This New Way

This episode is a full “build a business in 40 minutes” demo showing how AI collapses what used to take teams (creative production + sales ops + support) into a handful of prompts. Samruddhi generates a high-production video ad in Google AI Studio using a JSON-style prompt framework, then spins up a working voice sales/support agent in Vapi via Claude Desktop + MCP—so the agent is created from a single prompt instead of clicking through the UI. The conversation also covers why “interfaces matter less” in an agent-first world, why workflow tools (like n8n) still have a role, and how memory layers like Mem0 unify context across channels (email/WhatsApp/etc.) so you can take actions without hunting.

Timestamps

0:00 — “Single person billion-dollar company” belief + AI driving 10x execution speed
1:57 — Plan: create the ad in Google AI Studio (Veo 3.1) + build a voice agent using Vapi MCP via Claude Desktop
2:42 — Smithery: marketplace for MCP servers
3:39 — MCP for non-technical listeners: “like an API, but agents use it to talk to external services”
4:22 — Inside Vapi MCP: tool list = APIs the agent can choose from
5:06 — AI Studio setup: video generation playground + select Veo 3.1
6:16 — JSON prompting framework begins (structure → production-level output)
6:28 — Keys: description, style, camera, lighting, environment, elements, motion, ending, text
9:05 — Prompts/scripts can be AI-generated (humans provide guardrails)
10:41 — Need an API key to generate videos in AI Studio
10:54 — Ad review: strong realism; last segment looks AI-ish → iterate prompt
13:05 — Install Vapi MCP via npx from Smithery + add Vapi API key
13:46 — Claude Desktop: Vapi MCP appears under Connectors/Tools (not Claude web)
14:05 — Prompt the agent build: “Fresh Pause” + role, tasks, FAQs, call flows
18:23 — Testing: “Talk to assistant” starts a live call simulation
19:20 — Deployment: assign a phone number; Vapi provides free/test numbers (up to a limit)
21:57 — Mem0 / Supermemory: memory layer across apps/agents to keep context
24:13 — Why memory layers help: fewer MCPs → less slowdown/hallucination; no need to specify where to search
26:36 — MCPs + slide decks: mention of Gamma MCP via Claude
27:34 — Future of n8n/Zapier: they persist, but prompting increasingly generates workflows
31:38 — Prediction market trading algos (Kalshi/Polymarket) + AI improves speed/decision-making
36:02 — Closing vision: help orgs 10x execution speed, especially non-technical leaders (40+) with domain expertise

Tools & technologies mentioned

Google AI Studio (Video Generation Playground) — Generate an 8-second video ad.

Veo 3.1 — Google video model used for “production-level” output.

JSON Prompting Framework — Structured key/value prompts for story, visuals, camera, lighting, motion, ending frame.

Claude Desktop — Runs connectors/tools (including MCP servers).

MCP (Model Context Protocol) — Lets agents call external services/tools based on intent.

Smithery — Directory/marketplace for MCP servers.

Vapi — Voice agent platform; create agents + assign phone numbers.

Vapi MCP Server — Enables Claude to operate Vapi via prompts (create/list/configure).

npx — Installs MCP server quickly from the terminal.

API Keys — Required for AI Studio generation + Vapi authentication.

Mem0 / Supermemory — Cross-channel memory layer to retrieve context automatically.

Knowledge Graph — Underlying structure for semantic retrieval across interactions.

Glean — Referenced as a comparison point for search/context retrieval.

Gamma MCP — Example of generating slide decks via MCP.

n8n / Zapier — Workflow automation tools discussed in an MCP-first future.

OpenClaw — Mentioned as agent tooling that can help with steps like obtaining API keys.

Kalshi / Polymarket — Prediction markets referenced in the trading/AI speed discussion.

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

Manuela Barcenas breaks down how marketing work has flipped from “writer + editor” to “manager of agents.” She shares two concrete workflows: (1) using Claude Projects to reposition and modernize 100 legacy blog posts in a week (including updated product messaging, AI-forward advice, and internal links), and (2) using Fellow’s “Ask Fellow” to mine anonymized customer-call transcripts for original quotes and pain points—then turning those insights into publish-ready integration/use-case articles in hours, not weeks. The throughline: output is easy now; taste, judgment, and review are the differentiators.

