Return On Intelligence: Implement AI Without Wasting Time & Money

Is ChatGPT becoming the front door to your company’s software?

📋 Find your company’s AI readiness score: https://www.ninetwothree.co/tool/find-your-ai-readiness-score

OpenAI’s announced Oracle integration raises a question for business leaders: how much of your company’s work could happen through ChatGPT?

In this episode of Return on Intelligence, Andrew Amann and Nahshon Cook-Nelson discuss connecting AI to enterprise software, keeping control as model providers add proprietary features, and maintaining the apps your team builds with AI. They also look at how individual AI workflows fit into changes across an entire company.

In this episode:

- What the Oracle and ChatGPT announcement could mean for enterprise software

- Using one conversational interface to access business applications

- Choosing between model-agnostic systems and provider-specific features

- Why vibe-coded apps become harder to maintain as features accumulate

- Company knowledge, security, and ownership

- Evaluating AI beyond a single person’s productivity gains

Topics covered: ChatGPT for business • Oracle • Enterprise AI • Model-agnostic AI • Vibe coding • AI maintenance • Company knowledge graphs • AI ROI

About NineTwoThree:

We design and build AI products and software for businesses. Return on Intelligence covers the decisions behind that work, including what to build, how to operate it, and how to measure whether it helps.

Learn more: https://www.ninetwothree.co

Would you run your company’s software through ChatGPT? Tell us what you’d need to trust it with first.

If this helped, subscribe and share it with a leader making AI decisions.

Giving everyone Claude can make one employee faster while leaving the rest of the company buried in more work. How do you roll out AI so the whole team benefits?

On this episode of Return on Intelligence, NineTwoThree CEO Andrew Amann and Head of Marketing Nahshon Cook-Nelson explain how to connect AI adoption to team workflows, company knowledge, and measurable business results.

We walk through the four-step process we use with clients: align leadership, validate the problem, build the solution, and monitor what happens after launch. That includes setting rules for data access, choosing which problems to solve first, and checking whether AI spending produces useful outcomes.

Take the AI readiness quiz:
https://www.ninetwothree.co/tool/find-your-ai-readiness-score

In this conversation:
• Why individual productivity gains can create bottlenecks for other teams
• How an AI constitution defines approved tools, data access, and security boundaries
• Why your busiest senior employees can fall behind during an AI rollout
• How to connect token costs to business outcomes
• Andrew’s prediction for Anthropic’s expansion into cloud infrastructure
• What “owned intelligence” means for your company’s knowledge and experience

CHAPTERS
00:00 Why leaders should look past AI tool hype
00:42 Welcome to Return on Intelligence
01:27 Claude Dashboards, Motion, and the unified experience
02:09 Individual AI vs. team-wide results
04:54 Why traditional AI rollouts leave people behind
08:37 Our four-step AI implementation process
11:18 Moving beyond a personal AI assistant
12:50 Connecting token spending to revenue
13:58 Andrew’s predictions for Claude and cloud infrastructure
17:38 OpenAI’s consumer focus
19:06 Dots, GrokBot, Muse, and the distraction of new tools
20:20 Company knowledge and owned intelligence
22:51 Next steps and the weekly newsletter

Where does work get stuck after someone on your team uses AI to finish their part? Tell us in the comments.

Subscribe for more conversations about implementing AI across your business.

#AIImplementation #EnterpriseAI #ReturnOnIntelligence

Join 5,000+ leaders rolling out AI to their team: https://www.ninetwothree.co/guides/newsletter
___
Should your company build its own AI harness instead of relying entirely on Claude, ChatGPT, or another model provider?

Michelle C., AI Product Manager at NineTwoThree, explains what a custom AI harness is, when building one makes business sense, and why companies want more control over their AI infrastructure, workflows, data, and costs.

In this episode, we cover:

• What an AI harness does
• When a custom AI system is worth the investment
• The risks of depending on one model provider
• Where AI skills, tools, and business logic should live
• How proprietary data and workflows create a stronger system
• A real enterprise AI example, including the challenge, solution, and impact
• Why cheaper AI models may still produce more expensive agent workflows
• How security, infrastructure, and human oversight affect AI performance

If your company is evaluating AI agents, Claude, ChatGPT, or a custom enterprise AI system, this conversation will help you decide what to own and what to buy.

ABOUT THE GUEST

Michelle C. is an AI Product Manager at NineTwoThree. She works with teams to plan and build AI products grounded in real business workflows.

ABOUT NINETWOTHREE

NineTwoThree helps companies plan, build, and launch custom AI, web, and mobile products.

