https://www.codestory.co/episodes/s13-bonus-the-ai-dba-shift-automating-database-reliability-performance-with-itamar-syn-hershko-founder-ceo-of-neverblink/

Itamar Syn-Hershko has been a tech guy since he was a wee lad - IE he was teaching his kindergarten teachers how to use their computers. He learned how to “do stuff” tech wise through playing games back in the day. Back in the day he was one of the original developers on a document based database, as well as building a consulting company. Outside of tech, he is the father of 5 kids, all of them 7 and under... so his hands are quite full. That said, he does enjoy learning new things and traveling, when he has the time.

Itamar founded a boutique consulting agency, helping businesses succeed in data related projects. Like most smart agencies, they built tools to help them to do their jobs better. One of these tools became highly popular, and after 1.5 years of selling this tool to customers, they hit a $1m in annual recurring revenue.

This is the creation story of NeverBlink AI.

Links

https://neverblink.ai/

https://bigdataboutique.com/

https://www.linkedin.com/in/itamar-syn-hershko

Current Sponsors:

• Tiger Data

• Protected Harbor

• Render

• Fitnexa

• Perplexity

• Entelligence

Steven Walchek is originally from the Bay Area in California, but lives in Denver now. He’s been there for 8 years with his family. He comes from a family of risk takers and entrepreneurs - for example, his Dad was a tech guy who did his thing and had success in Silicon Valley. Outside of tech he is a family man, married with kids. He enjoys doing anything he can with them, as they are his “why”. He plays hockey, coaches hockey, and rides dirt bikes and mountain bikes.

Steven and his team built an application security platform, and ended up hitting every branch on the way down from the startup tree. When they sat down to figure out what they needed to do next, they combined their deep security tooling with a multi-modal platform, to enable complete AI governance in one central hub for regulated companies.

This is the creation story of Liminal.

Sponsors

Tiger Data

Protected Harbor

Render

Links

https://www.liminal.ai/

https://www.linkedin.com/in/swalchek/

Brief lets your whole team, and their coding agents, work with the judgment of a great product leader. Founder Drew Dillon built Brief after 20 years leading product orgs (and years as a fractional CPO for companies from two founders in a garage to 2,000-person late-stage startups), because he believes product judgment should scale the way code now does.

So… why should you use Brief?

An Opinionated Product Graph, Not Enterprise Search

Brief doesn't just ingest your tools and hope retrieval works. A set of opinionated agents distill what matters (personas, decisions, features, research, competitive signal) into a product graph, then link it all so the answer to "why are we building this?" is one hop away, not forty search results.

Completely Ambient. Not Another Dashboard.

Brief behaves like a peer, not a portal. Ask it in Slack or Teams, hit it from your CLI, or let your coding agents talk to it directly over MCP. Many engineers on Brief teams never open the website at all. Their agents just ask Brief and skip the meeting.

Decisions Your Agents Actually Follow

Every trade-off your team makes gets captured in a living decision ledger, with provenance. Connect Brief to Claude Code or Codex and those decisions are enforced, so nobody (human or agent) ships something the team decided against three weeks ago.

Feed Your Coding Agents Real Context

Agents are only as good as the context you give them. Brief's graph delivers just the relevant slice, which means better output and lower token spend. New research from the Brief team shows the opinionated graph beats open-ended memory systems on token utilization, execution speed, and rework.

Your Process, Enforced Automatically

Build your own agents on top of Brief: require acceptance criteria before a ticket moves, check every PRD against your template, review work as it flows through Linear or Jira. Consistency without the process police.

Trusted From Two-Person Startups to the Enterprise

Technical founders use Brief as their co-founder in product. Enterprise product orgs use it to cut the alignment tax across thousands of people. One customer account has grown from 10 to 56 seats.

Ship the Right Thing. For Free.

Your agents are already fast. Give them judgment.

Start your free trial of Brief today.

https://www.codestory.co/episodes/s13-e1-the-token-winter-is-coming-why-code-inflation-fails-to-ship-real-business-value-with-dave-garcia-founder-co-ceo-of-pensero/

Dave Garcia is originally from Barcelona, and recently moved stateside, to the East Coast. He started coding when he was 7 years old, on a Commodore 64, in BASIC language - and when he programmed an infinite loop, he was hooked. He’s been curious about machines, and fixing them - even when they weren’t broken. Outside of tech, he is married with a young family. He enjoys playing video games - mostly Dad games these days - and loves to do anything on the water, like kite surfing.

A few years ago, Dave decided to take some time off to spend it with his family. When the AI boom started to happen, suddenly the technology became ubiquitous, where AI was changing the life of generations around him. After some investigation, he decided wanted to build something world changing, relevant to the world the knows - which is within the tech mgmt. world.

