prompts as IP, and envisions a future where everyone at every level has a digital assistant.

What You'll Learn:

- Why every employee (not just execs) should have their own AI assistant

- How AI agents can be “managed” like employees—with hierarchy and QA

- The role of MCP (Model Context Protocol) in unlocking personalized, high-context automation

- Why context is king in building truly useful AI workflows

- Why prompt libraries should be treated like company IP

Timestamps:

00:00 – Imagine if interns had better EAs than execs

01:00 – Nick’s background: From high-frequency trading to AI consulting

02:20 – The origin of Leverage and the obsession with efficiency

03:10 – Inbox Zero, RAD framework, and AI-powered email agents

04:50 – What is MCP and why it matters

06:20 – The CPR framework: Communicate, Plan, Resource

08:30 – Orchestrator agents and agent

11:00 – QA as the new job for every role

12:00 – Why execs are adopting AI faster than junior employees

13:40 – The “Sniper Agent” and building executive briefs

16:00 – Personalized, context-rich email drafting

20:00 – Prompt optimization strategy as a business asset

22:30 – Real-time battle card generation and agent chaining

23:30 – Custom summaries using frameworks, not just transcripts

24:00 – Behind the scenes: building and deploying agents

Tools & Technologies Mentioned:

Claude – Anthropic’s AI model, used here with MCP and custom agents

MCP (Model Context Protocol) – Allows secure access to private data and agent orchestration within AI platforms

Asana – Project management tool integrated with agents

Zapier / NADN / Crew AI – Automation platforms for building AI workflows

Perplexity – Used to scan public web/news as part of the AI brief generation

Coda / Notion – Popular tools for knowledge capture, now evolving into AI-integrated workflows

HubSpot – CRM used to integrate and personalize AI-generated content

Fellow – AI meeting intelligence tool for smarter call summaries

Nick's website: https://www.getleverage.ai/

Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

Solon Angel is the founder of MindBridge and now Remitian, and he’s been at the forefront of applying AI to deeply unsexy but powerful domains like accounting and tax compliance. In this episode, he shares the origin story of MindBridge, how a DeepMind demo changed his life, and what it’s like to build a modern startup where AI plays the role of a product manager, podcast producer, and even financial advisor. Solon also demoed his newest AI agent that proactively manages tax remittances before late fees hit. If you're wondering what the future of AI-powered businesses looks like, this is a masterclass.

Timestamps:

00:00 — The DeepMind demo that inspired Solon

01:00 — Solon’s background and the early days of MindBridge

03:00 — The “dumb rule” state of AI in financial auditing

04:30 — Selling AI to skeptical accountants in 20150

6:00 — The staggering cost of late tax fees ($60B/year!)

08:00 — Remitian: an AI agent that pays your taxes for you

10:00 — Why Fellow is a core part of how Remitian runs

11:30 — How AI helps eliminate the need for a product manager

13:00 — Rewriting 3 years of code in 3 months with AI

16:00 — The shift in what matters: creativity over code

18:00 — Calorie-tracking app Cal AI and teen founders

19:00 — Solon’s AI-powered investment tool (+21% YTD)

20:00 — Live demo: AI agent managing tax payments

23:00 — Future vision: AI offering instant tax loans

25:00 — How Remitian uses Notebook LM for internal podcasts

27:00 — AI updates for board members in 10-minute clips

28:00 — Notion AI’s “research mode” vs. “ask” mode

30:00 — Predicting the rise of startups for content auto-archiving

33:00 — Solon’s final thoughts: beating billion-dollar firms with AI

Tools & Technologies Mentioned:

Fellow – Used for meeting AI transcripts, pre-reads, and knowledge sharing

Notebook LM (Google) – Turns transcripts into internal podcasts

Notion AI – Used for deep research, summarizing objections, and discovering product insights

Slack – Centralized communication, connected with other AI tools

Cursor – AI coding tool used to rewrite years of code in months

Soft Type 2 – Mentioned in relation to efficient AI-based prototyping

Cal AI – Food photo calorie tracker built by a 17-year-old founder

ChatGPT Vision – Used by Solon to interpret emotions via facial expressions

Custom AI Trader – Built by Solon for sentiment-based trading, outperformed the market

Remitian's AI Agent – Calls users, checks funds, splits tax payments, and offers loans

Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

In this episode, Aydin sits down with Greg Shove, CEO of Section, to unpack how AI isn't just a productivity tool—it's a new cognitive layer for modern organizations. Greg shares how Section pivoted from executive education to AI enablement after a single eye-opening session with ChatGPT. He dives deep into what it really takes to embed AI into workflows, culture, and decision-making—and why “talking to AI” is now mandatory at his company. From building a company-wide second brain with Claude to simulating board meetings with GPT, Greg offers a masterclass in practical AI integration.

Timestamps:

1:35 – Greg’s background: from flameouts to $250M in exits

2:00 – Section’s pivot from exec ed to AI enablement

3:01 – The 6-month internal resistance to AI

4:50 – Why training isn’t enough: the real AI challenge is change management

6:07 – Why treating AI like regular software is a strategic mistake

8:26 – What successful AI deployments have in common

10:02 – Lessons from Shopify, Duolingo, and Fiverr on AI expectations

11:45 – The price of AI is too low—why that might change

14:03 – AI vs. analyst time: “an hour becomes a minute”

15:31 – Section’s 25% productivity gain with AI

18:58 – Measuring productivity impact without perfect data

21:24 – Clever metrics: output per headcount, OKRs, AI shoutouts

24:51 – Using Claude as a company “second brain”

26:11 – Greg’s AI desktop setup: Perplexity, GPT, Claude

27:43 – The Section Expert: maintaining company context for AI

29:27 – “Working with Greg” manual: how to humanize your AI input

31:00 – The difference between values and operating principles

34:42 – Roleplaying board members with AI before real board meetings

36:05 – Claude vs. ChatGPT vs. humans: who gave better board insights?