Learn more: https://ninetwothree.co

SUBSCRIBE

Subscribe for practical conversations about AI product development, AI agents, custom software, and the systems companies need to put AI into production.

What part of your AI system does your company need to own? Share your answer in the comments.

#ArtificialIntelligence #AIAgents #EnterpriseAI

Download our free AI Implementation guide non-technical leaders: https://www.ninetwothree.co/guides/ai-playbook
___
Self-hosting an LLM sounds like a way to stop paying for tokens
forever. The token bill turns out to be a small part of the cost.

Nay Cook-Nelson talks with Vitalijus, a machine learning engineer
with a PhD in natural language processing and 7+ years in data
science. He has deployed self-hosted models for tens of thousands
of businesses, starting years before GPT existed, back when
training a model from scratch on your own data was the only
option in machine learning.

⚡ IN THIS EPISODE

▸ What "self-hosted" means in 2026, and how it differs
from a serverless API call
▸ The 15 things you inherit the moment you leave a
third-party provider
▸ Why an H100-class GPU at $25K to $40K only gets you
to your first few concurrent users
▸ How a model priced 50% lower per token can cost more
in production
▸ Hybrid routing, and how to evaluate before you swap
models
▸ The team you need: ML engineer plus DevOps or MLOps,
on duty for the life of the project
▸ Where self-hosting is the only option: sovereignty,
edge, IoT, and zero-connectivity environments
▸ The token pricing trap, and what model deprecation
does to a product built on prompts

🕒 CHAPTERS

00:00 What you get from this episode
00:53 Seven years in data science, a PhD in NLP, and
self-hosted deployments before GPT
01:58 Self-hosting vs. serverless APIs
03:37 Why every ML model used to be trained from
scratch, and what BERT changed in 2018
06:13 Model distillation, and how DeepSeek got
competitive at a fraction of training cost
07:50 What you inherit: serving, scaling, load
balancing, autoscaling, failover, monitoring,
batching, caching, quantization, GPU
utilization, rollbacks, upgrades, guardrails
10:35 A 120B open-weight model needs an H100-class
GPU at $25K to $40K. Then a fifth user shows up
12:45 Verbose models eat your margin. Prompt caching
cuts cost up to 90%, and you build it yourself
13:48 Hybrid routing by use case instead of one
provider for everything
14:53 Three ways to evaluate an LLM system: humans in
the loop, performance metrics, models as judges
17:10 Zero-retention agreements, and when proprietary
or medical data rules out an API
19:07 Who you need to hire, and why it is a department
rather than two people
20:38 A healthcare client self-hosted Langfuse,
provisioned 7 to 8 resources for one dashboard,
then moved to a paid zero-retention tier
23:26 When self-hosting is the only option: 100%
uptime, aircraft, IoT, EV charging in garages
with no signal
24:50 US-hosted open-weight inference, and how to get
cheaper tokens without buying GPUs
26:17 Infrastructure carries about 95% of this decision
27:14 The token pricing trap. Under 5% separates the
top model from the tenth, per Stanford's AI Index
30:27 A version bump does not mean a better model for
your workload

📌 THE TAKEAWAY

Open-weight models are good enough for most workloads, so
picking one is the straightforward part. Infrastructure,
evaluation, and staffing decide whether the move pays off.
Vitalijus's rule: know your load, know your user count, build
a baseline with real evaluation, then experiment. Swapping
models without that baseline will cost you.

8,000+ industry leaders trust our newsletter to maximize their return on investment on AI: https://www.ninetwothree.co/guides/newsletter

What if you didn't have to pay for tokens? What if you could run models locally on your machine and use an open-source model with your own weights? This is already happening.

In today's episode, we break down what Sovereign AI is and how you can take advantage of owning your intelligence, not being over-reliant on closed models from Anthropic and OpenAI

If you're interested in building a custom AI solution but aren't sure where to start, book a discovery call with us here: https://www.ninetwothree.co/discovery/ai

Everyone's arguing about whether AI makes engineers faster. That's the wrong question. The real one: Is AI making your company more predictable?

In this episode of the Return on Intelligence podcast, Andrew Amann and Nay Cook-Nelson break down the "tokenmaxxing" trend — why Uber reportedly burned its entire 2026 AI budget in four months, whether tools like Claude Tag are a Trojan horse for your company data, and why "owned intelligence" (not your token count) is the real competitive edge. If your AI strategy is still measured in tokens saved or tools adopted, this episode is for you.

