The idea behind Profile Context is simple: Teach Outcome about you and your business once, so you don't have to keep explaining it. The Agent would gather as much context as it can about you: your business, offers, audience, goals, expertise, writing style, positioning, products, and anything else useful. It could learn this from your website, content, existing funnels, and questions it asks you. Everything it learns would be visible and editable, so you're always in control of the context. That profile would then follow you throughout Outcome. When you create a new funnel, the Agent already knows your audience, what you sell, how you communicate, and what you're trying to achieve. When it creates an Outcome, it can make recommendations that actually fit your business. When you're reviewing leads, it can understand their answers in the context of what you offer and help identify the right next step. The goal is to move from: "Here's what I want you to build, and let me explain my business again..." to: "Build me a funnel for this." And Outcome already understands the rest. Over time, I think this context could become one of the most important parts of Outcome—not just helping AI generate better funnels, but helping the entire experience better match you, your expertise, your audience, and your goals.

Dylan Jones 8 days ago

I want to build a library of Outcome Funnel templates and examples that you can explore, try, and see working before creating your own. One of the hardest parts of a new tool is the blank page. What should I build? What makes a good Outcome? What questions should I ask? What can I even do with this? Templates should help answer those questions by showing, not telling. You could explore funnels for things like a personalized marketing plan, fitness assessment, business scorecard, content strategy, quiz, lead magnet, or product recommendation. Each one would be a real working Outcome Funnel you can go through as a lead, see the Outcome it creates, and then use it as the starting point for your own. The goal isn't just to provide templates to copy. It's to create a library of ideas that makes you think: "Oh... I could use Outcome for that." Then click a button, customize it, and make it yours.

Dylan Jones 8 days ago

Otto is currently in beta and hidden away, but the idea is pretty simple: An Agent inside Outcome that can do things for you. Think of Otto as an extra pair of hands (or maybe an extra mouse). You notice a headline that needs changing? Ask Otto. Want to rearrange a form step, add a new Outcome Block, change some copy, or make a few small tweaks? Ask Otto. The goal isn't to replace the interface. Sometimes clicking something and changing it yourself is faster. The challenge is designing Otto so he's there when useful, but never gets in the way when he's not. The more interesting part Where I think Otto gets especially useful is when you can have a conversation with your data. Open a lead and ask: "Tell me about this person." Or look at a Funnel and ask: "What are the most common problems people are telling me?" "How are beginners different from advanced leads?" "What should I change about this funnel based on the last 100 responses?" Eventually, that could extend across all the leads and funnels in your workspace. Instead of digging through profiles, answers, segments, and analytics, you can just ask. So I see Otto having two jobs: Do things for me — Make edits, move things around, add blocks, update copy, and handle those little changes you notice while working. Help me understand things — Talk to a lead, a funnel's audience, or eventually your entire workspace and help you find useful patterns. There's already a beta version of Otto working behind the scenes. The main thing I need to figure out now is the experience. The best Agent isn't the one you have to use. It's the one that's there exactly when it's faster than doing it yourself.

Dylan Jones 8 days ago

Right now, most Outcome Funnels use a simple intake model: Question → Answer → Question → Answer → Outcome. That works well for a lot of funnels, but it's not always the best way to get someone to an Outcome. Sometimes you want to take a lead or client through a more structured experience. A Workbook might take them through lessons, questions, and activities. A Challenge might spread those activities across five days, with each step building on the last. A Quiz might test their knowledge, grade their answers, and use the score as part of their Outcome. These are all different ways of collecting context and moving someone toward a result. I already have a lot of the infrastructure for this, along with beta versions of each model. The next step is giving them more love and polish so they can go live. The bigger challenge: AI Funnel Creation The interesting complication isn't just building the different intake models. It's making sure the Agent can create them really well. My goal with Outcome is that you should be able to give the Agent your content, framework, or idea and get a genuinely good first draft of the funnel you want. Adding Workbooks, Challenges, Quizzes, and eventually other intake models makes that creation process more complicated. The Agent needs to understand not only what questions to ask, but what kind of experience should be used to get someone to the Outcome. A five-day challenge shouldn't feel like a stretched-out form. A workbook shouldn't feel like a quiz. A quiz shouldn't just be a bunch of multiple-choice questions. Each needs its own structure. But I think solving this will make the entire Funnel creation process better. Because that's ultimately one of the standards I want to hold Outcome to: If Funnel creation is bad, you spend your time deleting what AI created. If Funnel creation is good, you spend a little time tweaking it and then you share it. That's the goal.

