“Does it have an opinion?”

LLMs are trained to be persuadable because persuadable is adjacent to useful. So they drift toward the answer you consciously or subconsciously want.

Every CEO has felt an AI agree with them too early and too eagerly. Every engineer knows exactly why it happens.

Naming it is at the foundation of how you graft AI into your business. Not naming it is how you don’t.

That question came from Joe Bradley, CTO of IDC, on AIWA E33.

IDC built a product, IDC Quanta, to resist that drift by design.

Opinionated and Open:

Joe and his team started where they could close a loop.A “partitionable” business with its own revenue line, its own customer and an outcome you can measure.

Then they evolved it from a pre-ChatGPT era business into an AI-first one.

One that wasn’t distracted by AI for AI’s sake. One focused on IDC’s alpha and its core, differentiated value proposition.

One that gets better as models get better. Not one rendered obsolete.

Not a chatbot. An intelligent system that “escaped the confines of its application” to meet people where they are.

Not a sycophant, an intelligence layer with a POV rooted in 60 years of IDC research.

Not dogmatic. “Opinionated and open.” Strong opinions loosely held meets foundational, research driven truths.

Two things can be true at the same time. It depends on the lens you’re using.

This is context and context is king.

What makes IDC Quanta intriguing is the balance. Preserving the integrity of IDC’s positions while letting the intelligence layer be interrogated and stress tested against alternative POVs, including their clients’.

The Mirror or the Moat:

No surprise IDC picked up on this. Their research is about perspective, knowledge and being right.

Most companies won’t catch the drift in the moment. They’ll catch it in the outcome. Eventually.

Regardless of your industry, encoding your POV is where it starts. You have to defend and evolve it.

Encoding it is a build task. Defending it is an operating discipline, because the model’s eagerness to please never ends.

As I wrote in June, an SME can catch the sycophant. That doesn’t scale. You can’t make every SME, analyst or other leader your last line of defense.

Resistance that lives only in your best people is inconsistent, unauditable and walks out the door the day they leave.

An intelligence layer that agrees with whoever is typing isn’t a moat. It’s a mirror with a search index.

And the defense is what makes the compounding worth having. A system that drifts toward every user compounds noise. One that holds its ground compounds judgement, at scale.

Which leaves one question. How do you know it held?

That’s evals, grounded in your context graph.

Encode the POV. Defend it. Evolve it.

The discipline is the floor. The art is the ceiling.

Joe Bradley and I get into all of it in AIWA E33.

Humans + Machines. Never Humans vs. Machines.

---

0:00 Intro: opera, physics and the road to CTO

1:42 CTO vs CIO vs Chief AI Officer

4:40 What "using AI well" actually means

7:34 Where IDC started: the Marketscape

9:43 From internal tool to IDC Quanta

11:01 "I just want a chatbot": chat system vs intelligence layer

14:42 Escaping the confines of the application

15:59 Does it have an opinion?

17:35 Trained to be persuadable, designed to resist

19:49 Opinionated and open: living with multiple truths

21:27 Open source vs closed, saturation and the edge

25:13 Two clocks: the harness and how we build now

29:40 Shared agentic presences and the PM agent

31:20 Who thrives with AI: the ownership model

34:20 The next six months: the data layer and unevenness

37:07 Why we're terrible at predicting the future

39:18 Carve away everything that isn't the sculpture

My man David Shaner had me the moment he described his podcast, [REDACTED], featuring entrepreneurs showing what they've built, the scar tissue they developed and the things they learned along the way.The things you can't teach but you most definitely can learn.The things you can only learn by doing.The feel. The intuition. The taste.The stuff that lets you see around corners.Show me, don't tell me. Not the smoke aka the "everything is perfect" posts on social. The hands in the dirt stuff I can't get enough of.David built a startup from scratch to 34 people, then wound it down to 2.Here's the hook. When the people left, the scaffolding stayed. Every process that created value was still mapped. So what did AI let him do with 2 people, standing on the structure built for 34?Our conversation on AIWA E32 was full of hard hitting, founder-led insights including:1: AI is very good at imitating what humans have already figured out that creates value. It's very bad at finding value from scratch. The scaffolding was the container it could imitate.2: Founders know "deep in your subconscious" that half of what they do every day is extraneous. The hard part isn't knowing. It's forcing yourself to prove it.3: The fastest path to getting your hands dirty is to start small and specific, using AI to solve your own problems, personally or professionally.Link to [REDACTED]: https://www.tweenertimes.com/s/redacted?utm_source=substack&utm_medium=menu

Could Tacnode be the next Databricks? Yes.

Imagine the world before writing. Before books. People had to learn everything from scratch.Thanks to writing, thanks to books, knowledge compounded. Everyone could learn from each other.

Today's AI agents don't yet have their version of writing or books. They operate in isolation, without the shared understanding or context of what previous agents and humans have experienced.

They have to start over. Learn everything from scratch.This is the context gap.Tacnode exists to solve the context gap.