This is the creation story of Pensero.

Sponsors

• Tiger Data

• Protected Harbor

• Render

Links

• https://pensero.ai/

• https://www.linkedin.com/in/davebcn

Key Takeaways

• The “Token Winter” Threat: Uncontrolled AI usage, retries, agent loops, and prompt inflation create a false sense of progress while quietly exploding enterprise API budgets without driving business value.

• Activity Does Not Equal Business Impact: High model usage or sheer volume of generated code masks inefficiencies, creating “code inflation” that increases code complexity and rework costs rather than shipping features.

• The Necessity of Agentic ROI and Measurement: Engineering organizations must move beyond raw token counts to measure actual outcomes, evaluating which AI workflows reduce rework and improve delivery speed versus those that add noise.

• Model Selection as an Engineering Strategy: Companies must avoid using frontier models for simple tasks (the “Lamborghini for grocery shopping” mistake) and instead match tasks to cost-effective, specialized LLMs.

• Shifting from Uncontrolled Experimentation to Governance: Long-term success in AI adoption depends on engineering visibility and financial feedback loops that attribute token spend directly to business value.

Frequently Asked Questions

What is the “Token Winter,” and why does Dave Garcia predict it?

The “Token Winter” refers to an impending financial reckoning for engineering teams as unmanaged, subsidized token spend explodes into a major enterprise cost center—mirroring the runaway cloud bills of a decade ago.

What is Pensero, and what problem does it solve for engineering leaders?

Pensero provides delivery intelligence and AI performance analytics, giving CTOs, CEOs, and engineering managers objective visibility into how AI usage translates into real business value, code quality, and ROI.

What is “code inflation,” and how does AI usage contribute to it?

Code inflation occurs when developers use AI to generate massive volumes of code without proper architecture, leading to overly complex, unmaintainable repositories that require significant human review and rework to fix.

How can companies prevent “the illusion of cheap AI” in production?

Organizations must establish clear attribution models and operational feedback loops, treating token consumption as an investment decision tied to key performance indicators like PR merge rates and cycle time reduction.

How should engineering managers evaluate whether an AI workflow adds real value?

Managers should measure outcomes rather than activity—focusing on whether an AI agent reduces cycle time, eliminates technical debt, or improves delivery quality, rather than tracking how many prompts or API calls a team uses.

What role does model selection play in controlling enterprise AI costs?

Instead of defaulting to top-tier frontier LLMs for every query, teams should strategically route prompts—reserving expensive, highly capable models for complex reasoning and deploying lightweight, cheaper models for routine tasks.

Code Story Ambassadors

We’ve officially launched our Code Story Ambassador Program, giving you a direct way to help us expand our community. By sharing your unique referral link with colleagues, friends, and fellow tech enthusiasts to get them subscribed, you can play a vital role in our growth while unlocking exclusive perks along the way.

To show our immense appreciation for spreading the word, we’ve put together four reward tiers based on the number of new subscribers you bring in:

• Tier 1 (5 referrals): Receive a direct Personal Thank You from me for stepping up to support the show.

• Tier 2 (15 referrals): Get a public Episode Thank You shoutout credited right in an upcoming podcast episode.

• Tier 3 (30 referrals): Represent the show in style with an exclusive Code Story T-shirt.

• Tier 4 (50 referrals): Level up your wardrobe with our premium Code Story Hoodie.

Share the Show

https://www.codestory.co/episodes/s12-bonus-rickards-deterministic-return-converting-scattered-data-into-autonomous-production-grade-apps-with-rickard-hansson-founder-ceo-of-gainable/

We have a special return episode, by our good friend Rickard Hansson. Rickard joined us previously on the podcast in Season 8to tell the creation story of Weavy - collaboration infrastructure for serious builds. Today, he makes a follow up visit to tell us all about Gainable, his new project - which removes data and engineering from being the middle man, and enables your team to build the apps they need now.

Questions;

• Last time we talked in Season 8, you were building Weavy. Whats happened since we last talked with that company?

• Tell me about Gainable - give me the pitch there, and tell me why this is the right approach to using AI.

• Most AI builders wire straight to a frontier model and wait for the next release to fix the gaps. I didn’t. Where does the model actually sit in Gainable product, and why only there?

• Why is an app factory that is deterministic important? Dig into that.

• You use the term “free-range coding”.. what does this mean? Unpack the phrase for us.

• You point out that tokens still appear to be heavily subsidized to me. What do you mean by that, and what happens to all these AI products when that ends?

