A Busy Sprint

We don't usually talk about what's happening behind the scenes this much, but the last few weeks have been genuinely exciting. Four substantial new features have shipped, and each one addresses something real that users were running into. Here's what's new and why it matters.

1. The Citation Probe: Your GEO Baseline in 60 Seconds

This one's the headline. The Citation Probe runs seven live queries across multiple AI engines and checks whether your brand gets cited in the response. It's your GEO baseline — the starting point that everything else in your strategy should be built around.

Before the probe existed, the only way to know whether ChatGPT or Perplexity was recommending your brand was to manually open those tools and test queries yourself. That's not a system. It's guesswork, and it doesn't give you anything you can track over time.

The probe changes that. You get a citation rate (the percentage of queries where your brand was mentioned), a breakdown of which specific queries are hitting and which are missing, and a timestamp so you can track how the number moves as you publish more content and add more vault facts. Run it now to get your baseline. Run it again in four weeks after you've published a GEO article or two. That delta is your proof that the strategy is working.

The missed queries are equally useful — those are your content gaps. Each uncited query is a topic you should be writing about. We've wired Citacious to read these results and recommend exactly that.

2. The Multi-Agent Orchestration Crew

The Agents tab has been in the platform for a while, but this sprint we rebuilt the underlying pipeline and added significant reliability improvements. Here's how the crew works:

Crawler Agent pulls live content from across the web on whatever topic you specify. It's using neural search, not keyword matching, so it finds genuinely relevant material rather than just pages that contain your search term.

Extraction Agent reads everything the crawler pulled and isolates the high-entropy facts — the specific, verifiable data points that AI engines are likely to cite. It discards the noise. Your Knowledge Vault context is injected here, so the agent knows what you already know about your brand and can build on it rather than repeat it.

Schema Agent takes those extracted facts and structures them as JSON-LD. This is the machine-readable format that AI crawlers prefer. It's not glamorous, but it's one of the highest-leverage things you can do for your citation rate.

Synthesis Agent writes the GEO article. Not a generic blog post. An article structured specifically to be cited — with a clear thesis statement, H2/H3 headers that match common AI query patterns, embedded statistics, and a Key Takeaways section at the end. The article is grounded in the facts the extraction agent found, which means it doesn't hallucinate.

When you hit Publish, the article saves to your database and, if you've set up the email webhook (more on that below), gets sent to you directly so you can copy it to your site.

3. Cost Audit: Understanding Your Unit Economics

If you're in the Superuser panel, you'll find a new Cost Audit section. This was built specifically to answer a question that comes up a lot in early-stage SaaS: what does it actually cost to serve one user?

The panel tracks every API call made by your account — Citacious conversations, Citation Probe queries, agent runs, content scoring, everything. It breaks down the total spend by feature, shows you a daily spend sparkline for the last 30 days, and calculates the average cost per call.

The pricing calculator at the bottom is where it gets interesting. Drag the subscriber slider to your target number of users. It calculates your total monthly API cost at that scale, then suggests a subscription price at a 3× cost margin (conservative for SaaS — most tools run 5-10×). It also shows projected net revenue and margin at that subscriber count.

This is the kind of data you need before you set your pricing. It's genuinely difficult to find this information anywhere else because it's specific to your usage pattern, not someone else's estimates.

4. Article Email Delivery: A Free CMS Substitute

One of the more practical things we shipped this sprint is a simple email notification system for generated articles. Here's the problem it solves: the Agents tab can generate a high-quality GEO article in a few minutes, but then what? If you don't have a CMS or a webhook endpoint set up, the article just lives in your Firestore database and you have to log in to read it.

We built a /api/notify-article endpoint that solves this without requiring any external tools. When you click "Publish to Database & CMS" in the Agents tab, it saves the article to your database and emails you the full article content, the extracted facts, and the JSON-LD schema — formatted and ready to use.

In Settings, under the Outbound Webhook section, there's a new "Use Email Notify" button that pre-fills the webhook URL with this endpoint. One click, save your settings, and from that point on every article you generate gets emailed to your account address. You can copy it straight into your website editor.

For most people at the early stages, this is all you need. Once you have a real CMS set up, you swap the URL out for your CMS endpoint and the workflow stays exactly the same.

What's Next

The Citation Probe is going to become more central to the platform as we build out the automated monthly re-run and trend tracking. We're also working on making the agent crew smarter about reading your existing vault before generating content — so over time, your articles get increasingly specific to your brand rather than the general industry.

More soon.