I interviewed Xiaowei Jiang, Tacnode founder + CEO on AI with Alec E31.

What follows are four of his arguments that have stuck with me, worth weighing in your own context.

1: The primary user of enterprise software is shifting from humans to agents.

Databricks. $5.4B revenue run-rate. 65% YoY growth. $134B valuation. 60%+ of the Fortune 500. The most valuable private enterprise software company in the world.That's the bar.In Xiaowei's view, lakehouse architecture became the standard because humans were the primary consumer of data. Databricks and Snowflake built extraordinary capabilities for the analytical workloads they were designed to serve.But humans are slow. A handful of decisions a day with long gaps. Pipelines had time to catch up. Caches had time to refresh.Agents collapse those assumptions. Thousands of decisions a second. No human in the loop. Zero tolerance for conflicting signals.This isn't about replacing analytical platforms. Tacnode sits alongside the lakehouse, purpose-built for real-time agent decisions.

2: Most "AI failures" are not model failures. They are context failures.

Xiaowei broke it into three patterns. AI invents things when context is missing. AI contradicts itself when sources conflict. AI commits confidently to information that is no longer true.The model is rarely the bottleneck. The pipeline behind it is.When an agent reads account balance from one system, transaction velocity from another, and behavior signals from a third, each with its own lag, the model is reasoning from "a snapshot that never existed in the world."In fraud detection and credit underwriting, that fictional snapshot shows up on the P&L.

3: The design starts with what must be true at decision time.

The intuitive approach is to wire together best-in-class components. A great stream processor. A great feature store. A great search engine. Each one correct in isolation. The composite system is not.Guarantees that hold inside one system erode the moment you cross to another.Xiaowei's team inverted the design. Start with what must be true at decision time. Build everything else on top of that contract.This is what first principles looks like below the waterline.

4: Shared context compounds. Isolation does not.

"If a database gives you application shared state, context lake is going to give agents shared memory in a compounding system."

Read that twice.

Every decision an agent makes generates a signal worth keeping. A fraud pattern. A predictive signal. A route cause. In an isolated stack, that learning evaporates with the session. In a Context Lake, it becomes every other agent's capability instantly.The early movers don't just deploy infrastructure. They accumulate institutional knowledge inside it."The cost of waiting is not linear. Every month you wait, the gap grows."

Early movers compound. Late movers start at zero.

Humans + Machines. Never Humans vs. Machines.

The promise of deploying autonomous AI agents at scale hinges on infrastructure and architectural truths. AI isn’t magic or a silver bullet. It most definitely isn’t one size fits all.

What did it accomplish? 

What steps were taken to get there? 

What was the general knowledge vs more specialized, subject matter expertise required?

What logic was introduced and how did it reason along the way?

Jason Dobbs, Founder and CEO of ICTI, is building AI-enabled tokenized issuance for sovereign debt.

He joined me on AI with Alec E30 and shared his truths using the fascinating vocabulary of a space I’m intrigued by. 

I was all ears.

Two of Jason’s quotes are stuck in my head:

1. “‘We don’t do tokenized assets.’ That’s the same as saying, ‘We don’t do the internet.’”
2. “That’s a real big sledgehammer approach, when sometimes you just need a tiny little jewelry hammer going at something.”

Below the waterline. Architecture and infrastructure is where it’s at.

There's a short list of people who shaped my AI journey before I even knew it was a journey. Henrik Werdelin is at the top of that list.

We met in NYC in 2011. A few months before he launched BARK.

I was the furthest thing from technical. He had a gazillion things going on.

But he made time, answered every question and never made me feel like I was behind.

Those conversations in NYC accelerated my learning curve in ways I'm still drawing from today.

Bark now ships millions of dog toys a month. Henrik has since built Audos.com, an AI-native incubator that can launch a business in 15 minutes.

He hasn't slowed down once.

My hope with this episode is simple.

What Henrik did for me back then, this conversation does for all of you now.

A few things he said that I can't stop thinking about:

1️⃣ Most enterprise leaders are about to hit a wall they don't have language for yet.

2️⃣ AI agents are going to free up enormous human energy inside your organization. The question nobody is asking soon enough: what does your organizational design do with it?

3️⃣ The unlock isn't a better AI strategy. It's a harder question: what is your company's strategy in an age of AI?

Humans + Machines. Never Humans vs. Machines.

Creativity is magic. 75% of your creative team’s week is logistics, 25% is actually creating. What happens when you invert that?

Shane Hegde has spent eight years fixing that problem and the implications for modern CMOs are staggering.

Most CMOs have no visibility into what content their org is creating, where it’s going or how it performs.

Shane drew the parallel to a CRO with no pipeline view, something unimaginable at this point.

What Salesforce is to CROs, Air is to CMOs.

Air is an operating system for your visual data. Images, videos, raw files, ad campaigns + brand assets.

Organized, enriched, searchable and connected to performance data so your CMO knows what content exists, where it lives and what it’s doing.

The 75/25 split? Air flips it.

