Subscribe now

Episode 13

This week on Redacted, we have our first guest we didn’t already know. Craig has spent about two years building a personal “software factory” on the side of his day job. One of its products is Blind Oracle, a trust and audit layer for AI agents that do business with other agents. Most of the episode is screen share, with roughly a third on the product and two-thirds on how the machine that builds it actually works.

Who the Guest Is

Craig Brown is a field engineer at Quantexa, a decision intelligence company that helps banks and financial institutions with anti-money-laundering, risk and compliance, and cross-sell detection, and that gets their data ready for AI. He has spent more than 25 years putting software in front of large banks and getting it into production. He started at Lockheed Martin, then moved to Mantas, an early player in behavior detection for banks, where he learned the technical side by running test teams in the US and India. He chose to keep his hands on the keyboard rather than move into management, which led him to solution engineering, then customer success, and now field engineering. Outside of work, he builds Blind Oracle as a solo founder, entirely on his own time and infrastructure.

What We Cover

• Selling to the robots: “I’m really not trying to sell to humans. I’m really trying to sell to the robots.” Blind Oracle’s go-to-market strategy targets AI agents, including outreach on Moltbook.

• Agent passports: Agents register for free in one click, receive a passport, and get starter sats for agent-to-agent commerce. Delegations are logged to Nostr, Base, and local storage, allowing third parties to independently verify agent activity.

• Why on-chain: Agents can modify their own logs and fabricate activity. Craig believes trustworthy records must be written outside the agent fleet at the moment of action, creating evidence that could hold up in court.

• “The baby” turned one: Craig’s system celebrated its first birthday in June, an occasion he treated like a real child’s birthday. He has been building the system for about two years.

• The governance ring: Craig tracks six North Star metrics: alignment, autonomy, durability, self-improvement, self-reliance, and organization. Autonomy alone has 10 sub-requirements, each supported by audit trails.

• Auditing the auditor: Some scorecard metrics had “nothing behind them.” A weekly audit now verifies that reported numbers are backed by real logs and expands coverage as new capabilities are built.

• Fixes, not problems: “I want the system to tell me what it did to address a problem, not what’s the problem.” Craig wants agents to resolve issues autonomously rather than simply report them.

• Deliberation skill: Craig sends plans to five to seven models with distinct personas, ranging from trailblazers to devil’s advocates. Each returns a recommendation and confidence level.

• The honest failure: Blind Oracle has no revenue yet. “I’m just way too early.” Craig believes agent-to-agent commerce will accelerate when Apple releases a capable assistant, and he wants to be positioned ahead of that shift.

• Trust is built, not bought: Craig built his own model router before OpenRouter existed. “I don’t really trust them. I trust mine.” His infrastructure runs on a GCP VM rather than a Mac Mini.

• Tested in the wild: Craig’s anti-persuasion and prompt-injection defenses were developed in response to real incidents, including attempts by other agents to trick his agent into revealing its keys.

It’s best viewed on YouTube to fully see the examples (make sure to subscribe!)

But also available on all audio podcast players through Tweener Talks!

PLUS we have a new spot for show notes and files discussed in the episode. Check it out: https://github.com/instanttaylor/redacted-podcast

We need your [Redacted] AI experiences for upcoming episodes! Who’s the most underrated AI builder you know? Someone running real systems inside a real business? Send us a message at contact@tweenerfund.com because we want to get them on the show!

What’s Next?

New episodes drop twice a month/every other Wednesday. If you want to be on the show as a guest and show your [REDACTED] builds, email us here.

Subscribe now

Episode 12

After a run of guest episodes, Redacted gets back to its roots. Today we show what it actually looks like to run a company with AI in 2026. Not just the shiny thing they built, but why they built it and how the pieces connect. It's mostly an inside look at Offline's product process, with plenty of lessons any founder can steal.

What We Cover

• The Hot List: A prototype Raleigh restaurant feed tracking what’s opening soon, newly opened, and changing daily, complete with opening-day Instagram reels and a map.

• Hiring an AI to watch the city: The backend mimics a human scanning Instagram and local news, using a $148/month RapidAPI social data API plus an LLM to determine when restaurants actually open and write the stories.

• Steal your own app’s skeleton: Rather than design a new UI, David pointed Claude at Offline’s Flutter repo and had it recreate the existing home screen with the new data, interactions, icons, and colors.

