There are things you know you need to fix in your business. You write them down, you revisit them, you tell yourself you'll get to them when things slow down. Sales pipeline structure and marketing operations have been on my list for a while. They were the two biggest gaps standing between Obsidian Rising and the kind of growth I know this company is built for.
What I didn't expect was how quickly that would change.
The Problem With "Getting To It Later"
Running a business means you are constantly triaging. Every day there is a fire to put out, a client to serve, a deliverable to send. The strategic work — the kind that requires you to sit down, zoom out, and build the infrastructure your business needs to scale — gets pushed, because the urgent always wins.
For me, that strategic work lived inside a pile of spreadsheets. Over a million data points spread across multiple files, representing potential clients, leads, and contacts accumulated over time. Some of it was gold. Some was outdated, misspelled, or null. All of it was sitting there waiting for me to do something with it.
I knew a full sales funnel lived inside that data. I just needed a way to get it out.
A Word on the Tool Itself
Claude Fable 5 launched on June 9, 2026, and it represents a meaningful step beyond what previous AI models have been able to do in a real work context. A few things stand out that are directly relevant to what I experienced.
It is built for long-horizon tasks. That means it does not just answer a prompt in isolation. It holds the full scope of a project in mind, works through complexity without losing context, and adjusts as it goes. For a project like mine — one million-plus data points, multiple interconnected deliverables, a specific business goal — that continuity matters enormously.
It self-verifies. Fable 5 reflects on and validates its own work as it runs. This is part of why the data accuracy surprised me. The model was not just processing. It was checking itself.
It is built for knowledge work at scale. Early testers noted it handles document-based reasoning, complex analytical tasks, and data interpretation at a level that outperforms previous models. One early adopter noted it beat their spreadsheet suite at every effort level, completing runs 25 to 30 percent faster. That tracks exactly with what I saw.
None of this is to say the tool does the work for you. The inputs matter. What Fable 5 does is execute with precision when given the right direction.
What Changed: Claude Fable 5 in Practice
I want to be straightforward about what happened here, because I think people underestimate what this technology is actually capable of when you know how to work with it.
In under three hours, Claude Fable 5 did the following:
- Parsed through over one million datasets across multiple spreadsheets. It identified which data was accurate, which was outdated, which had errors, and which was missing values altogether. That level of data hygiene work alone would have taken days of manual effort — and even then, human error would have crept in.
- From that cleaned data, it identified the most relevant contacts to build a targeted sales pipeline — a pipeline built around real criteria specific to Obsidian Rising's goals.
- Built out the full funnel structure, from cold outreach all the way through to scheduling calls with prospective clients. Every stage documented and accounted for.
- Created on-brand documentation for each component of the system, written in Obsidian Rising's voice and aligned with how we actually operate.
- Drafted a comprehensive warm handoff document for Kami, my outreach agent who is now managing our mailing list, daily operations, and this new outreach initiative.
- Built out operational structures for the sales pipeline and marketing workflows that I can now work from, refine, and eventually replicate for others.
Kami's first automated outreach campaign is live.
This is her first time handling outreach at this scale — 40 to 80 targeted emails per day on the cold outreach component alone. Having a clear, detailed handoff document made the difference between a chaotic launch and a clean one. As of this week, the system is running exactly as designed.
The Part That Actually Surprised Me
Speed was impressive. The accuracy was remarkable.
When you are working with data at this scale, you expect some margin of error. You expect to have to go back in and clean things up. What I experienced instead was a level of precision that made me rethink what I thought I understood about what AI tools are capable of right now.
The model did not just process data. It reasoned about it. It flagged anomalies. It made decisions about relevance. It held the broader goal in mind the entire time: building a functioning outreach system, not just executing tasks in isolation. That is a different category of tool.
One more thing worth saying: the results I got were not just a product of the tool itself. Getting an AI model to perform at this level — accurately, on-brand, and aligned with a specific business goal — requires knowing how to communicate with it. I work at a fairly advanced level of prompt engineering, which means I know how to structure inputs, layer context, and guide a model toward outputs that are actually useful. That expertise was part of what made this work. The same tool in different hands would produce a very different outcome.