Timestamps

0:00–0:00 - Intro

1:18–2:54 Early Fellow days: one blog/week, months-long ebooks, craftsmanship vs scale

3:06–3:26 Scale expectations now: Amazon’s ebook upload limit anecdote (3/day)

3:40–4:30 Fellow previously managing an “army of writers” → now mostly AI/agents

4:36–5:00 “Taste” as the differentiator: what good content is + standing out

5:53–7:12 The 100-post update explained: not link swaps—full repositioning + modernized advice

7:25–9:36 Switching from ChatGPT to Claude; LinkedIn poll results + “context retention” theme

9:48–10:21 Claude Projects setup: separate projects to maintain context and instructions

14:43–15:29 Prompt versioning: internal links, new features, and repeated refinement cycles

18:55–19:20 Demo: paste URL → Claude fetches page → follows checklist automatically

19:26–20:24 Manuela’s QA: she reads/edits everything; “taste” = final layer (like editing writers)

21:38–23:17 Claude Skills discussion: turning repeated workflows into reusable MD “skills” (personal vs company-wide)

25:42–26:26 SEO myth: focus isn’t “AI penalty,” it’s originality and substance (quotes, stats, real insight)

26:38–28:39 Original content engine: Ask Fellow pulls anonymized customer-call insights by feature/integration

28:39–31:21 Building documents from transcripts (pain points, best practices, FAQs, quotes) → export to Doc/PDF

31:21–33:29 Feed exported insights into Claude Project to draft a tight article rich with customer quotes

33:29–36:06 Why it works: management loop (outcomes → constraints → review → feedback) at faster cadence

36:18–37:30 What’s next: Claude Code / Claude “co-work”; projects as “mini employees”

37:02–38:06 Personal brand workflow: Claude analyzes best LinkedIn posts → style guide + voice-based drafting (Whisper Flow)

38:28–39:12 Wrap: AI speed is real; staying current requires constant learning

Tools & technologies mentioned (with brief descriptions)

Claude (Anthropic) — LLM used for higher-quality long-context writing, structured rewrites, and content systems.

Claude Projects — Workspace feature to keep persistent instructions/context per workflow (e.g., content optimization agent).

Claude Skills — Reusable capabilities packaged as uploaded markdown files (personal or org-wide) to standardize output.

Claude Code / Claude “co-work” — Anthropic workflows/webinars referenced for deeper automation beyond writing (emerging).

ChatGPT — Baseline comparison model; Manuela notes switching due to Claude’s perceived context + output quality.

Excel + Claude — Mentioned via finance demo: using Claude in Excel to build financial models.

Fellow.ai — AI meeting assistant used for transcripts, summaries, action items, and cross-tool integrations.

Ask Fellow — Fellow feature that queries meeting knowledge (calls/transcripts) to generate anonymized insight docs.

Anonymization (in Fellow) — Removes identifying customer details while preserving job titles/quotes for safe content use.

Integrations (examples named) — Slack, Asana, HubSpot, Salesforce, Linear, Jira, Confluence (tools Fellow connects with).

Whisper Flow — Voice-to-text capture tool used to speak ideas, then convert into styled writing (e.g., LinkedIn drafts).

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

AI is pushing knowledge work toward a world where “leaders manage agents”—and eventually, where some management functions themselves are handled by AI. Shweta Kamble and Hari Iyer (founders of HaloVision) unpack that future and demo what it looks like today: an AI “third-party” that runs confidential 1:1-style conversations with employees, synthesizes themes into quantified “case files,” and creates a bidirectional channel between executives and the org.

