SolCrys Logo

How SolCrys Works

Inside the Action Hub—from finding to shipped, re-measured work

The Action Hub is where SolCrys findings become owned work. A Deep Analysis recommendation, a Content Audit finding, or a Signals idea becomes a task that keeps its evidence, gets an owner, a due date, and a named reviewer, and can move through a recorded review and approval before your team publishes the change. SolCrys never publishes to your site. Once a task is marked published with its live URL—in the dashboard, or by an AI assistant with an opt-in MCP permission—the same tracked prompts keep running on the same engines, so you can see whether answers start citing the page, and a page fixed from a Content Audit is re-audited, so you can see which lost points came back. This guide walks through the loop, the team layer it runs on, where it fails, and what it deliberately doesn't do.

Published · Updated

Questions this guide answers

  • What is the SolCrys Action Hub?
  • How does SolCrys turn AEO findings into tasks?
  • Can a team assign and review AEO tasks in SolCrys?
  • How does SolCrys check whether an AEO fix worked?
  • What's the difference between an AEO dashboard and an action engine?

What is live today

Everything on this page is in production as of September 2026: promotion into the queue from Deep Analysis, Content Audit, and Signals; the task layer added in August and September (owners, due dates, named reviewers, comments with @mentions, checklists, reminders, and archive); brief documents in Google Drive; recording publication in the dashboard or through MCP; the Impact tab; and Content Audit re-audit score recovery.

Two things the Action Hub does not do, and we repeat them wherever they matter: it does not write to your CMS or publish anything to your site, and it does not claim that a before-and-after change was caused by your fix. The re-measurement is evidence you read, not a verdict the platform issues.

Direct answer

The Action Hub is the Execute and Verify half of Measure → Diagnose → Execute → Verify. When measurement and diagnosis surface a gap—a prompt where competitors are cited and you aren't, a page losing audit points on checks you can fix, a timely topic nobody in your category has answered—the Action Hub turns it into a task with the evidence attached, routes it through human review and approval, records where the change went live, and re-measures the same prompts and the same page afterward.

It is not a dashboard with a task list bolted on. A task here carries the finding that created it and the measurement that closes it. If your current platform's answer to "what do I do now?" is "go write something," it stopped at Diagnose.

Why this is the half most AEO platforms skip

Building a dashboard for AI visibility is a tractable engineering problem. Probe a set of prompts, capture the responses, parse citations, aggregate, render. Most vendors in the category build a credible version of that.

The half that comes after is harder. It means choosing which of many possible gaps deserve work this month, grounding recommendations in a brand's approved facts rather than a model's guesses, routing work through an approval step that matches how marketing teams already operate, and re-measuring after a delay long enough for engines to pick up the change. None of those steps is glamorous; each is where the work either gets done or doesn't.

We invested in the loop before the dashboard was perfect, and we have also cut part of it. Until May 2026 the Action Hub reported automatic before-and-after deltas at fixed intervals after publication. We removed them because they had no control prompts and no significance testing, and a number that looks like proof when it isn't is worse than no number. What replaced them—re-running the same prompts and re-auditing the same page, presented as evidence to read—is described below.

The loop, step by step

Every task follows the same 6 steps. Measure and Diagnose happen before a task exists; the Action Hub picks up at the moment a finding is worth someone's time.

Step 1: A finding becomes a task

Tasks come from 4 places. A Deep Analysis on a tracked prompt explains why the answer came out the way it did—which sources the engine cited, which competitors it named, what is missing—and proposes actions with a recommended channel (owned, editorial, or community) and a content type. A Content Audit names the checks costing a page points, with the concrete current-versus-updated edit and the points each fix can recover. A Signals market brief proposes timely content ideas with an angle, timing, and sources. And anyone on the team can write a task by hand. Add to Actions creates the task with its evidence attached, and recommendations are grounded in your organization's Corporate Context, so they start from your real capabilities rather than invented ones.

One structural detail worth knowing before you trust the loop: an audit finding has exactly one tracked task, whether it was promoted when the audit ran or attached later, and the evidence behind it is snapshotted onto the task. The task keeps its evidence even after the audit run that produced it is gone, so a fix shipped in November can still be traced to the November finding, not to a re-run that no longer shows the problem.