41:00 – AI for owner-operators: create your own board

42:26 – What Greg’s most excited about: how AI unlocks new opportunities

44:08 – Where to find Greg & Section + listener discount

Tools & Technologies Mentioned:

Claude (Anthropic): Used to build a company-wide second brain and simulate board member personas

GPT (OpenAI): Used as a daily thought partner and board advisor

Perplexity: A go-to AI for fast, accurate information lookups

Section Expert (Claude project): A centralized AI project workspace housing all of Section's key documents for brainstorming

ProfAI (Section’s product): An AI-powered coach designed to teach people how to use AI effectively

ChatGPT for Teams: Mentioned as a better, paid alternative to free-tier tools

Gemini Pro: Noted for its screen-sharing and future context-awareness potential

Copilot (Microsoft): One of several LLM tools tested during board simulations

Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

What if your AI agent could send emails, check your calendar, and even text people on your behalf—all securely and with your permission? In this episode, Aydin and guest co-host Alexandra from Fellow talk with Sam Partee, co-founder of Arcade, about how AI agents are actually becoming useful in the real world.

Sam breaks down how Arcade enables LLM-powered agents to act on your behalf across tools like Gmail, Slack, Salesforce, and more, without sacrificing security. He also shows us how he automates his own workflows, from email triage to iMessage replies, and shares how tools like Cursor and Claude are reshaping how engineers work day-to-day.

Whether you're technical or not, this episode is packed with actionable insights on what it means to work in an AI-native company—and how to start doing it yourself.

Timestamps

0:00 – The future of agents impersonating people

01:20 – Meet Sam Partee and his background in high-performance computing

02:50 – What Arcade is and how it powers AI agents

05:10 – Use case: ambient social media agents

06:50 – “YOLO mode” vs. human-in-the-loop agent workflows

07:30 – Building a lean AI-native company

08:00 – Engineers are now 1.5x more productive—with caveats

12:00 – Why the whole team (PMs, QA, etc.) should use tools like Cursor1

4:00 – How Markdown became the LLM-native format

17:00 – Sam’s iMessage agent and calendar automation

18:45 – His AI-powered inbox (email triage + drafting)

21:00 – Live demo: using Slack assistant “Archer” built with Arcade

24:00 – How non-technical people can use these tools too27:00 – Cursor vs. Copilot: What’s better?

30:00 – Cursor agent mode and example developer workflows

34:00 – Vector databases and prompt design

35:00 – Using LLMs to redesign error handling and generate docs

38:00 – Advice for teams adopting AI: start by building

Tools and Technologies:

Arcade – Let AI agents act on your behalf (email, Slack, calendar, etc.) with secure OAuth.

Cursor – LLM-native IDE with full-codebase context. Ideal for AI-assisted development.

Claude – Chat interface + agent orchestration, paired with Arcade.

LangGraph – Multi-agent orchestration framework with human-in-the-loop support.T

ailScale – Secure remote networking; enables Sam to access agents from anywhere.

Twilio – Used for SMS reminders and notifications.

Obsidian + Markdown – Sam uses Markdown + AI for personal notes and research.

GitHub Copilot – Used in tandem with Cursor for inline suggestions and PR reviews.

Subscribe to the channel for more behind-the-scenes looks at how top teams are rethinking work with AI.

Subscribe at thisnewway.com to get the step-by-step playbooks, tools, and workflows.

In this episode, Edmundo Ortega, partner at Machine & Partners, joins to explore how AI is transforming the way we build and ship software. He shares real-world use cases—from cutting document processing time by thousands of hours to prototyping apps with tools like Bolt in under 15 minutes.

We also dive into:

• How non-technical team members can now contribute to development

• How Ed uses Claude to write in his own voice

• Why leaders should think beyond cost savings to unlock AI’s full potential

Whether you're a developer, founder, or team leader, this episode will reshape how you think about building with AI.

Timestamps

00:00 – The promise of AI beyond cost savings

01:10 – Edmundo's background and the founding of Machine & Partners

04:40 – How Machine & Partners identifies and builds AI use cases

05:45 – Personal shift to an AI-powered workstyle

07:00 – Building internal tools using Bolt

09:20 – Demo: Creating a chart parser app using Bolt

12:00 – Claude vs. ChatGPT for writing with custom instructions1

5:15 – Use case: Automating document extraction for analysts

21:00 – Workflow transformation with AI and report generation

25:15 – PMs using AI to prototype and inspire devs

33:00 – AI's impact on team structure and dev productivity

38:24 – How to get in touch with Ed and his advice for the next 18 months

Tools & Technologies Mentioned

• Bolt (bolt.new)

• Claude 3.7 (Anthropic)

• ChatGPT / GPT-4.5

• Cursor (AI coding assistant)

• Chart.js• Google Sheets

• OpenAI API

• Replit

• N8N

• Make.com

• Gumloop

• Relay

• SectionSchool.com

Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.