TL;DR
Greg Isenberg and guest Imran walk through building a practical AI chief of staff using a platform called Nebula, demonstrating live agent creation for tasks like blocking radar, project status tracking, vision accountability, daily agenda briefings, and ICP lead sourcing. The video shows non-technical users how to deploy autonomous agents connected to Gmail, Slack, Linear, Calendar, and other tools without writing code.
Key points
1
Nebula is a no-code agent creation platform that abstracts away terminal setup and server config required by tools like Hermes or Open Claw, making it accessible to busy non-technical founders
2
Agents in Nebula are defined by a system prompt, a set of goals, and tool integrations (Gmail, Slack, Linear, Jira, Confluence, Google Calendar, Granola) — mirroring how human team roles work
3
Five agents are built live: Blockage Radar (who is waiting on you), Project Status (what shipped, what is due, what is at risk), Vision Tracker (offsite goal accountability with motivational quotes), Daily Agenda (calendar briefing with second-brain cross-reference), and Late Checkout Prospector (10 ICP leads sourced daily)
4
Nebula agents can be made public and shared via URL so others can clone and remix them without rebuilding from scratch, solving a gap where previously only skill files or CLI configs could be shared
5
Nebula also supports spinning up hosted mini web apps (called mini apps) that connect live to agents, enabling visual dashboards with deep links rather than text-only Slack digests
Actionable insights
→
Start by listing the three to five things a chief of staff would do for you, then map each to an agent with specific tools — this mental model (role + goals + tools) is the fastest way to get productive with agents
→
Use cheaper models like QN 3.6+ or the Nebula model for routine briefing and triage tasks; save expensive frontier models (Claude Opus, GPT-4 class) for coding or deep reasoning tasks to cut costs significantly
→
Spend one week watching yourself work and identify three to five repeatable tasks you do manually, then automate them — frame it as working on your job, not just in your job, and treat it as an ongoing muscle not a one-time project
→
When building outbound lead-sourcing agents, add shared context signals (same city, same university, mutual LinkedIn connections) as bonus scoring criteria to meaningfully increase response rates
→
If you use a messaging automation agent via SendBlue, read their usage rules carefully before deploying outbound sequences to avoid getting your number flagged as spam or rate-limited
Notable quotes
“There is no reason why a human should be looking at your calendar, your email, your LinkedIn messages. We can build agents to do most of that.”
“I think all careers are going to become like that. You need to allocate a percentage of your time to work on your job.”
“I know I can be doing so much more with agents that I am not doing already. That is the biggest thing I have been wrestling with mentally.”
Worth watching?
✅
Worth watching the full video?
Watch if you want to see the agents built live with real prompts and real tool connections — the key setup logic and agent templates are all captured here, but the screen walkthrough adds useful visual context for replicating this yourself.
Saved you some time? The creator still deserves a like.
Found this by searching? That's the hard way.
summree watches the AI channels you follow and sends you the summary the moment a video drops. Nothing gets past you. From £4/month.
Try it free