Hub
Framework library
All frameworks and models named by guests, grouped by area.
Every model, method or checklist a guest laid out, merged where different people describe the same thing. 3 appear in two or more episodes and have their own page; the rest are listed here with their source.
Starting with agents (8)
- AI maturity curve — Progression in how people use AI: ask, co-work (scheduled tasks), delegation, then teams of shared agents ending in a fully autonomous loop. (presenter · Grok Bot 201 (Oct 2026) ▶ 33:33)
- Five-day starter plan — Practice routine: process-map your own life, write one process down, sort steps into the four buckets, then offer one workflow to a small business. (Vas · Vas (Oct 2026) ▶ 45:41)
- Four-week agent-team rollout — Build the coordinator and about three specialists in week 1, execute without tinkering in week 2, fill gaps in week 3, add routines in week 4. (Billy Howell · Billy Howell (Aug 2026) ▶ 13:37)
- Hire the first employee — To choose a first agent, imagine one free full-time hire for 40 hours a week, build that role, and check the tool can actually do it. (presenter · How Grok Bot Is Going To Make (Oct 2026) ▶ 11:57)
- Levels of automation — Taxonomy of how automated an agent task is: not automated (ask and respond), semi-automated (human approves), fully automated (runs on its own). (presenter · Grok Bot for Admins (Oct 2026) ▶ 18:55)
- Process-first deployment sequence — Map the real process, baseline KPIs, re-engineer, build, deploy, and prove results before and after, typically a four-week audit, four weeks of building and checks at three and six months. (Vas · Vas (Oct 2026) ▶ 10:00)
- Sidekick agents versus background agents — Chat-driven assistants versus agents that do defined work and ping a human; background agents depend on a mapped process. (Vas · Vas (Oct 2026) ▶ 39:20)
- Three sources for mapping a process — Combine interviews, mining systems of record, and existing documentation; the split varies with company size. (Vas · Vas (Oct 2026) ▶ 10:18)
Use cases and business models (6)
- Give an agent a phone — Five-step setup for a calling agent: install the template, create and fund an API key with the calling provider, paste it into the secure vault, set a voice, test by calling yourself. (presenter · Phone calls with Grok Bot (Sep 2026) ▶ 00:27)
- Niche newsletter plus niche directory flywheel — Pair a niche newsletter with a niche directory; the directory earns referral revenue and referral leads become newsletter sponsors. (Billy Howell · Billy Howell (Aug 2026) ▶ 41:12)
- Nine stages of the war on agents — Host’s prediction of how sites and agent platforms settle: block, arms race, market split, trusted middle layer, welcomers win, competition for agents, ads return. Stage 6 is not described. (presenter · The 9 Stages of AI Agent Adopt (Oct 2026) ▶ 18:23)
- Skills of a top forward deployed engineer — Understand how work is done, ship auditable production code against the systems, know the AI layer (models, evals, rollbacks), and communicate with leadership. (Vas · Vas (Oct 2026) ▶ 40:30)
- Strong LinkedIn post structure — Hook, establish the problem, strongest evidence, why it matters, memorable close. (presenter · How I use Grok Bot to go viral (Sep 2026) ▶ 05:38)
- Three camps on agent access — Companies block, accidentally block, or welcome agents depending on how they make money. (presenter · The 9 Stages of AI Agent Adopt (Oct 2026) ▶ 04:43)
Teams of agents (4)
- Coordinator plus specialists team — A coordinating (chief-of-staff) agent owns the mission and routes work to specialist agents, each with one role (research, sales, per-channel writers, designers, platform expert). The human works mainly through the coord (Billy Howell · Billy Howell (Aug 2026) ▶ 11:40)
- Fan out — Send one identical brief to several models in parallel, compare, and merge the best ideas. (presenter · Grok Bot 201 (Oct 2026) ▶ 29:36)
- Regular team audit — Periodically re-run the coordinator audit: review how the team worked, decide which agents to add or remove and what to turn into a routine or skill. (Billy Howell · Billy Howell (Aug 2026) ▶ 31:09)
- Scheduled orchestrator loop (self-improving) — An orchestrator on a recurring routine asks a metrics agent for data, removes the worst performers, has a designer agent make replacements and a developer agent deploy them, then repeats every 24 hours. (presenter · Grok Bot 201 (Oct 2026) ▶ 38:17)
Permissions and human control (3)
- Ladder of trust — Three levels of authority given to an agent, from reading to spending: research, prepare, pay. (presenter · The 9 Stages of AI Agent Adopt (Oct 2026) ▶ 06:47)
- Layered control of agent actions (admin guardrails) — Decide which tools and MCP servers an agent may use, whether it may work on local machines, and which network destinations are reachable; then layer team rules in context, per-action auto-review rules, and an external al (presenter · Grok Bot for Admins (Oct 2026) ▶ 09:19)
- Story-to-post workflow with human gates — Find a story, research and verify, generate angles, human chooses, briefs to writers and designers, human reviews, final package approved and posted manually. (presenter · How I use Grok Bot to go viral (Sep 2026) ▶ 15:43)
Cost and infrastructure (3)
- Building blocks of an agent — What an agent is made of: a model, a sandbox or computer, web search, an action space, connectors to data, skills, secrets, routines and memories, plus observability, feedback loops and security. (Parag Agrawal · Parag Agrawal (Oct 2026) ▶ 27:26)
- Quality, cost and latency as simultaneous constraints — Agent infrastructure is pushed on quality, cost and latency at once; voice agents made latency hard. (Parag Agrawal · Parag Agrawal (Oct 2026) ▶ 24:19)
- Why teams add a dedicated tool layer — Bundled tools limit quality; switching models needs model-agnostic tools; at scale, optimise cost and tokens. Presented as why customers buy the speaker’s search product. (Parag Agrawal · Parag Agrawal (Oct 2026) ▶ 43:56)
Rolling out to a team or company (2)
- Shared specialist agent with one maintainer — One person builds and tunes a specialist agent and shares it; others use it unconfigured and receive later updates. (presenter · Grok Bot 201 (Oct 2026) ▶ 12:19)
- Staged rollout to a team of people — Pilot group while guardrails are set, then provision seats and invite everyone, then share team agents as departmental defaults. (presenter · Grok Bot for Admins (Oct 2026) ▶ 15:28)
Designing an agent (1)
- Anatomy of an agent instruction — What a good instruction to an agent contains: a five-part version (role, goal, inputs, process, quality bar) and a short delegation version (goal, rules, spending or approval limit). (presenter · How I use Grok Bot to go viral (Sep 2026) ▶ 04:25)
Running agents day to day (1)
- Delete, code, agent or human — Sort every process step into: delete, plain code, agent (judgement plus enough data) or human decision (riskiest tasks). (Vas · Vas (Oct 2026) ▶ 18:22)
Sources
- (01)
Masterclass: How FDEs make $1M/yr deploying AI agents
10:00 into “Masterclass: How FDEs make $1M/yr deploying AI agents”
- (02)
- (03)
How I use Grok Bot to go viral on X, LI, and IG (full workflow)
4:25 into “How I use Grok Bot to go viral on X, LI, and IG (full workflow)”
- (04)
- (05)
- (06)
AI Agents Could Drive 10,000x More Web Activity | Former Twitter CEO
24:19 into “AI Agents Could Drive 10,000x More Web Activity | Former Twitter CEO”
- (07)
- (08)
- (09)