00:00 - Intro
01:02 — “Undercover Boss” analogy: AI can surface ground-truth operational fixes at scale.
02:01 — “No ICs anymore”: the shift to managing armies of agents.
03:06 — AI can outperform average managers at listening, context, and coaching—at scale.
04:47 — Introducing Halo Vision: management + AI as a core intersection.
05:19 — What Halo does: confidential 1:1 conversations, analyzed into exec-ready insights.
06:00 — Key difference: not a suggestion box—Halo quantifies impact and outcomes.
06:36 — 1:1 controversy (e.g., “don’t do 1:1s”) and why time cost matters.
08:11 — Third-party confidentiality: why employees share more with Halo than internal tools.
09:30 — SurveyMonkey comparison: blending “survey + 1:1 + executive alignment.”
10:50 — Feedback loop requirement: employees must believe feedback leads to change.
12:06 — Founders’ backgrounds (Zoom AI/data products; CS/product design; Cisco ventures).
16:28 — Building Halo = “several companies in one”: auditing, privacy, PM estimation, infra.
18:03 — “Telephone game” across agents: why infra/evals matter for compound accuracy.
19:47 — Defining evals: correctness, reasoning tests, summarization/synthesis checks.
23:32 — Concrete eval example: summaries must trace back to transcript evidence.
27:03 — Added complexity: longitudinal context and time relevance (“6 months ago may not matter”).
30:39 — Prompt → context engineering: getting the right info to the model at the right time.
32:16 — Why off-the-shelf tools weren’t enough: auditability and tracing across abstraction layers.
37:18 — Live demo setup: Halo’s internal “case file” view with quantified issues.
38:01 — Example case files: exec jumping into low-level decisions; burn rate + delay cost estimates.
41:16 — Live call begins: confidentiality disclaimer + agenda choices.
41:50 — Halo’s questioning style: reflective, probing, tailored follow-ups.
46:17 — Positioning: Halo doesn’t replace 1:1s—it makes them more effective and focused.
47:00 — What they’re excited about next year: science/research advances + shifting human work.
Tools & technologies mentioned

Halo Vision — AI “third-party” that conducts confidential employee conversations, synthesizes insights into quantified exec recommendations, and helps align understanding across the org.
Evaluation frameworks (Evals) — Methods to test AI outputs (reasoning, summary accuracy, grounding) to prevent misleading conclusions and compounding errors in agent workflows.
LLM-as-a-judge — Using an LLM to grade another model’s output for correctness, grounding, or quality; often paired with other checks.T
racing / auditability / evidence links — Attaching each summary claim to specific transcript excerpts so you can prove where conclusions came from and debug errors.
Speech-to-text / transcription — Converting conversations into text artifacts that can be analyzed, summarized, and traced.
Fellow.ai — AI meeting assistant that joins meetings, summarizes, tracks actions/decisions, integrates with common work tools, and supports sensitive meetings with privacy/security controls.
Gemini (Google) — Mentioned as performing strongly for some use cases relative to other models at the time of recording.
GPT-4 / GPT-5 (and “5.2”) — Used as examples of model shifts affecting product behaviour (reasoning chains, tone/EQ, evaluation requirements).

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 Hai Nghiem from AGI Ventures Canada to explore how Claude Code is changing the way teams build software, automate workflows, and even run go-to-market operations—without requiring everyone to be a developer.Hai walks through real, hands-on examples of using Claude Code as a terminal-based AI agent to qualify inbound leads, generate follow-up emails and statements of work, manage internal context with skills and sub-agents, and even automate browser-based tasks like filling out applications. The conversation dives deep into go-to-market engineering, context engineering, and why skills are becoming one of the most powerful primitives for scaling AI across an organization.If you’re curious how non-technical teams can start using agents today—or how technical teams can dramatically compress GTM and sales workflows—this episode is a must-listen.Key Timestamps00:00 - Intro00:08.334 – “What’s the killer AI product everyone should be using?”00:25.582 – Hai introduces Claude Code and why it’s blowing up01:10.900 – Claude Code as an agent running in your terminal01:45.600 – Go-to-market engineering and reducing SDR teams02:10.222 – Industry trend: shrinking sales teams with AI agents03:45.976 – Claude Code vs Cursor for coding workflows04:32.100 – Writing 90% of production code with AI (safely)05:45.300 – Non-coding automation with Claude Code, Zapier, and n8n06:01.645 – What AGI Ventures Canada does06:45.900 – AI Tinkers community and the origins of AGI Ventures07:38.958 – Automating inbound lead qualification08:50.839 – Live role play: discovery call walkthrough09:12.607 – Using Notion as a live note-taker and context store10:03.350 – Example GTM automation use cases at Fellow11:52.973 – Running Claude Code with “dangerously skip permissions”13:07.050 – Sub-agents vs skills explained16:40.851 – What Claude “skills” actually are17:15.359 – Email writer skill walkthrough20:19.750 – Auto-updating skills from real GTM learnings22:19.592 – How Claude pulls context from Notion automatically25:42.632 – Generating follow-up emails using skills30:08.595 – Generating Statements of Work with scripts31:35.478 – Browser automation with the Claude Chrome extension32:16.870 – Auto-filling applications using personal skills34:56.562 – AI-powered Discord bot for community support37:18.114 – Live fact-checking inside Discord38:09.159 – How to contact AGI VenturesTools & Technologies MentionedClaude (Anthropic)An AI assistant positioned as a business-focused alternative to ChatGPT.Claude CodeA terminal-based AI agent that can write code, automate workflows, manage files, and interact with browsers—used heavily for GTM and internal automation.Claude SkillsLightweight, reusable instruction sets that teach Claude how to perform specific tasks (e.g., writing sales emails) without permanently consuming context.Claude Sub-agentsDelegated agents used to manage context and offload complex tasks without bloating the main agent’s context window.NotionUsed as a lightweight CRM, document store, and central source of truth for agent context.DiscordPrimary internal and community communication platform, integrated with AI bots for automated responses.Chrome Automation (Claude Extension)Allows Claude Code to control the browser and complete web-based tasks like filling out forms.ZapierNo-code automation tool for connecting apps and workflows.n8nOpen-source workflow automation tool often used for advanced AI and agent pipelines.GPT Models (OpenAI)Currently used in AGI Ventures’ Discord bot, with plans to migrate to Claude models.

