Want to validate your next product, find early buyers, and know exactly what to sell before you build it? Market research interviews are the answer. And now they're stupidly easy to run. I did this for a few partner-clients in the past. And looking back, it was the first and best step to start. Here's the setup: 1. Offer to do market research to test a product idea, entirely for free upfront, any broke creator would say sure. 2. The ONLY work they do is a a calendar link to the bio of a engaged audience to go book a call (and claim a free gift)... and to promote it, mention it, etc hard for a week or two. 3. Use a structured script (AI writes it, creator approves it, we run it) to interview, ask for pain, life,etc 4. Tell your audience to book a slot to claim a free gift, access to new community 5. AI agents + human backup conduct the interviews, using script, offering upsells w/ a free lead mag, this acts as a soft onboarding and they get happy. Send them all to claim free gift (free skool) 6. Eventually enough data (30 interviews) collected represents the whole of the audience, with AI it may be in the hundreds 7. Finalize with turn common ain points, and product positioning addressing the audience handed to you 8. Feed to an AI "make a course to solve X" and make slide decks. Give it to the creator to run (or run eleven labs) to have an AI version of them record it, say it, and post it to Skool immediately, and say "our suspecions were. correct, we launched the course for free to everyone inside 9. Use the initial feedback and hype, and make it paid, everyone who missed out FOMOs in, and you announce another wave... You keep the calendar link or skool link. Founder wouldn't kick you out as they just lose money. The magic: - Costs you $0 to start (just add the link) - Takes 1 minute to implement - People want to talk about their problems - Your CTA becomes "Book a research call" - Calendars fill up FAST (we've seen them book solid in hours)

https://x.com/naval/status/2019430858248347951 Naval is one of the best communicators and thinkers of this modern area. Here's my take With public access to AI, AI basically gives anyone high level skills and can get unlimited capacity to handle well defined automatic or decision tasks. So the "next" step is "what exactly is the right thing to work on next". Constraints based thinking solves that. Building a team of AI agents to think about how to develop, what to solve, and have humans in charge of making sure they follow the mission and directive their humans give them. We have this QUEST framework and traditionally has been able to give humans the best possible decision making. I see it like generals surrounding a war plan, deciding where to place their resources, time, money, manpower. But soon the decision making also needs to be handled by AI. And virtually most organizations haven't even built their own "War room" yet let alone even think about "how" they can have AI help them. Right now they use it as simple chat bots, "should I do x or y" but not automated decision makers, proactively doing work, fleshing out projects and processes, and actually carrying them out. We are easily a few years ahead of traditional operators and can build something that shortcuts that struggle.

Instead of "sub agents" where you have to individually call each one, and have them "hand off" to the next role without deep context and sharing and discussing, we can develop a team lead bot, that spins up its own sub agents and acts as the "brain" of the group. Why is this game changing? This QUEST protocol, can be explained to a main team lead agent, that says "OK I see that we need 6 agents, one to check in to each area, one to track data and make statistical suggestions based on real data, one to map out the steps and flows of that workflow and one to test how to improve them, and one to handle the actual tasks assigned as customers flow, and one to make sure all the other supporting files and media are still relevant and used. If theres a problem, the sub agents goes back and asks all the other agents (I found a problem, what do you think) and they all have their own unique view, and the ENGINES bot share "Oh I think the problem stems from this step" and the UNITS bot says "this data point backs it up, its been falling since the last change", and the Solve Problems bot says "Let me make a statistical experiement to undo what we changed and see if it worked". This x the fact that a human based team might take a week or two, with AI is not just 10x faster, but 100-1000x more man power hours as theres no costs needed. All of this can be done with a $20/mo claude subscription. And we're the people to build it. Discuss. People can go play with claude teams rn with a pro sub.

I made a post about how to use agents.md to build agents that build agents to complete an outcome. But someone asked me how to host it (they have an unreliable computer) Modal is a super cheap thing -- they give you $5 (or $30 credits) of AI time at billed at 0.0000001 per second which might last until when the sun dies out. And if it does run out, just pull it to a new instance. 1) Make a bot 2) Host it there 3) ??? 4) Profit Here's a vid if you need to be hand held.

I'm considering making an ultra minimal discord channel to replace Sowork. Sowork is great and fun, but only when the team are all there. So when you're managing a team, it would be cool to see people. Each person on the team gets their own "channel" aka their own room. And an admin / manager can move people around e.g. to make sure people show up. We track how long people are "active" and if people can't show up and sit in their "spot" then its hard to manage people if those people can't even commute to work. This would be a simple "How long have you been active in disco". And theres no auto-AFK like I get in sowork. And its free + unlimited and easy to customize And it lets us enable camera views, which people have told me thats what they need. This seperates customers from the team, and keeps the team online at the same time, and builds a culture. ThoughtS?