• We’ve all read the headlines - Fable 5 got switched off by the government for 18 days. Why do you see this as a turning point, not a footnote?

• You suspect flat subscriptions for the top models are done, and it all drifts to credit-based. What signal are you seeing that tell s you this?

• If the model is a commodity everyone rents, where’s the moat?

• What is next for Gainable, and how can someone get started using the platform?

Sponsors

• Unblocked

• TECH Domains

• Mezmo

• Braingrid.ai

Links

• https://www.gainable.dev/

• https://www.weavy.com/

• https://www.linkedin.com/in/rickardh/

https://www.codestory.co/episodes/s12-e30-wastewater-guardians-automating-biology-to-protect-clean-water-with-virginia-szepietowski-co-founder-of-nyad-ai/

Virginia Szepietowski grew up in the UK outside of London, and now lives in Alabama. She’s had a winding path to her current venture, including body building, triathlons, law, and entrepreneurship. She comes from a family of entrepreneurs, who are deeply ambitious, tenacious, and deeply humble. She finds the feeling of a deep safety net from her family, and she pursues her adventures. Outside of tech, she is married to her now co-founder. She is still a competitive body builder, and likes to push herself to the limit.

Through a series of life events, Virginia got interested in water treatment. She started discovering the world of wastewater operators, and the fact that they were the last line of defense before toxic wastewater moved into our waterways (rivers and such). Using AI, her and her team started to build a platform for these operators to quickly detect organisms in these water streams.

This is the creation story of Nyad AI.

Sponsors

• Unblocked

• TECH Domains

• Mezmo

• Braingrid.ai

Links

https://nyad.ai/

• https://www.linkedin.com/in/virginia-szepietowski/

And that’s where Protected Harbor comes in.

Application-aware infrastructure, owned with accountability

Application-Aware Infrastructure (AAI) is Protected Harbor’s philosophy - that infrastructure should be designed around the way applications actually operate, not forced into a one-size-fits-all environment.

Traditional infrastructure treats every workload the same, but AAI takes a different approach. Protected Harbor begins by understanding how an application behaves and then engineer, manage, and continuously optimize the infrastructure specifically for those needs. The result is better performance, greater reliability, stronger security, and more predictable costs.

These are the five core pillars of AAI:

• Application Understanding: Protected Harbor analyze’s application behavior, database activity, workload characteristics, and usage patterns to ensure infrastructure is aligned with how the application actually runs.

• Purpose-Built Design: Every environment is right-sized and architected for optimal performance, security, resilience, and scalability — without unnecessary complexity or wasted resources.

• Unified Outcomes: One trusted, accountable partner is responsible for the entire environment, eliminating finger-pointing and ensuring faster resolution when issues arise.

• Continuous Optimization: Infrastructure is monitored 24/7 using customized dashboards, with proactive tuning and ongoing improvements to maximize uptime and efficiency.

• Cost Engineering: Costs are designed to be predictable, transparent, and aligned with actual business needs rather than fluctuating cloud consumption.

Talk with Protected Harbor

If infrastructure is becoming a larger operational or business concern, a conversation can help clarify where pressure is building and what improvements would make the biggest difference.

Request an infrastructure review to identify hidden risk and improvement opportunities.

https://www.codestory.co/episodes/s12-bonus-the-perishable-supply-chain-crisis-why-generic-erps-fail-fresh-food-logistics-and-how-ai-agents-are-transforming-error-free-order-intake-with-sid-dixit-chief-technology-officer-at-itradenetwork/

Sid Dixit is originally from central India, and came to the states for college. He is a technologist and builder at heart, serving in leadership roles across major companies. He has built and managed a fleet of satellites, built robots at Amazon, worked at Microsoft on surface tablets, and finally, at Google working on Android. Outside of tech in lives in the Bay Area with his wife and kids. He loves water sports, especially sailing. He spent 10 years in San Diego, and stumbled on the sport.

Sid’s current company started in 1999, and was acquired in 2010. A few years ago, Sid joined the company, at a time when the company was wanting to rebuild its network from the ground up - starting with a powerful index.

This is Sid’s creation story at iTradeNetwork.