Contact Hai:agiventures.ca
[email protected]
https://ca.linkedin.com/in/haiphunghiemSubscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.

Amir (Co-Founder at Humblytics) shares how he builds an “AI-native” company by focusing less on shiny tools and more on change management: assessing AI fluency across roles, setting the right success metrics, and creating shared context so AI can reliably ship work. The big theme is convergence—engineering, product, and design are collapsing into tighter loops thanks to tools like Cursor, MCP connectors, and Figma Make. Amir demos workflows like: AI-generated context files + auto-updated documentation, scraping customer domains to infer ICPs, turning screenshots into layered Figma designs, then converting Figma to working React code in minutes, and even running an “AI co-founder” Slack bot that files Linear tickets and can hand work to agents.Timestamps0:00 Introduction0:06 Amir’s stance: “no AI experts” — it’s constant learning in a fast-changing field.1:59 Cursor as the unlock: not just coding, but PM/strategy/design work via MCPs.4:17 The real problem: AI adoption is mostly change management + fluency assessment.5:18 The AI fluency rubric (helper → automator → augmentor → agentic) and why it matters.8:13 Cursor analytics: measuring AI-generated code and usage across the team.9:24 “New code is ~99% AI-generated” + how they keep quality via tight review + incremental changes.10:58 Docs workflow: GitBook connected to repo → AI edits docs and pushes live fast.14:02 ICP building: export Stripe customers → scrape domains with Firecrawl → cluster personas.17:45 Hallucination in the wild: AI misclassifies a company; human correction loop matters.34:43 Wild move: they often design in code and use an AI-generated style guide to stay consistent.38:10 Best demo: screenshot → Figma Make → layered design → Figma MCP → React code in minutes.45:29 “AI co-founder” Slack bot (Pixel): turns a bug report into a Linear ticket and can hand off to agents.48:46 Amir’s wish list: we “solved dev”; now we need Cursor for marketing/sales → path to $1M ARR.Tools & technologies mentionedCursor — AI-first IDE used for coding and product/design/strategy workflows; includes team analytics.MCP (Model Context Protocol) — “connector” layer (Anthropic-origin) that lets LLMs interface with external tools/services.ChatGPT — used as a common baseline tool; discussed in the context of prompting practices and workflows.Microsoft Copilot — referenced via the law firm incentive story; used as an example of “usage metrics” gone wrong.Anthropic (AI fluency framework) — inspiration source for the helper/automator/augmentor/agentic rubric.GitBook — documentation platform connected to the repo so docs can be updated and published quickly.Firecrawl (MCP) — agentic web scraper used to analyze customer domains and infer ICP/personas.Stripe — source of customer export data (domains) to build ICP clustering.Figma — design collaboration tool; used here with Make + MCP to move from design → code.Figma Make — feature to recreate UI from an image/screenshot into editable, layered designs.Figma MCP — connector that allows Cursor/LLMs to pull Figma components/designs and generate code.React — front-end framework used in the demo for generating functional UI components.Supabase — mentioned as part of a sample stack when generating a PRD.React Router — mentioned as part of the sample stack in PRD generation.Slack — where Amir runs internal agents (including the “AI co-founder” bot).Linear — project management tool used for creating tickets from Slack/agent workflows.CI/CD — their deployment/review pipeline; emphasized as the human accountability layer.Subscribe at⁠ thisnewway.com⁠ to get the step-by-step playbooks, tools, and workflows.