Step 2: Triage—priority, type, and owner

Each task arrives with a proposed priority (high, medium, or low) and a type: Create for new content, Update for an existing page, or Fix for a specific page defect. You set the owner and the due date. Priority is a proposal, not a ruling. A Signals idea tied to a prompt where you are measurably absent arrives as high, an idea on an unmeasured prompt as medium, and you can change either. The Action Hub does not estimate revenue at risk; if your team ranks work by pipeline value, that judgment stays with the people who own the pipeline.

Step 3: Draft where your team already writes

For content tasks, the working document lives where your writers already work. With Google Drive connected, starting a Create task generates a brief document in Drive from the recommendation and links it to the task; you can link any other document URL instead, such as a Google Doc or a Notion page.

The drafting is done by your team, or by an AI assistant your team connects through the SolCrys MCP server, which can read the task, the Deep Analysis reasoning or audit detail behind it, and your Corporate Context. Our open-source Skills package 2 of those workflows: one drafts content for a queued task, the other drafts the page fix for an audit finding. Both are written to ground every claim in workspace data and the brand's own facts, and to return an outline and a fact checklist rather than invent facts when that grounding is thin.

Step 4: Review and approval

Requesting review names a reviewer and timestamps the request, and only the named reviewer can decide it. The reviewer sees the draft next to the gap evidence and can approve it, send it back for revision, or reject it; each decision is recorded on the task's timeline.

Approval is a recorded step rather than a lock. A task your team finished and approved elsewhere, in a CMS workflow or a legal review, can be marked published straight from In Progress. What never happens is SolCrys publishing anything: every change goes live because a person on your team shipped it.

Step 5: Your team publishes, then records it

Your team ships the change in your own CMS, through your own release process. Then someone opens the task's Publish tab, enters the live URL, and marks it published, from In Progress or Approved.

An AI assistant can record the same step through MCP if an admin grants the opt-in write permission for actions: it marks a Deep Analysis task published with its live URL. Content Audit tasks are published from the dashboard, because that step also triggers the page's re-audit. Either way, marking a task published updates the task in SolCrys; it does not touch your site.

Step 6: Re-measure

After publication, the task's Impact tab tracks the page against the prompts you already measure. For any task with a live URL, it plots per engine how many answers cite that page each day and how many tracked prompts cite it, with the publication date marked against a baseline window, and it records the first date each engine cited the page.

For a task promoted from a Content Audit, the Impact tab leads with the re-audit: the page's score before and after, how many of the task's checks recovered, and a per-check before-and-after, with a button to re-audit again. Give it days to weeks, because engines pick up changes on their own crawl schedules, and read any movement as an observation. The Impact tab itself notes that citations can change for reasons beyond the update.

The task layer: running the loop as a team

A loop that only one person can operate is a demo. Between August and September 2026 the Action Hub grew the parts that let a marketing team run it the way it runs everything else, without moving the work into a separate project tool where it would lose its evidence.

An action moves through 6 states: todo, in progress, in review, approved, published, completed. Publication and completion are deliberately separate: shipping a page is not the same as closing the loop on it, and re-measurement happens between the two.

  • An owner and a due date. Every action can be assigned and dated, so the queue is a workload rather than a list of suggestions.
  • An explicit review handoff. Requesting review names a reviewer and timestamps the request; the reviewer sees what is waiting on them specifically, not a shared pile.
  • Comments with @mentions. Discussion lives on the action, next to the gap evidence that prompted it, rather than in a thread somewhere else that loses the context by the second reply.
  • Checklists. A single action, such as refreshing a canonical page, usually contains several concrete edits. Sub-items keep partial progress visible instead of a card that sits at "in progress" for 2 weeks.
  • A blocked flag, not a seventh state. A blocked card keeps its column and carries a badge and a reason, so the board shows what is stuck without pretending it moved backward.
  • Reminders you control. Per person, per workspace, in your own time zone: due soon, overdue, pending review, and an optional daily digest. Email is opt-in and off by default.
  • Archive that is independent of status. Clearing the board does not require pretending a task was done.
  • The original recommendation is preserved. When someone edits an action's description, what the platform proposed is kept alongside what the team decided to do, so a quarter later you can tell which outcomes came from the recommendation and which came from the edit.

