The Electricity Test: Is Your AI a Feature or a Resource?

You don't buy electricity to light a room. You just turn the lights on. You don't evaluate electricity providers based on how well they power your kettle versus your laptop. You plug things in and they work. Electricity is a resource. Ubiquitous, invisible, and useful everywhere simultaneously. Now look at how the financial advice industry buys AI. "We're evaluating AI note-takers." "We've shortlisted three AI compliance tools." "We're looking at AI-powered report writers." The industry treats AI as a product. Something you buy, evaluate, compare, implement, and manage. Another line on the tech stack spreadsheet. Another vendor relationship. Another integration to maintain. That's the wrong framework entirely. AI-as-product versus AI-as-resource When you buy AI as a product, you get a tool that does one thing. A note-taker takes notes. A compliance checker checks compliance. A document generator generates documents. Each one is a discrete purchase, a discrete subscription, a discrete login. When AI is a resource, it's just there. You don't think about it any more than you think about electricity when you flip a switch. The meeting finishes and the facts are already extracted. The document drafts itself because the AI already knows the client, the compliance framework, and the firm's house style. The marketing content writes itself in your firm's voice because the AI has absorbed your brand, your philosophy, your tone. AI should be a thing that's ubiquitous around us, deployed wherever we need it. Not something we buy and bolt on, one function at a time. The question is: how do you tell the difference? Why most adviser AI is a feature, not a resource For AI to work as a genuine resource, like electricity, two things must be true. First, the AI and the platform must share the same data. An AI note-taker that can't see the client record is like a light bulb that generates its own electricity. It works, technically. But it's doing everything the hard way, reinventing context that already exists somewhere else in your stack. Second, the AI and the platform must be built by the same people. When your AI is a third-party integration, there's always a gap. The AI vendor doesn't fully understand the platform's data model. The platform vendor doesn't fully understand the AI's capabilities. The integration is a compromise; a lowest-common-denominator pipe that moves just enough data to technically function. We're the only advice tech company in the world, as far as we're aware, that has both a proper back office and its own AI. Everyone else is either a back office bolting on someone else's AI, or an AI company pretending to be a platform. That distinction matters more than any feature comparison. What AI-as-resource actually looks like At Ningi, the meeting note-taker isn't a product. It's one manifestation of the same AI that powers everything else. The same intelligence that processes your meeting transcript also: Generates documents - pulling from the client record, previous correspondence, and your firm's templates, without anyone copying and pasting context. Handles compliance - understanding your firm's specific framework, not a generic industry checklist. Extracts facts - updating the client record in real time as the meeting progresses, not after someone manually reviews a transcript. Writes marketing content - in your firm's voice, reflecting your investment philosophy and brand positioning. Powers the client portal - so your clients interact with an AI that actually knows their situation, not a generic chatbot. Runs operations - from workflow automation to MI reporting to task management. One resource. Many applications. No integrations required. The three tiers Not all AI should work the same way. Ningi runs three distinct tiers of agents, each with a different purpose. Ningi's own agents handle operations, sales support, and customer success. These work behind the scenes. They're for us, not for your clients. System agents do the heavy lifting during your day: fact extraction during meetings, document generation, workflow automation, data analysis. They activate when needed, do their work, and get out of the way. You don't manage them. You don't configure them. They just work. Firm-configurable agents are the ones your clients and your team interact with directly. Your marketing agent knows your brand's tone of voice and investment philosophy. Your compliance agent knows your firm's specific framework and regulatory approach. Your client's AI coach knows their situation, their goals, their risk profile. Each firm's agents are genuinely theirs. Not a generic model with your logo on it, but an AI that has absorbed your brand, your philosophy, your rules, your preferences. It gets smarter as you use the platform, because every interaction adds to the context it can draw from. The Electricity Test Here are six questions to ask any tech provider who claims to offer "AI-powered" advice technology. The answers will tell you whether you're looking at a feature or a resource. 1. Does your AI share the same data layer as your back office? If the AI and the platform are separate systems connected by an integration, the AI is working with incomplete information. Every integration is a filter. Some data makes it through, some doesn't. The AI can only be as good as the data pipe allows. 2. Can your AI see the full client record when it generates a document? Not "can it pull in some client data." Can it see everything - the record, the history, the previous documents, the meeting notes, the compliance flags, the investment philosophy - simultaneously, in real time, without someone manually providing context? 3. Is your AI built by the same team that built the platform? Third-party AI bolted onto someone else's platform will always be a compromise. The AI vendor optimises for their product. The platform vendor optimises for theirs. Nobody optimises for the join between them. 4. Can your AI work across multiple functions, or just one? If you need separate AI products for compliance, documents, marketing, and client communication, you don't have AI-as-resource. You have four products, four subscriptions, four data silos, and four vendors who don't talk to each other. 5. Can your AI be configured to know your firm's brand, philosophy, and rules? Generic AI gives generic outputs. If your compliance agent doesn't know your firm's specific framework, it's just a fancier checklist. If your marketing agent doesn't know your brand voice, it's just ChatGPT with a logo. 6. Does your AI get smarter as you use the platform, or is it static? A resource improves with use. Every meeting transcript, every compliance decision, every document generated, every client interaction should make the AI more useful. If your AI is the same on day 300 as it was on day one, it's a feature, not a resource. The honest scorecard Most adviser tech will score one or two out of six. That's not a criticism of the vendors. It's a structural limitation of how their products were built. You can't retrofit AI-as-resource onto a platform that was designed before AI existed. If your current setup scores three or fewer, you're buying features, not a resource. You're buying light bulbs that generate their own electricity instead of plugging into the grid. And that matters, because the firms that figure out AI-as-resource first will operate at a fundamentally different level. Not 10% more efficient. Structurally different. The kind of different where a seven-person firm with the right platform outperforms a twenty-person firm with the wrong stack. The electricity test isn't about Ningi versus anyone else. It's about asking the right questions before you sign the next contract. The industry is spending millions on AI features when what it actually needs is AI infrastructure. Stop buying light bulbs. Plug into the grid.