Sponsors

• Unblocked

• TECH Domains

• Mezmo

• Braingrid.ai

Links

https://www.itradenetwork.com/

• https://www.linkedin.com/in/siddharthdixit/

https://codestory.co/podcast/ai-throughput-illusion-expensive-models-fail-measure-real-engineering-output-emilie-schario-head-of-product-kilo/

The AI Throughput Illusion: Why Splurging on Expensive Models Fails to Ship Code and How to Measure Real Engineering Output with Emilie Schario, Co-Founder & Head of Product & Engineering at Kilo Code

Emilie Schario grew up in New Jersey, outside of Newark, and attended college in the state. Currently, she lives in Columbus, Georgia, outside of Atlanta. She mentions she got into technology so she could easily follow her husband’s career geographically, and has much success in the industry. Outside of tech, she is married with 3 boys (all 5 and under)… so there is a lot of wrestling in her household. She admits she is often quoted staying she does three things in her life – work, parenting, and if she is lucky, attends CrossFit 3 times a week. In fact, she finds a great sense of community in that world, and brings her kids with her to cheer her on.

A year and a half ago, Emilie’s current venture was started, to build the open source orchestrator (or “harness”) for AI coding agents. Through some shuffle in the early team, Emilie joined and started in building the fastest AI coding app on the market.

This is the creation story of Kilo.

Sponsors

Unblocked
TECH Domains
Mezmo
Braingrid.ai

Links

https://kilo.ai/
https://www.linkedin.com/in/emilieschario/

.TECH is giving founders building in AI, SaaS, devtools, and Web3 the chance to apply for premium domains — complimentary for 10 years. Names like host.tech, bot.tech, dev.tech. The kind of name that makes investors lean forward.

Your competitors are paying for theirs. You could apply for yours. → startup.tech

https://codestory.co/podcast/bonus-aggregate-metrics-hide-revenue-leaks-autonomous-agentic-analytics-bhaskar-sunkara-bicycle-ai/

Bhaskar noticed that business teams were drowning in dashboards, and as such, were not sure how to take the next steps in the business. He and his team realized that what people needed was not a retroactive view, but a proactive one – something more akin to a 24×7 analyst.

https://codestory.co/podcast/the-ai-control-loop-the-enterprise-ai-accountability-moment-with-shayne-higdon-of-wallarm/

Today, we are dropping our final episode in our series The AI Control Loop, How enterprises govern the AI they’ve already deployed – sponsored by our friends at Wallarm.

Wallarm is the AI Control Platform for Enterprise AI, protecting every AI workload, API, and application in production, giving CISOs the governance they need and CIOs the speed they demand. Organizations choose Wallarm for a complete inventory of APIs, AI agents, and AI apps, patented AI/ML-based threat detection and blocking that operates at production traffic speeds.

In our final episode, we are joined by Shayne Higdon, Wallarm CEO, who closes the series by examining what the accountability moment demands from enterprise leaders, what a mature AI governance model needs to prove rather than promise, and what the next 12 to 24 months look like for organizations that get this right.

https://codestory.co/podcast/e27-real-time-scaling-websockets-anycable-decoupled-irina-nazarova-evil-martians/

Irina is the CEO of Evil Martians, a well known design and engineering consultancy. During the time of the company, she and the team noticed that websocket solutions don’t guarantee delivery. They decided to build a new solution, one that does guarantee delivery, through automatic recovery of messages during connection issues.

Introducing… Fuzzball.

Fuzzball is CIQ’s HPC and AI orchestration platform, a single control plane that sits above the infrastructure you already operate and routes every job to where it runs best.

• No rip-and-replace. Fuzzball layers on top of your existing PBS and Slurm schedulers, your existing parallel file systems (Lustre, GPFS, BeeGFS), and your existing workflows. Nothing migrates and nothing breaks.

• One workflow definition, every environment. The same job runs on-premises, on AWS, GCP, Azure, Oracle Cloud, or CoreWeave, unmodified. When a job needs more capacity, it moves to cloud automatically with no rewrites.

• HPC and AI, unified. Run simulations and model training on the same cluster from one control plane. When training finishes, Fuzzball provisions the inference endpoint with no manual handoff and no separate tooling.

• Data sovereignty built in. Placement policies control where regulated workloads run, so models and data stay on infrastructure you own and control for the full lifecycle.

• Built for every user, not just CLI experts. A browser-based interface handles job submission, monitoring, and results, putting high-performance compute in reach of every researcher on the team.

Get a Fuzzball demo

https://codestory.co/podcast/ai-control-loop-whats-missing-ai-security-craig-thomas-wallarm/

Today, we are dropping another episode in our series The AI Control Loop, How enterprises govern the AI they’ve already deployed – sponsored by our friends at Wallarm.

Wallarm is the AI Control Platform for Enterprise AI, protecting every AI workload, API, and application in production, giving CISOs the governance they need and CIOs the speed they demand. Organizations choose Wallarm for a complete inventory of APIs, AI agents, and AI apps, patented AI/ML-based threat detection and blocking that operates at production traffic speeds.