Failure modes (the part the demo doesn't show)

Five failure modes show up in real use. We document them because a platform's posture on failure is the real test of whether it's safe to build a workflow around.

Reading noise as a result

AI answers vary from run to run. SparkToro's January 2026 study of 2,961 runs found less than a 1 in 1,000 chance that 2 runs return the same brand list in the same order, while visibility percentage across many repeated runs held up as a reasonable metric. A single day's citation of a newly published page is a data point, not a result; the Impact tab earns its keep as a trend across daily runs on a fixed prompt set.

Slow review

If approval sits for weeks, the gap a task addressed has often shifted by the time the change ships, and the re-measurement reflects the staleness, not the quality of the work. Due dates, pending-review reminders, and the named-reviewer handoff make that wait visible; they can't shorten it for you.

Disagreeing with the recommendation

When a reviewer rejects or rewrites a recommendation, the disagreement is often about what Corporate Context says, or fails to say, rather than about the task itself. The original recommendation stays next to the edited version so the disagreement remains visible. The durable fix is to correct Corporate Context at the organization level, so the next recommendation starts from the right facts.

The engine moved, not your page

An engine can change what it cites for reasons that have nothing to do with your work. In the week of Aug. 10, 2026, ChatGPT all but stopped citing Reddit, a shift no content task caused, while Google AI Overviews and Perplexity were largely unaffected. A task published that week could look like a success or a failure depending on its source mix. The Impact tab shows the timeline per engine; separating your change from the engine's change is a judgment the platform leaves to you rather than fakes.

Work with no page to track

Some tasks don't end in a URL you own, such as pitching an editorial source or correcting a fact in Corporate Context. Those tasks close with a completion note instead of a published URL, and their effect shows up, if at all, in the prompt set's broader numbers rather than on a page-level chart.

What's not in the Action Hub

Four things we explicitly don't try to be:

  • A CMS or a publisher. SolCrys has no write path into WordPress, Webflow, Contentful, or any other CMS, and it never publishes on your behalf. Approved work ships through your own release process, and the content stays where you store everything else.
  • A general SEO backlog. Tasks come from AI-answer evidence, Content Audits, and market signals. SEO foundations still matter, especially for Google's AI surfaces, which build on Google's Search index and ranking systems, and the Content Audit checks several of them, such as indexing, canonical URLs, and sitemap inclusion. Rank tracking and link building belong in your SEO tools.
  • A creative agency. Recommendations and briefs speed up the operational content a marketer would otherwise write from a brief. Manifesto essays, original research, and founder bylines still come from people.
  • A replacement for your project tracker. If your team schedules everything in Jira, Asana, or Linear, keep doing that. The Action Hub is where the evidence and the measurement live.

How it differs from a dashboard with a task list

Some AI visibility platforms have added task or recommendation tabs to their dashboards. The useful distinction is whether the task is open-ended or closed-loop.

Open-ended: the platform says a prompt is losing share, suggests creating content, and your team takes it from there. The platform never sees the work again, so nobody checks whether the answer changed.

Closed-loop: the task keeps the evidence that created it, passes through a review the platform records, knows where the change went live, and is re-measured on the same prompts and, for a page fix, the same audit checks. That doesn't make the platform smarter on its own. It makes your team's record of what worked, and what didn't, reliable enough to learn from.

Both designs are legitimate. Open-ended tools are cheaper to build and lighter to run; closed-loop tools ask more of onboarding and of your team's discipline in recording publication. SolCrys is the second.

There is a third possibility worth naming, because it is where the category is heading: a task list bolted onto a dashboard, with owners and comments but no measurement behind the task and no re-measurement after it. That is a project tool wearing a dashboard's clothes. The test is not whether a platform has assignment and comments; those are table stakes, and we added them late. The test is whether the task carries the evidence that created it and the measurement that closes it.

Where to look in the product

Open Actions Hub from the workspace navigation (the product labels it that way; we call it the Action Hub here). List view gives a compact review, Board view shows work by stage, and Mine filters to what you own. Each task opens on 4 tabs that follow the loop: Plan, Content, Publish, and Impact.

To see it on your own brand, the Free plan includes 1 Deep Analysis and 1 Content Audit a month. Add their recommendations to the Action Hub and walk one through review, publication, and the re-audit.

