Todd
@remotebranch
Followers
2K
Following
10K
Media
196
Statuses
7K
Founding Engineer @hellodigit (acquired) & early @groupon (IPO). Reliable AI Agents Without N8N, ChatGPT, or messy "vibe coding" (even if you're non-technical)
Bentonville, AR
Joined October 2016
two big themes of AI in 2026 will be enterprise agent adoption and scientific acceleration
436
585
6K
Gemini nano banana for images Claude for everything else
Someone educate me. - OpenAI - Claude - Manus - Gemini How do you determine when to use each one?
0
0
1
Your distributed system just failed halfway through a transaction. Now you have: - Money debited but order not placed - Inventory reserved but payment failed - Users angry. Data corrupted. Here's how to fix this forever. A thread on the Saga Pattern:
10
1
2
founder led marketing
Meek mill becomes a ai founder!!!
0
0
3
Without packaging, onboarding, and adoption playbooks, great tools sit unused. The same is true for AI. Leverage comes from frameworks that make it usable, reliable, and tied to real outcomes. In my latest article, I break down: → Why adoption lags behind potential
0
0
1
AI is no longer about flashy demos. It’s about whether teams can actually use it at scale. The surprising thing? The bottleneck isn’t the models. It’s the systems around them. Engineers already know this. Code is never the whole product.
1
0
3
Stop guessing. Start asking your data bigger questions Most teams still scrape through text looking for keywords, hoping to stitch together some insight But that approach misses the real shape of your domain ⇒ With a graph database like Kuzu, you actually map how everything
1
0
1
Why it matters: - Less runtime chaos • Fewer brittle hacks • Cleaner, testable interfaces • Real pipeline reliability You stop babysitting the model. You start building with confidence.
1
0
0
That changes everything. You stop asking the model what it thinks the structure should be. And start telling it: "Here’s the structure. Fill it in. Don’t drift."
1
0
0
BAML is worth a closer look. It’s not just a better format. It’s a contract for AI output. You define the structure up front: strict types, required vs optional fields. No guesswork.
1
0
0
You’ve probably asked ChatGPT (or another LLM) to give you clean JSON. Maybe it looked fine. Until you tried using it: - Wrong fields • Partial values • Markdown wrappers • Hallucinated structures Here’s a better path:
1
0
0
Your AI is Missing Something Big (And It’s Costing You Accuracy) If you're dropping PDFs directly into ChatGPT hoping for clear, actionable insights, you're probably disappointed Generic outputs won’t cut it — especially if you're in a regulated field or niche industry ChatGPT
0
0
1
Use @linear? Try the MCP server with Claude Code Go from idea -> PRD -> tickets in minutes that can be implemented by real devs or agents
0
0
1
Nick Talwar and I are looking for 7 freelancers, consultants, or agency owners Who are looking to deliver better work, faster (without working more hours) So if you: • Want to learn how to use the new AI workflow to deliver better work in half the time (without
0
0
2
Why “simple” no longer cuts it in AI builds You used to get away with sticking to a single cloud Pick your poison—AWS, GCP, Azure—and stay there. The services were close enough. The trade-offs weren’t that serious But today? That’s a good way to get left behind AI models are
0
0
0
Curious about the full list? Check my Substack for 10 books to transform your startup. Read, act, win. What’s your go-to business book? https://t.co/YVMGOaL0IB
0
0
0