Air backs it up with six months of forward deployed engineers on their dime, not yours.

My man Shane Hegde put his actual cell phone out there, multiple times, on my podcast.

That’s what 15 years of friendship and 8 years of obsession looks like.

He’s expecting your call. I guarantee he will get back to you.
Quickly.

Humans + Machines. Never Humans vs Machines.

s/o Tyler Strand, Ping Ma, Ariel Rubin

Would you believe me if I told you there’s an autonomous AI agent platform that enables anyone with a business idea to outsource to a swarm of AI agents to handle everything except what the founder wants to focus on?Would you believe me if I told you this company has seen its ARR run rate grow from $0 to $100K to $1.5M in a few weeks, with a trajectory that looks less like a hockey stick and more like an elevator shaft?Allow me to introduce you to Polsia and the founder + CEO, Ben Cera.There are 3 reasons Polsia and Ben are the focus of #10.1: Polsia epitomizes how AI is unleashing a sonic boom of entrepreneurial and human potential by dialing up our ability to “focus on making the beer taste better” instead of all the other important but tedious workEntrepreneurs spend 26-50% of their work week on administrative tasks. First time founders have an 18% success rate and the leading reasons startups fail are no market need (42%), running out of funding (29%) and the wrong team (23%) (link). There are 28.5 million solopreneurs in the US, 81% of all small businesses are solo, non-employer firms, yet less than 4% ever break $1M in revenue (link).Ben’s framing is simple and surgical: AI handles 80% of the operational grind. Humans focus on creativity, taste and direction. What used to be inaccessible infrastructure for a bootstrapped one-person company is now table stakes.2: Solo founder + AI stack is a new start-up archetypeOne person. No employees. No engineering team. $0 to $1.5M ARR. Ben never looked at the code. Polsia was built by AI and works in production.Team size as a proxy for ambition is becoming a thing of the past.We’ve moved from AI enabled promises to reality. The future arrived yesterday. 3: What’s the difference between the democratized AI-enabled infrastructure an Entrepreneur vs Intrapreneur has access to? Hint: nothingPolsia is enabling Entrepreneurs to capitalize on the AI Technical Overhang. No different than the way Intrapreneurs in established companies can lead tiger teams, departments or even entire companies by harnessing the technology.Remember the one word (“Goose”) Jack Dorsey left out of his memo (AIWA “The One Thing” #08)? Remember Anthropic’s AI labor market research describing “what AI is theoretically capable of doing versus what’s actually happening in the workplace” aka The Gap is the Game (AIWA “The One Thing” #09)?You know what I mean?Less strategy, “more hands in the dirt” doing.The 18-year-old with a great idea who couldn’t afford a team? They can now build like a funded company. That’s not disruption. That’s democratization.Never run from it. Run at it.

AI with Alec E26 with Greg Boone. 5 reasons to watch?

1: Greg says the quiet part out loud

2: The era of the lone “idea guy or gal” is over, now everyone can build

3: You can’t mandate your way through AI transformation, never underestimate psychological safety

4: AI isn’t an intern, it’s an everything machine, an “IDE for business people”

5: AI native interns crush month long projects in days, are you enabling them aka “feeding the beast?”

BIG s/o to Tracey Franklin at Moderna, leading the integration of technology + human + digital labor.

The way Chase Clark, Lead UX Researcher at Calm, demonstrates how AI is augmenting his workflows with real, tangible examples is outstanding. Personification of “hands in the dirt” use of AI that produces real, measurable results. The kind that all leaders want to generate.

1: Time Compression at Scale: 2 hours → 5 minutes, 2 weeks → 2 days

Not just time saved, it’s all about accelerated learning.

2: AI Prototyping: Previously impossible research is now possible

What happens when teams test 5-10 concepts in the time it took to test one? Innovation cycles compress. Risks get de-risked before production.

3: AI Augmentation: Calibrating collaboration between Humans + Machines

Never underestimate the importance of human taste. AI processes data. Humans are the strategic and creative layer.

While a lot of people are debating AI strategy, Chase and his team are shipping real results.

s / o Lovable, Supabase

Want to see AI-first at scale? Check out Thomas Menard and the rockstar team at L’Oreal showing us how it’s done.Embracing AI to drive “idea to impact” in less than 90 days for a 100,000 person enterprise is impressive.To do it in May 2024 is outrageous. My man Thomas Menard is a “good pirate” who personifies the “ship fast, learn, iterate” approach. No analysis paralysis. No tech for tech’s sake. No science projects. Just a ruthless commitment to augmenting people to create business value.Intelligent systems that abstract complexity so everybody, engineers and non-engineers, can create value without becoming AI experts.What’s next?From Gen-AI-as-a-Service to a proper Agentic AI platform. Realizing the dual benefits of AI infrastructure while balancing the buy vs build equation accordingly.Think context engineering, AI-first architecture, platform design and continuous improvement.1% better every day. This is the way.s/o Harrison Chase, LangChain + GCP