Hey there,
Megan here. I'm a Product Manager at Secfi, and while I typically spend my days building products and tools rather than writing about them, I wanted to share something we've been working on that gets at a trend I've been watching closely in the AI space.
If you’ve been keeping track of the evolution of AI tools over the past few years, you’ve probably noticed a shift. Generalist AI tools wowed us in the beginning and are still immensely helpful, but a new market for more specific, purpose built AI tools has emerged.
⬆️ The rise of purpose-built AI
We're entering what I'd call the "specialized AI" era. Yes, the generalist tools are still incredibly useful (I use them daily), but they're increasingly being complemented by AI tools purpose-built for specific domains.
Think about it: Harvey for legal work. Copilot for coding. Midjourney for visuals. Jasper for marketing. These aren't just ChatGPT wrappers with a different interface. They're AI systems built from the ground up to solve specific problems, connected to domain-specific data, and designed to handle the nuances that general-purpose AI often misses.
And here's why this matters for anyone with equity compensation: When you're making six or seven-figure financial decisions, "close enough" isn't good enough.
🧐 The problem with general AI for equity planning
I love Claude, I really do. But we noticed that sometimes Claude and other AI tools would get it wrong when it came to equity related calculations. We decided to make it better by combining claude with Secfi’s specialized tech, access to market data and your equity & tax data, to improve the quality of response for equity related questions.
The issue isn't that these models are not performant. Quite the opposite, actually. The problem is that they're trained to be helpful and confident, even when they shouldn't be. Ask ChatGPT to calculate your AMT liability, and it'll give you an answer. Ask it again with slightly different phrasing, and you might get a completely different number.
For recipe recommendations or vacation planning? That may be totally fine (I mean if your cake doesn’t turn out quite right you can always throw it out and try again without too much loss). For decisions involving hundreds of thousands of dollars in tax implications? Not so much.
The fundamental challenge is this: Generalist AI tools lack context. They don't know your grant details, your company's current 409A valuation, how your company is trading on secondary markets or know how to factor in the latest tax code nuances. They're making educated guesses based on patterns they've seen, not calculations based on your actual data.
🤝 Meet Maeve: AI built for your equity
This is why we built Maeve.
Maeve is specifically designed to help answer equity and tax questions with the kind of precision that high-stakes financial decisions require. We chose Claude Sonnet 4.6 as a foundation and built specialized functionality around it. Here's what makes Maeve different:
1. It knows your equity and tax situation
Maeve has access to your grant data, vesting schedules, exercise history, and tax profile. When you ask about exercising ISOs, it's working with your real numbers, not making assumptions about a hypothetical employee.
2. It uses our proprietary tax engines
When Maeve calculates your potential AMT liability or models an exercise scenario, it uses the same tax calculation APIs and equity modeling systems that our wealth advisors rely on, systems we've built and refined over years of serving employees at 90%+ of US unicorns. It’s not relying on prompting a language model to generate this number. We cite every number that’s been calculated by Secfi calculations APIs, so you can have confidence in the numbers provided by Maeve.
If you've used any of our calculators (AMT calculator, equity planner, etc.), Maeve is connected to those same underlying engines. The difference is now you can ask questions conversationally and more tailored to your unique situation.
3. It keeps a pulse on market conditions
Maeve integrates with our market intelligence data, pulling real-time company valuations and insights so your decisions are informed by current market conditions, not outdated assumptions. When you're thinking about whether to exercise before a potential down round, that context matters.
📈 The broader trend
I think what we're seeing with Maeve, and with specialized AI tools across industries, is a maturation of how we think about AI assistance. The initial wave was about demonstrating that AI could be helpful across many domains. This next wave is about making AI reliable in specific domains where accuracy really matters.
For legal work, that means AI that understands case law and regulatory frameworks. For healthcare, it means AI that can access patient records and evidence-based treatment protocols. For equity planning, it means AI connected to real financial data and calculation systems.
🫵 Try Maeve
If you're a Secfi user, Maeve is live in your dashboard now. If you're not yet a Secfi user, you can sign up and start asking questions for free at secfi.com.
A few things to try:
- "Should I exercise my ISOs before year-end?"
- "What would my tax bill be if I exercised 5,000 options today?"
- "How does a down round affect my equity value?"
- “How many ISOs can I exercise tax-free this year?”
Maeve is designed to be conversational, so don't worry about phrasing questions perfectly. And if it doesn't understand something or needs more information, it will ask.
And if you're thinking "Wait, how do I know Maeve isn't hallucinating too?", fair question. That's exactly why we built it on our proprietary calculation APIs rather than just prompting a language model. Every number it gives you from our system is cited, so you know it’s being calculated by Secfi, not from pattern matching.
❗️ The bottom line
The AI landscape is evolving from "one tool for everything" to "the right tool for the job." For equity planning, that means moving beyond general-purpose AI to systems specifically built to handle the complexity and precision that financial decisions require.
We built Maeve because we don’t want people making costly mistakes, either by trusting AI that hallucinated calculations or by avoiding AI entirely and missing out on helpful guidance.
Your equity compensation is likely one of your most valuable assets. It deserves AI that actually does the math.
Try Maeve and let me know what you think. You can respond to this email or find me on LinkedIn. I'd genuinely love to hear what questions you're asking and how we can continue to improve Maeve to make it more helpful.
Things we’re digging:
- 🤖 A startup called Sunday just raised $165M to put a robot that does laundry and clears the table in your house by Thanksgiving. Guess we've really reached The Jetsons’ territory...
- 📞 A yellow phone booth called "Call a Boomer" near BU lets you call a random senior citizen in Nevada for free. It's a social experiment by a biotech company trying to fight loneliness 😭
- 🚀 NASA's first crewed moon mission since 1972 is scheduled for April 1. Apparently not an April Fool’s joke.