**⏱️ Timestamps**

```
00:00 – The wrong question everyone's asking about AI
00:24 – Is Claude Tag a Trojan horse for your company data?
02:29 – The economics of paying per token
04:05 – Are your Pro & Max accounts really safe?
06:09 – Owned intelligence: API access vs. the "harness"
08:34 – Why swappable models matter (Notion, GLM, DeepSeek)
09:23 – Controlling telemetry: chat apps vs. the API layer
10:35 – Tokenmaxxing & Uber's 4-month budget burn
12:50 – Measuring real ROI instead of token counts
13:56 – Assigning tasks to AI agents
15:40 – Taste, hiring, and project selection
17:21 – How to measure "taste" when hiring
19:18 – The "Pokémon" candidate-profile tool
20:23 – Final takeaway + call to action
```

**🔗 Links & Resources**

- 📰 Subscribe to the AI newsletter (5,000+ readers): https://hubs.li/Q048GVnH0
- 📞 Work with NineTwoThree (companies doing $10M+): https://www.ninetwothree.co/discovery/ai

- Uber burned its 2026 AI budget in 4 months (Fortune): https://fortune.com/2026/05/26/uber-coo-ai-spending-tokens-claude-code/
- Anthropic launches Claude Tag (Fortune): https://fortune.com/2026/06/23/anthropic-claude-tag-virtual-employee-tool-slack/

**📌 About**

NineTwoThree helps business owners and leaders implement AI without wasting time and money — focused on predictable, stable outcomes and owned intelligence.

👍 If this helped, like and subscribe for more practical AI strategy.

#AI #ArtificialIntelligence #AIStrategy #Tokenmaxxing #EnterpriseAI #OwnedIntelligence #AIAgents #AIforBusiness

Download Our Free AI Guide: https://hubs.li/Q04myJpC0
___
AI is amplifying engineers for better and worse. In this episode, we unpack vanity metrics, the -10x engineer, and why taste may be the most important skill in an AI-native engineering team.
___
Resources Mentioned:
Case Studies & Resources: https://hubs.li/Q03KQ_zG0
___
About NineTwoThree Studio:

We've built custom AI solutions for 100+ companies, delivering 7-figure returns from 6-figure investments. Our team combines AI expertise with proven product management to create solutions that actually work.

Book a discovery call to see if your AI project has real potential:
https://hubs.li/Q03KQ-Nd0

Download Our Free AI Guide: https://hubs.li/Q04myJpC0
___
How do you ACTUALLY become an AI native company? What are the step-by-step moves you need to make in your organization? In today's episode of Return On Intelligence, we walk you through those steps.
___
Resources Mentioned:
Case Studies & Resources: https://hubs.li/Q03KQ_zG0
___
About NineTwoThree Studio:

We've built custom AI solutions for 100+ companies, delivering 7-figure returns from 6-figure investments. Our team combines AI expertise with proven product management to create solutions that actually work.

Book a discovery call to see if your AI project has real potential:
https://hubs.li/Q03KQ-Nd0

Join 5,000+ AI leaders receiving our newsletter: https://hubs.li/Q048GVnH0
___
In this episode, we unpack Elon Musk’s surprise compute deal with Anthropic and what it signals about the real AI power shift. We also share the “$2,500 test” for deciding if your Claude Code idea is worth building, and why most teams who feel stuck with AI are usually reacting to FOMO instead of building durable systems.
___
Resources Mentioned:
Case Studies & Resources: https://hubs.li/Q03KQ_zG0
___
About NineTwoThree Studio:

We've built custom AI solutions for 100+ companies, delivering 7-figure returns from 6-figure investments. Our team combines AI expertise with proven product management to create solutions that actually work.

Ready to move beyond generic tools like ChatGPT? Book a discovery call to see if your AI project has real potential:
https://hubs.li/Q03KQ-Nd0

Apply to get a FREE AI Transformation Audit: https://www.ninetwothree.co/discovery/ai
___
In this episode of the Return on Intelligence podcast, Stephanie Leathe, an AI product manager with one exit and over 18 AI products on her roster, shares her RVS framework that allows business owners and leaders to validate their AI ideas quickly and land on the highest impact use cases in less time.
___
Resources Mentioned:
Case Studies & Resources: https://hubs.li/Q03KQ_zG0
___
About NineTwoThree Studio:

We've built custom AI solutions for 100+ companies, delivering 7-figure returns from 6-figure investments. Our team combines AI expertise with proven product management to create solutions that actually work.

Ready to move beyond generic tools like ChatGPT? Book a discovery call to see if your AI project has real potential:
https://hubs.li/Q03KQ-Nd0