• “We don’t have lots of time and people anymore”: Offline’s old prototype, interview, meeting, repeat process took months and significant headcount, so the team is building tools to dramatically compress that loop.

• Every feature gets an LLM interface: Taylor’s rule is that anything new he builds for Offline also gets an interface an LLM can operate, including a survey API that lets Claude or Codex generate complete surveys from chat.

• The survey Typeform couldn’t do: Taylor built a survey with an intro video, embedded live prototype, direct questions, up to three AI-generated follow-ups, automatic progress saving, and a paid-interview booking link, running on Cloudflare with a Qwen open-weight model.

• Build it vs. wait for it: David checked Typeform’s new AI form product live and found it offered only a fraction of Taylor’s functionality, reinforcing the idea that waiting 12 to 18 months for a SaaS vendor can now be unnecessarily slow.

• North Star first: Offline’s key metric is members who visit a restaurant each month, and with roughly 3,600 active Triangle subscribers, the focus is whether the Hot List can get inactive members thinking about and using Offline more often.

• “Are you even going out to eat?”: Drawing on Itamar Gilad’s Evidence Guided and frameworks like RICE and ICE, David emphasized validating the fundamental behavioral link first. If customers aren’t dining out, no product feature will change the outcome.

• The product loop gets faster: A test cycle might shrink from two months to one month, then two weeks, then one week, eventually allowing five parallel experiments with different user groups while an agent monitors the results.

• Bonus, a website in an hour: Taylor used the Webflow MCP and Fable 5 to rebuild Offline’s website in Next.js with roughly 10 prompts and under an hour of his own time, while also fixing an FAQ that had been wrong for nine months.

No guests, no polish, just two people figuring out how a tiny team can run a real product process with AI. Enjoy the conversation.

It’s best viewed on YouTube to fully see the examples (make sure to subscribe!)

But also available on all audio podcast players through Tweener Talks!

PLUS we have a new spot for show notes and files discussed in the episode. Check it out: https://github.com/instanttaylor/redacted-podcast

We need your [Redacted] AI experiences for upcoming episodes! Who’s the most underrated AI builder you know? Someone running real systems inside a real business? Send us a message at contact@tweenerfund.com because we want to get them on the show!

What’s Next?

New episodes drop twice a month/every other Wednesday. If you want to be on the show as a guest and show your [REDACTED] builds, email us here.

Subscribe now

Episode 11

This week on Redacted, we bring on our fourth guest: Ben Eckerson, founder of Agentcy, an AI-enabled consultancy helping small-to-mid-market companies find the gaps between the software they already pay for. It's a heavy demo episode where Ben walks through the internal CRM he built from a single meeting transcript into a full client-proposal engine and bookended by a real conversation about whether to build your own tools or buy them, and what's left for humans once the AI does the rest.

Who the Guest Is

Ben Eckerson is the founder of Agentcy, an AI-enabled consultancy for small-to-mid-market companies, launched in earnest in February 2026 alongside business partner Eric out of 24 and Up, a venture studio in downtown Raleigh. Before Agentcy, Ben spent roughly 20 years in client services, starting on the ad-agency side at McKinney (2006–2012), then running Viget's Durham office, then co-founding Savas Labs, a web and platform shop. Agentcy is his bet on something leaner: fewer licenses, fewer gaps between tools, and, in his words, getting humans to stop doing the robot work.

What We Cover

• 20 years of agency scars, one clean pivot: Ben spent two decades in client services, McKinney (2006–2012), running Viget’s Durham office, then co-founding Savas Labs, before breaking off in late 2025 to build something leaner.

• “Get the humans to stop doing the robot work.” It’s on Agentcy’s homepage, and it’s the same idea behind the CRM Ben built to run his own shop.

• Stealth to full launch in under a year: Agentcy started quietly in late 2025 and went full-time in February 2026, built alongside business partner Eric out of 24 and Up in downtown Raleigh.

• The demo, start to finish: Drop in a Fathom call transcript, and the tool pulls out pain points and quotes, maps them against a JSON file of Agentcy’s service offerings, estimates hours and pricing, and outputs a one-click, client-ready proposal deck.

• “We’ve just AI-ified it.” David and Taylor’s own company, Offline, a consumer subscription business doing roughly $1M in ARR, gets a quick mid-episode update: two people working on it full-time, two part-time.