Dylan Jones 8 days ago

Dynamic Image Templates are currently in beta. The goal is to make personalized image generation fast, cheap, and reliable. The AI Image Block is great when every image should be unique. Think an infographic for someone's workout plan, launch strategy, or personalized roadmap. But sometimes you don't need to generate an entirely new image. You need a certificate, event badge, scorecard, report cover, or something else that has the same basic design for everyone—but personalized with their information. That's where Dynamic Image Templates come in. You design a template (or have AI help design it) and define a few dynamic areas: their name, date, score, recommendation, a few lines of text, an icon, etc. When someone reaches their Outcome, AI generates the values and fills in the blanks. The finished image is generated almost instantly, saved, and displayed to the lead. Compared to generating a new AI image—which can take 15–30 seconds and use premium AI credits—this should be about as cheap and fast as it gets. AI generates the content, not the entire image. Things to finish Image Template Editor — I have a beta editor that does the job, but it needs a lot more love. Better UX, better AI integration, and more “taste” built into the Agent so you don't need to be a graphic designer to create something that looks good. It should also be really obvious which parts of an image are dynamic and how to edit them. Dynamic Content — Right now text and colors can be dynamic, along with image and icon variants. Variants are particularly interesting. You could have three different badges or icons and let AI choose which one to display based on someone's score or Outcome. There are more dynamic properties I want to support, but the bigger challenge is making all of this feel simple rather than turning it into a complicated design tool. Preview & Testing — An easy way to preview different people, values, scores, and variants before publishing. The bigger idea is to have two ways to create personalized images in Outcome: AI Image: Generate something completely unique for this person. Dynamic Image: Start with a great design and intelligently personalize the parts that matter. Different tools for different Outcomes.

Dylan Jones 8 days ago

The Code Widget Block is currently in beta, but I want to finish it and polish it for production. The goal is to create an escape hatch for anything Outcome doesn't support natively yet: calculators, custom tools, interactive widgets, niche functionality, and other small experiences. The way it works today is pretty simple. You add a Code Widget Block, “vibe code” what you want with the Agent, and it generates the block for you. This isn't meant to be a full app builder. It's a small, focused piece of functionality that lives inside your Outcome page. When you publish, the code is saved and renders like any other block. One part of this that I think is especially useful is that the widget can save and update data on the lead's profile. So if someone uses a calculator, fills in values, changes settings, or interacts with the widget, that data can persist. If they come back later, their state can still be there. It's subtle, but I think this becomes important as Outcome Funnels get more interactive. Things to finish Design & styling — I want generated widgets to feel as native to the Outcome page as possible. Fonts, spacing, cards, inputs, buttons, and other styling should inherit the look and feel of the page. There should also be enough “taste” built into the Agent that it's difficult to accidentally create something ugly. Unless you ask it to. :) Templates & examples — A blank prompt can be stressful. I want to add useful starting points like calculators, assessments, planners, comparison tools, estimators, and other widgets you can instantly use and tweak. More efficient editing — Right now, even a small change can cause the Agent to rewrite the entire widget. Prompt caching helps, but I want to explore more targeted editing so small changes only touch the parts that need to change. This one may come later, but it's an interesting problem. The bigger idea is simple: If Outcome doesn't have the exact block you need, you should be able to make it.

Dylan Jones 8 days ago

I want to build an AI Outcome Block that can recommend the most useful resources to a lead based on their goal, answers, and context. The basic idea is simple: you probably already have a bunch of useful things—videos, articles, links, webinar replays, downloads, products, recipes, guides, or other resources. Instead of showing everyone the same things, Outcome could surface the 3–5 resources that are most useful for this person, right now. For example, a health food blogger might have 100 recipes. Someone goes through their Outcome Funnel and says their goal is to eat healthier, they only have 20 minutes to cook, and they're feeding a family. Instead of giving them a giant recipe library, Outcome recommends the four recipes that best fit. Your library + their context → the right resources. Things to figure out The biggest challenge is making this generic enough to work with any type of resource. The simplest version would be a Resource Library where you can manually add things and give each one a title, description, URL, image, tags, and other useful information. But ideally, an AI Agent does most of that work for you. Give it your YouTube channel, website, list of links, or other source and let it find, import, describe, and tag the resources automatically. Then the Resource Recommendation block has one job: Find the most useful things you already have for the person in front of you. This could be especially useful because the best next step isn't always another AI-generated answer. Sometimes you've already created exactly what they need. Outcome just needs to find it.

Dylan Jones 8 days ago

I want to add a way to quickly test an Outcome Funnel without having to go through the entire funnel as a normal lead would. This would make it easier to test Outcome Blocks, see how the AI Agent is behaving, and make sure the instructions and context are producing useful results. A big part of building good AI experiences comes down to the quality of the prompts + inputs + context. So the goal here is to make testing those things much simpler. Most funnels probably won't need a ton of testing. But if you're building a bigger "Hero" funnel, being able to really test different scenarios and responses could be the difference between an outcome someone finds genuinely useful and one they ignore. We don't want them ignoring the outcomes! :) Features to build Testing — Quickly run through a funnel, jump between steps, and create test profiles for different scenarios, segments, answers, and types of leads. For example, create a "Beginner" profile and an "Advanced" profile, then see how the same Outcome changes for each without filling out the funnel over and over again. Observability — Reveal more of what's happening behind the scenes: what context was used, what instructions were given, what the AI considered, and how the final Outcome was generated. The goal isn't to make Outcome feel more technical. It's to make it easier to answer one simple question: "Is this giving my leads the best possible Outcome?"

Dylan Jones 8 days ago