AI makes starting easier, not finishing automatic.
AI can create a first draft quickly. Shipping useful work still requires judgment, review, timing, and careful follow-through.
The gist
Episode 2 connects creative resistance—the difficulty of beginning and finishing meaningful work—with the operational resistance that appears after a promising AI demo. AI can make it easier to start, but useful results still need review, ownership, and a reliable place to live.
Will's field-sales workflow makes the idea concrete. AI helps identify promising accounts in Pennsylvania and Ohio, prepare for meetings, capture notes through Telegram, and organize the results in Notion or a CRM. Those notes then become follow-ups, proposals, product requests, calendar updates, and assigned tasks.
Key takeaways
Six practical lessons from this episode.
Plan sales trips around the best opportunities.
Before the rep travels, use AI to rank territories, companies, contacts, outreach ideas, and likely next steps.
Turn fresh voice notes into structured CRM updates.
A short Telegram memo after a meeting can become contact notes, a follow-up date, an assigned task, and a cleaner shared record.
Give agents a dependable view of the business.
Agents become more useful when they can read trusted information from Notion, Vercel, GitHub, Sentry, analytics, search data, and customer notes.
Protect time for customers.
Use AI to reduce administrative work so more time goes to buyer conversations, better preparation, and thoughtful follow-up.
Better context produces better creative work.
Generated images, thumbnails, copy, and strategy improve when the model understands the brand, audience, and specific goal.
Tactical guidesWhat to try next
Build an AI-assisted field-sales trip plan.
Use AI before the trip to turn a broad region into a ranked, realistic plan for who to meet and why.
- Define the territory, dates, ideal customer profile, product-fit signals, and firm constraints.
- Ask AI for candidate companies by geography, vertical, size, trigger event, and buying likelihood.
- Rank accounts with a simple score: fit, urgency, proximity, relationship path, and deal size.
- Identify the people to meet, then write one useful reason to talk to each person.
- Turn the ranked list into a travel route, outreach queue, meeting brief, and CRM view.
Run a Telegram-to-CRM capture loop.
Capture each meeting while the details are fresh, then let AI organize the notes for review.
- Create a private Telegram bot or agent chat that can receive text and voice messages.
- Connect the agent to a safe Notion or CRM workspace with scoped read and draft permissions.
- After each meeting, record a short memo: who, what happened, objections, next step, and date.
- Have the agent draft contact notes, task records, follow-up reminders, and missing-field questions.
- Review the draft before the agent writes to the source of truth or sends anything externally.
Create a daily company-brain digest.
Give operators a short, evidence-backed morning view of what changed and what needs attention.
- Connect only the systems that matter: deploys, code, crashes, analytics, search, customer data, and tasks.
- Schedule a morning digest with sections for wins, risks, anomalies, pending decisions, and owner-ready tasks.
- Require evidence links for every claim, especially traffic spikes, broken pages, bug reports, or campaign movement.
- Rank items by urgency and business impact instead of presenting every metric with equal weight.
- End with a short action list: what to inspect, fix, expand, or ignore.
Turn calls into clear follow-up actions.
Use AI to prepare the administrative work after a call, then review it before anything is sent or assigned.
- Import the transcript, meeting notes, and CRM context into one reviewable workspace.
- Ask AI to extract needs, objections, promised follow-ups, stakeholders, timing, and deal risk.
- Draft the follow-up email, proposal outline, calendar invite, quote request, and internal handoff.
- Create tasks for sales, product, engineering, or support only when the transcript supports them.
- Save personal details only when they are relevant, appropriate, and useful for future service.
Capture inspiration without project sprawl.
Capture promising ideas quickly without allowing every idea to become a new project.
- Make the capture step one gesture: voice memo, quick note, Telegram message, or action button.
- Ask the agent to produce a tiny artifact: sketch, brief, outline, landing page draft, or task card.
- Store the artifact in Notion with tags for customer value, timing, effort, risk, and owner.
- Let the agent compare new ideas against existing commitments, bugs, deadlines, and revenue work.
- Promote only the ideas that survive review; archive the rest for later.
Make creative assets with real brand context.
The thumbnail discussion points to a broader rule: creative AI works best with specific brand and audience context.
- Collect the brand kit: colors, type, image rules, prior thumbnails, tone, and audience.
- Start from the episode theme or transcript, then generate a few precise visual directions.
- Use image generation for backgrounds, objects, art styles, mood, and variations that used to require stock-photo hunting.
- Review for taste: composition, brand fit, readability, and whether the idea matches the episode.
- Save effective prompts and approved assets so future work stays consistent.
ChaptersTimestamp map
00:00↗
01:27↗
03:25↗
06:31↗
08:20↗
10:57↗
13:01↗
15:48↗
17:15↗
19:31↗
23:00↗
26:57↗
29:24↗
31:00↗
33:48↗
36:08↗
39:26↗
42:00↗
Rick Rubin, The Creative Act, and resistance
The episode opens on Rick Rubin's creative philosophy and the surprising resistance that appears before meaningful work.
Grind mode, flow mode, and multiple paths to output
The hosts compare artists, athletes, engineers, and founders who arrive at good work through different rhythms.
Find your workflow by testing the tools
AI advice becomes useful only when you test it with your own business, customers, schedule, and working style.