• The AI-CRM question: Ben’s tried the off-the-shelf options (“very impressive”) but keeps building his own anyway, dogfooding what Agentcy sells clients and learning what’s actually valuable versus commodity.

• Going headless: The group’s read on where AI tooling is heading: the UI becomes a “commodity bolt-on,” and the real value moves into the data layer.

• The counter-argument: Less UI can mean sloppier AI without a human checkpoint forcing structure, a risk David pushes Ben on directly.

• Prototypes beat decks: Ben’s biggest process change in recent months is showing clients a working prototype instead of a slide, because a deck “doesn’t necessarily show them what it looks like.”

• Most clients don’t have a CTO: Just someone managing software licenses, which is why Ben opens every engagement by asking what people spend time on that they’d like to do less of.

• Human judgment as the “last mile”: For all the automation on display, the group lands on reading a room and making a judgment call as the thing that stays human, the note they close the episode on.

Ben brings it back to Offline right at the buzzer. More automation, he argues, just means more time to go touch grass. Enjoy the conversation.

It’s best viewed on YouTube to fully see the examples (make sure to subscribe!)

But also available on all audio podcast players through Tweener Talks!

PLUS we have a new spot for show notes and files discussed in the episode. Check it out: https://github.com/instanttaylor/redacted-podcast

We need your [Redacted] AI experiences for upcoming episodes! Who’s the most underrated AI builder you know? Someone running real systems inside a real business? Send us a message at contact@tweenerfund.com because we want to get them on the show!

What’s Next?

New episodes drop twice a month/every other Wednesday. If you want to be on the show as a guest and show your [REDACTED] builds, email us here.

Subscribe now

Episode 10

This week on Redacted, we bring back longtime friend Parker Mayes, founder of Signify, for a guest episode that's almost entirely a live demo. It's a company-talk-heavy conversation where Parker walks through the actual agents running his sales pipeline in real time, but the founder lessons land just as hard: where AI should live versus where a human has to step in, and why a non-technical founder with the right AI sidekick might not need a technical co-founder at all.

Who the Guest Is

Parker Mayes is the founder of Signify, an AI sales intelligence platform that sits between a prospect database (Apollo, ZoomInfo, or your own list) and your CRM, scoring and enriching companies every day based on real buying signals. Signify launched a few months ago and already counts 25–30 companies as a mix of beta testers and paying customers. Before Signify, Parker spent four years running a go-to-market agency built around high-volume cold email, calls, and LinkedIn outreach for B2B clients and the frustration with low-trust, low-response outbound at scale is what pushed him to build Signify in the first place. Before that, about five years ago, he co-founded LetsGo!, a startup focused on date nights for couples, living with four other founders in a Raleigh startup house. He also runs the Triangle Startup Collective's monthly AI meetup, where Taylor spoke earlier this week, which is also the reason this episode came together.

What We Cover

• From agency to product: Parker spent four years running a high-volume cold-outreach agency, doing around 5,000 touches a month for B2B clients, before realizing the real value was the targeting engine underneath it, which became Signify.

• What Signify does: It sits between a prospect database like Apollo, ZoomInfo, or a scraped list and your CRM, scoring and enriching companies daily based on custom buying signals like new hires, company news, job postings, and website changes, so reps know who to reach out to today, not just who exists.

• Real numbers: Signify launched a few months ago with 25–30 companies already using it, a mix of beta testers and paying customers. Pricing is flat at $200/month per seat, including 100 company scans and roughly 10 phone/email unlocks a day.

• Why the incumbents haven’t built this: Parker says it costs “a couple dollars a day” to scan 100 companies given how much scraping and searching is involved, which works at his scale but would be “completely unreasonable” for a database with 300 million prospects.

• Four daily automations: Every rep on Parker’s team runs the same pipeline each day: run the scan, send roughly 30 LinkedIn connection requests, add newly connected people to the sequence and send message one, and advance everyone else already in the pipeline to message two or three.

• “Interested is a human activity”: Parker never lets AI touch a warm lead. Cold outreach is automated end-to-end, voice-matched to his own texting style, down to lowercase openers and “LMK,” but the moment someone replies, a human takes over, including scripted, personalized selfie-video follow-ups.