Where We Are Now
As of this week, the outreach system is live. Kami's first automated mailing send has been a success. Obsidian Rising now has a solid infrastructure and a functioning pipeline — both of which had been sitting on that to-do list far too long.
The system is running exactly as designed.
What This Means for You
If any part of this sounds familiar — if you have data you are not using, a funnel you have been meaning to build, or marketing operations running on instinct instead of systems — I want to talk to you.
This is not just something I figured out for my own business. It is a process I now understand well enough to build for others.
If you are ready to stop leaving structure on the table, fill out the contact form and let's start a conversation. Tell me where your blind spots are. Let's see what three hours could do for you.
Your Data Deserves a System Behind It
If your business has a contact list you haven't fully activated, a sales funnel that exists only in your head, or marketing operations that depend entirely on you showing up manually — this is the engagement. We take your existing data, build a functioning pipeline, and configure an AI outreach agent to run targeted, on-brand campaigns on your behalf. Forty to eighty personalized outreach touchpoints a day, without adding to your workload. If you are serious about scaling your marketing without scaling your hours, reach out. The conversation starts with your data.
Frequently Asked Questions
What is Claude Fable 5 and what makes it different from previous AI models?
Claude Fable 5, launched June 9, 2026, is Anthropic's most advanced AI model to date. It is built for long-horizon tasks — meaning it holds the full scope of a complex project in mind and works through it without losing context. It self-verifies its work as it runs, which produces significantly higher accuracy on data-intensive tasks. Early testers reported it completing knowledge-work runs 25 to 30 percent faster than prior models while outperforming traditional spreadsheet analysis. For businesses with large datasets, interconnected deliverables, or complex operational goals, Fable 5 represents a meaningful step beyond what previous AI tools could accomplish.
How did Obsidian Rising build a sales pipeline in under three hours?
Obsidian Rising had over one million data points across multiple spreadsheets — contacts, leads, and potential clients that had never been organized into a functioning funnel. Using Claude Fable 5, the process involved parsing all of that data to identify what was accurate, outdated, or missing; extracting the most relevant contacts based on specific ICP criteria; building a complete sales funnel from cold outreach through discovery call scheduling; creating on-brand documentation for each stage; and drafting a comprehensive handoff document for Kami, the AI outreach agent. The entire build took under three hours.
What is Kami and what does she do?
Kami is Obsidian Rising's AI outreach agent. She manages the company's mailing list, handles daily marketing operations, and runs a targeted cold outreach initiative sending between 40 and 80 personalized emails per day. Kami operates from a detailed handoff document specifying voice, goals, pipeline stages, and outreach parameters — which is what allows her to execute consistently and on-brand without constant human oversight. As of June 2026, her first automated mailing send has been a success and the system is running exactly as designed.
Can Obsidian Rising build an AI-powered outreach system for my business?
Yes. The process starts with your existing data — contacts, leads, CRM exports, or spreadsheets — and builds from there: data hygiene, ICP identification, funnel structure, on-brand documentation, and an AI agent configured to run outreach at scale on your behalf. If you have data you are not using, a funnel you have been meaning to build, or marketing operations running on instinct, this is the engagement. Reach out at obsidianrisingllc.com/contact to start the conversation.
Why does prompt engineering matter for AI-powered business systems?
Prompt engineering is the practice of structuring how you communicate with an AI model to produce outputs that are accurate, on-brand, and aligned with a specific goal. Without it, even the most powerful AI tools produce generic results that require significant cleanup. With it, those same tools can execute complex, multi-step business tasks with high accuracy and minimal revision. The same tool in different hands produces very different outcomes — which is why working with someone who understands how to direct AI models at an advanced level is a meaningful advantage, not a commodity skill.
What does AI-assisted data hygiene actually involve?
AI-assisted data hygiene involves parsing a dataset to identify duplicate records, outdated contact information, null or missing values, and inconsistent formatting. Once clean, that data is filtered against specific business criteria — ideal client profile, industry, geography, company size — to extract the contacts most likely to convert. At Obsidian Rising, this process transformed what would have been days of manual spreadsheet work into a structured, accurate output in a fraction of the time. That clean data then becomes the foundation for a functioning sales pipeline and targeted outreach system.