In today’s episode, Craig Thomas, Sr. Solutions Engineer at Wallarm, returns to the show to dive into why runtime behavior is the critical blind spot, and what CISOs should demand if they want to move from policy to control.

https://codestory.co/podcast/crm-texts-spam-folder-fix-john-wright-truedialog/

In the past, John and his team built a platform around email, which they sold in 2001 to a company that is now apart of Google. Post that, he started to noticed the proliferation of SMS in the messaging world, in similar patterns as to what email did – and they decided to build a platform to serve the enterprise in this capacity.

Introducing Rocky Linux from CIQ Pro (RLC Pro)

Built on Rocky Linux, with 2.75 million actively deployed instances worldwide, RLC Pro is a commercially supported Enterprise Linux subscription that treats enterprise-grade capabilities as the baseline, not a premium tier.

• LTS, bundled as standard. Every RLC Pro subscription includes Long-Term Support. Platform teams plan on 3 to5 year infrastructure horizons without forced upgrades, on your roadmap, not the upstream release calendar.

• FIPS 140-3, included. NIST-certified cryptographic modules ship standard on LTS minor versions, so regulated industries including government, defense, financial services, and healthcare can achieve validated environments without resource-intensive internal certification.

• Direct bug fixes, not a queue. When a bug affects your environment, CIQ’s engineering team fixes it and ships the patch directly, with no waiting on upstream. Bug remediation is a subscription feature, not an escalation process.

• Purpose-built variants for specialized workloads. RLC Pro Hardened ships 95%+ DISA STIG and 99% CIS compliant at first boot. RLC Pro AI delivers measurably higher inference throughput on identical hardware versus stock Ubuntu, validated across NVIDIA and AMD GPU architectures.

• Zero lock-in. Full Enterprise Linux binary compatibility means existing applications, automation playbooks, and ISV certifications carry over. The switching cost from RHEL is effectively zero.

RLC Pro is available now on AWS, Azure, and Google Cloud Marketplace.

Learn more about RLC Pro

You need Tiger Data.

Tiger Data started as an IoT platform, but when off-the-shelf databases couldn’t handle their sensor data, they built a PostgreSQL extension that could. They quickly realized they were solving a problem a lot of other people had too, and pivoted to become a full database company.

Fast Ingest, Efficient storage & Real-time Analytics at Scale

Today, Tiger Data is the fastest PostgreSQL cloud platform that natively supports full-SQL, combining the power, reliability, and ease-of-use of a relational database with the scalability typically seen in NoSQL systems. It is built on PostgreSQL and optimized for fast ingest and complex queries.

From manufacturing and oil and gas to finance, logistics, space, and telecom, Tiger Data powers mission-critical applications where data speed and reliability aren’t optional. If a company is running complex monitoring, financial risk management, or industrial data analysis, Tiger Data was built for that.

Postgres That Doesn’t Break

Data is everywhere and it’s only compounding. Whether it’s AI agents, SaaS, or the physical world, data is the substrate for all of it. Tiger Data’s bet on PostgreSQL has proven right — it’s the leading developer database — and they’re going to keep building on that.

Tiger Data is doubling down on the physical world, targeting industries like manufacturing, AI data centers, electric vehicles, energy, battery technology, nuclear fusion, and chip design. These are some of the most data-intensive operations on the planet, like Nvidia and Spacex, generating massive volumes of time-series data that Tiger Data was built to handle. In a lot of ways it is a return to their IoT roots, just at a much bigger scale.

DB’s for Agents - One Per Task, No Blast Radius

Tiger Data has a strong conviction that AI agents are the new developer, and they are building PostgreSQL for that world. Their latest product, Ghost, is a database built specifically for AI agents and represents that vision in action. The philosophy behind it is the same one that has guided them from the start: when a new wave comes, you don’t throw everything out. You take what is working and you make it better.

https://codestory.co/podcast/the-ai-control-loop-detection-is-not-enough-with-tim-ebbers-of-wallarm/

Today, we are dropping another episode in our series The AI Control Loop, How enterprises govern the AI they’ve already deployed – sponsored by our friends at Wallarm.

Wallarm is the AI Control Platform for Enterprise AI, protecting every AI workload, API, and application in production, giving CISOs the governance they need and CIOs the speed they demand. Organizations choose Wallarm for a complete inventory of APIs, AI agents, and AI apps, patented AI/ML-based threat detection and blocking that operates at production traffic speeds.

In his follow up appearance on the Code Story podcast, Tim Ebbers, Field CTO at Wallarm, discusses why detection alone is insufficient for AI-driven systems, what real enforcement looks like at the runtime level, and what accountability becomes possible once all four stages are in place.