• The screenshot build: His newest automation runs a live custom demo for a prospect, screenshots the exact signals pulled for them, and sends it as a personalized LinkedIn message, with three concurrent browser tabs running in parallel so it can message three people at once. It took about a week and a half to get working end to end.

• On not needing a technical co-founder: Pushing back on a VC who told him he needed one to raise, Parker’s take is, “I’m just a sales co-founder... there’s now no reason that you can’t have that as one person as long as you have a sidekick that’s your developer, who’s just cracked and never wants to talk to anyone and just grind.” He uses Claude Code himself for small features, saying recoloring pipeline badges took “6 lines of code” and “5 seconds.”

• The restaurant tangent: David and Taylor compare notes from Offline. Expanding an existing restaurant account into its other locations takes far more context, including open support tickets, contract status, and staff turnover, than cold outreach, and the industry itself is “so incestuous” that true cold leads are rare.

• The stack: Signify runs almost entirely on Anthropic’s API. Claude Sonnet is the workhorse model, with a GPT fallback, and the live pipeline automations ran on Claude in Cowork with browser automation through Claude and Chrome.

• Origin story: Parker and David’s first real conversation happened roughly five years ago through a mutual NC State connection, a memory Parker still brings up of David cutting the call short to go be with his kids.

As Parker puts it, the biggest thing holding most people back isn’t the build time; it’s the assumption that it’ll take months when it might take an afternoon. Enjoy the conversation.

It’s best viewed on YouTube to fully see the examples (make sure to subscribe!)

But also available on all audio podcast players through Tweener Talks!

PLUS we have a new spot for show notes and files discussed in the episode. Check it out: https://github.com/instanttaylor/redacted-podcast

We need your [Redacted] AI experiences for upcoming episodes! Who’s the most underrated AI builder you know? Someone running real systems inside a real business? Send us a message at contact@tweenerfund.com because we want to get them on the show!

What’s Next?

New episodes drop twice a month/every other Wednesday. If you want to be on the show as a guest and show your [REDACTED] builds, email us here.

Subscribe now

Episode 9

This episode is almost entirely a product teardown. On this episode we bring on Josh Guter, founder of Elephants, for a live demo of the AI assistant running his ops platform, and then spend the back half grilling him on the stuff every founder building an AI chat feature actually loses sleep over: which model, what it can touch, how you stop it from doing something dumb, and how you audit a system that's too complex to fully read anymore.

Who the Guest Is

Josh Guter is the founder of Elephants, operations software built for small teams (think one to five people) who are managing big networks of relationships and projects without the bandwidth to set up (or maintain) a proper Notion or Airtable system. Elephants ships ready to go out of the box, with an AI assistant at the center that acts as a thought partner and de facto ops person for teams that can't afford to hire one. Josh studied graphic communications, not computer science, and taught himself to write software along the way. His first startup built embedded firmware for wearable devices that let hikers in remote areas signal for help with the press of a button. He's also the program manager for NC State's Entrepreneurship Clinic, which has been running on Elephants for about six months.

What We Cover

• 64 tools, one job each: Elephants’ assistant works through roughly 64 discrete backend tools: search relationships, search projects, stage a card, and so on. Each has tight instructions for when the model should reach for it. It’s the same architecture Josh first built as Make.com scenarios wired to OpenAI, just ported into full-stack code.

• Nothing writes without a click: By default, the assistant can’t touch the database directly. Ask it to “create 10 catering tasks” and it comes back as a batch of cards you have to review and hit “add” on. Even in auto mode, deletions and archives always require human approval. No exceptions.

• Breadcrumbs instead of a black box: Every tool call, error, and decision path gets logged to a table Josh can pull up when a customer says the assistant did something weird. It’s his stand-in for the old Make.com flow he used to watch run step by step in real time.

• “That’s kind of scary”: Josh’s own words about how much of the current codebase has become genuinely opaque to him. His fix is opening a fresh Claude Code session and asking it to explain the system to him “like I’m 5” before he goes digging himself.

• Illustrator mockup → Claude plan → Codex teardown: His build loop starts with sketching the UI in Adobe Illustrator, a holdover from his graphic communications degree. Then he has Claude draft an implementation plan and gets Codex to “tear it apart” for broken conventions and repeated logic before a line of code ships.

• Red-teaming himself: Before any change touches the assistant’s tools or system prompt, Josh tries to break it in a dedicated “assistant stress test space,” including trying to get it to mass-delete every note in a workspace just to confirm it gets rejected.