Creative work still has a hard middle
Training, coding, school, design, and content creation all include days where getting started is the hardest part.
Rapid AI demos versus production systems
AI can create a concept quickly, but production still requires testing, polish, reliable data, safe permissions, and follow-through.
Planning a real sales trip with AI
Will explains how AI helped narrow Pennsylvania and Ohio into companies, people, and outreach angles worth pursuing.
Plane Wi-Fi, Telegram, and a laptop at home
Text-only Wi-Fi becomes useful because Telegram can reach OpenClaw on a laptop connected to the work systems at home.
Unstructured meeting notes become structured tasks
Voice notes after meetings turn into contact updates, follow-up dates, task records, and CRM-ready data.
How agents connect AI models to business systems
The hosts compare OpenClaw, Hermes, Codex, and Claude Code as tools for connecting capable AI models to real work.
OpenClaw as COO and CTO
Enzo describes using agents to watch deploys, GitHub, Sentry, analytics, search data, Hotjar, and daily business signals.
Reducing the delay between idea and action
Agents can help people act faster when the right tools, information, and permissions are already in place.
Inspiration is perishable
The conversation returns to capturing ideas immediately while still respecting the toil needed to finish anything valuable.
Notion as the prioritization layer
Meeting transcripts, email, deadlines, and project ideas feed a company brain that can rank what matters next.
Remove small barriers to useful action
Action buttons and EV charging show how every extra step can make capture, follow-up, or review less likely.
Sales reps, admin drag, and customer time
The hosts focus on the gap between high-value customer work and the CRM updates, proposals, quotes, and coordination around it.
Create a personalized audio briefing before a call
AI can brief a traveling rep on the buyer, company, relevant history, competitors, and goals for the next conversation.
Thumbnails, image generation, and brand fit
Creative production shifts from hunting for approximate assets to generating the right image in the right style.
Every function should ask where AI fits
The closing argument is broad: sales, content, coding, finance, strategy, and operations all have daily workflows worth rebuilding.
The stackTools & concepts
OpenClaw + Telegram + Notion
Field sales · company brain
Send a text or voice note to an agent from your phone. The agent uses trusted company information to draft structured updates, which a person reviews before anything reaches the CRM or task database.
The Creative Act
Rick Rubin's book frames the opening discussion about resistance, flow, taste, and creative work.
Open publisher pageTetragrammaton
Rick Rubin's interview show, referenced during the episode's discussion of creativity and technology.
Open TetragrammatonOpenClaw
A persistent assistant that can connect messages, tools, memory, and workflow actions.
Open OpenClawHermes Agent
A self-hosted framework for agents that need memory, tools, messaging, and scheduled work.
Open Hermes AgentModel Context Protocol
An open standard that lets AI applications connect to tools, data, prompts, and other resources.
Open MCP docsNotion API
The official interface for connecting agents to Notion pages, databases, task fields, and shared company knowledge.
Open Notion developersTelegram Bot API
The official API path for building a Telegram bot that can receive messages and route them into an agent workflow.
Open Telegram APISentry Cron Monitoring
Monitoring for scheduled jobs, useful for confirming that automated reports and checks ran as expected.
Open Sentry cronsGoogle Analytics
A source of website and campaign data for daily reports, trend checks, and operational decisions.
Open Analytics docsSearch Console API
Search-performance data that can support an AI-generated daily or weekly website health report.
Open Search Console APIHotjar
Behavior analytics for seeing where users click, scroll, struggle, or leave.
Open HotjarSalesforce State of Sales
The episode cites the low-customer-time problem; Salesforce's public write-up reports reps spend just 28% of their week selling.
Open Salesforce researchCodex
A coding agent that works with repository context, development tools, tests, and review evidence.
Open Codex docsClaude Code
A command-line coding agent discussed as another way to work directly with repositories and development tools.
Open Claude Code docsOpenAI image generation
Tools for generating or editing specific visual assets instead of searching for approximate stock images.
Open image docsLinks & references
EpisodeWatch the full episode — YouTube↗
GuideBuild an AI-assisted field-sales trip plan — rank accounts and prepare the route↗
GuideRun a Telegram-to-CRM capture loop — voice notes to structured records↗
GuideCreate a daily company-brain digest — metrics, unusual changes, and next actions↗
GuideTurn calls into clear follow-up actions — emails, proposals, and assigned tasks↗
IndexReturn to the Shoreline resource library — homepage↗
“
A practical AI sales workflow is more than one prompt. Capture the conversation while it is fresh, add the right context, structure the result, review it, and save the next action in the right system.
Shoreline Ep. 2 · distilled operating principleListener checklist
Pick one field workflow.
Choose trip planning, meeting prep, post-call notes, follow-up drafting, or daily ops reporting. Keep the first workflow narrow.
Choose the source of truth.
Decide where contacts, tasks, transcripts, follow-ups, and account notes should live before connecting any agent.
Add one mobile capture surface.
Use Telegram, voice memos, or a private form so real-world notes can enter the system immediately.
Draft, then review.
Let AI create structured updates, but keep the first few weeks behind a human approval gate.
Schedule a daily summary.
Ask the agent to report wins, risks, changed records, missing information, and suggested next actions.
Keep only what creates real value.
Keep workflows that reduce administrative work or improve customer conversations. Archive demos that add more review than value.