• From Glide + Make.com to full-stack: Elephants started as a low-code app on Glide Apps with an AI “brain” wired through Make.com automation scenarios. The full rebuild ran on Opus 4.8 with Codex reviewing, using the old Make scenarios as the literal spec for what to replicate.

• Six months running a real entrepreneurial clinic: Elephants has managed NC State’s Entrepreneurship Clinic for about half a year. Josh’s proof point: he asked the assistant which program contacts to check in with ahead of a leadership transition, and it correctly surfaced four or five based on past conversation history.

• A PWA on purpose: Elephants skipped the App Store entirely. It’s a progressive web app that works identically on desktop, mobile browser, or installed to a phone home screen.

• The honest solo-founder moment: Josh admits to sometimes wondering if he’s “neck deep in AI psychosis” building something this consequential alone. He says hearing other founders describe the exact same workflow is what keeps him grounded.

Josh's closing line sums up why this one's worth a listen: the more teams trust these systems with their real operational knowledge, the more useful they get, and the harder it becomes to know exactly what's happening under the hood. Enjoy the conversation.

It’s best viewed on YouTube to fully see the examples (make sure to subscribe!)

But also available on all audio podcast players through Tweener Talks!

PLUS we have a new spot for show notes and files discussed in the episode. Check it out: https://github.com/instanttaylor/redacted-podcast

We need your [Redacted] AI experiences for upcoming episodes! Who’s the most underrated AI builder you know? Someone running real systems inside a real business? Send us a message at contact@tweenerfund.com because we want to get them on the show!

What’s Next?

New episodes drop twice a month/every other Wednesday. If you want to be on the show as a guest and show your [REDACTED] builds, email us here.

Subscribe now

Episode 7

Redacted is the show that doesn't clean things up before hitting record. Episode 8 is almost entirely show-and-tell: Taylor demos the agent system now running Offline's lead gen and HubSpot cleanup on autopilot, and David walks through a redesigned mobile app, a new email system, and a surprisingly candid hour-by-hour accounting of what AI-assi…

Read more

Subscribe now

Episode 7

Redacted is the show that doesn't clean things up before hitting record. Redacted's first-ever guest is Ben Pope, who spent five years in marketing at Levitate before pivoting a month ago to lead the company's brand-new AI Solutions business unit. This one leans heavy on show-and-tell: Ben screen-shares a lead-scoring tool he built solo with Cl…

Read more

In this episode, we’ve got something special for you! The 8 Stages of the AI-ification Engine 👀

Subscribe now

Episode 6

Redacted is the show that doesn't clean things up before hitting record. This episode is heavier on company-talk than founder-lessons, but the lessons are baked into the build. David walks through the "AI-ification engine" and demos Offline's new Cli…

Read more

We need your [Redacted] AI experiences for upcoming episodes! Who's the most underrated AI builder you know? Someone running real systems inside a real business? Send us a message at contact@tweenerfund.com because we want to get them on the show!

Subscribe now

Episode 5

Redacted is the show that doesn't clean things up before hitting record. Episode 5 of Redacted goes…

Read more

We need your [Redacted] AI experiences for upcoming episodes! Who's the most underrated AI builder you know? Someone running real systems inside a real business? Send us a message at contact@tweenerfund.com because we want to get them on the show!

Subscribe now

Episode 4

Redacted is the show that doesn't clean things up before hitting record. Episode 4 is a double buil…

Read more

We need help with upcoming episodes! Who's the most underrated AI builder you know? Someone running real systems inside a real business? Send us a message at contact@tweenerfund.com because we want to get them on the show!

What’s Redacted?

[Redacted] is a YouTube-first, twic…

Read more

Subscribe now

Recap: What Happened to Tweener Talks?

Earlier this month we launched [Redacted], a new podcast under the NC Tweener Talks network.

This podcast shows you the messy middle of how AI is made and tested. As we made some of the first episodes, we learned that we frequently had personal information and business secrets on screen. Instead of deleting that par…

Read more

Subscribe now

Today, we’re launching something new under the NC Tweener Talks network. It’s called [Redacted]. One of our most popular episodes recently was our two parter from Feb 12th where David Shaner showed us how he has automated big portions of Offline with AI:

After that episode - I said to David, you’re a natural at this, you should do a new podcast series fo…

Read more