How this connects to the rest of the product

The Action Hub is where the rest of the loop converges. The Visibility Measurement Methodology explains how the prompts it re-measures are run and captured. What a SolCrys Content Audit looks like covers the checks and points behind audit tasks and how re-audit recovery is computed. Corporate Context grounds the recommendations, and SolCrys MCP and Skills covers how an assistant reads tasks and records publication. Each piece alone is incomplete.

Sources

FAQ

Does the Action Hub auto-publish content to my site?

No. SolCrys has no write access to your site or CMS. A task can move through a recorded review and approval inside SolCrys, your team publishes the change through its own process, and a person, or an assistant you have explicitly granted the opt-in write permission, records the live URL afterward.

Where do Action Hub tasks come from?

Four places: Deep Analysis recommendations on tracked prompts, Content Audit findings on specific pages, Signals content ideas from the market brief, and tasks your team writes by hand. Add to Actions creates the task with its evidence attached, and each Content Audit finding maps to exactly one tracked task.

Does SolCrys write the content for a task?

It prepares the brief, not the page. With Google Drive connected, starting a Create task produces a brief document from the recommendation. Your team drafts from it, or an AI assistant connected through the SolCrys MCP server drafts using the task's evidence and your Corporate Context; our open-source Skills package that workflow for Claude and other MCP clients. Every draft goes through your review before anything ships.

Can an AI assistant mark a task as published?

Yes, if an admin opts in. The SolCrys MCP server has a narrow write permission for actions: with it, an assistant in Claude, ChatGPT, or Cursor can mark an approved or in-progress Deep Analysis task as published with its live URL. It cannot publish content to your site, create audits, or start measurement runs, and Content Audit tasks are published from the dashboard so the re-audit runs.

How does SolCrys tell whether a fix worked?

Two ways, both observational. The same tracked prompts keep running on the same engines, and the task's Impact tab shows, per engine, whether answers start citing the published page and on how many prompts. For a page fixed from a Content Audit, a re-audit compares the page's score and the task's checks before and after. Neither proves your change caused the movement, because engines change what they cite on their own, so read them as evidence rather than a verdict.

Can I try the Action Hub before paying?

Yes. The Free plan is a real workspace on your own brand, with 10 tracked prompts, on-demand ChatGPT checks, and 1 Deep Analysis and 1 Content Audit a month; it needs email verification and no credit card. Promote a finding and walk it through review, publication, and the re-audit. Start Free.

Can I assign Action Hub tasks to specific people and review each other's work?

Yes. An action carries an owner and a due date, and requesting review names a reviewer and timestamps the handoff, so the reviewer sees what is waiting on them rather than a shared pile. Discussion happens in comments on the action itself, with @mentions, next to the gap evidence that produced it. Reminders are per person and per workspace in your own time zone (due soon, overdue, pending review, and an optional daily digest), and email notifications are opt-in and off by default.

Why not just run AEO tasks in Jira, Asana, or Linear?

You can, and for pure scheduling those tools are better. What they cannot do is carry the evidence and the measurement. A SolCrys action holds the prompts and engines that produced the gap, the cited sources that beat you, the original recommendation alongside whatever the team edited it into, and the re-measurement after the fix ships. Copy that into a generic tracker and you get a title and a due date; the reason the work existed stays behind. Teams that want both usually keep scheduling where it lives and keep the evidence here.

Free ChatGPT visibility check

See where AI answers skip your brand — then fix it, free

Start a free workspace with your domain: 10 buyer-intent prompts through ChatGPT show where you are mentioned, cited, or skipped, and who gets recommended instead. A free content audit in the same workspace hands you the first fix to ship.

Start Free

Free · No credit card · About 5 minutes

Related guides

How SolCrys Works

What a SolCrys Content Audit Looks Like

What a SolCrys Content Audit checks on one URL, how the score and recoverable points work, and how a fix is promoted, published, and re-audited.

How SolCrys Works

How to Set Up a SolCrys Workspace: A Best-Practices Guide

What a SolCrys workspace is, the two ways to create one, and the best practices that keep your AI-visibility measurement clean: scoping the category, when to use multiple workspaces, engines, competitors, Corporate Context, and the prompt set.

How SolCrys Works

AI Visibility Measurement Methodology

How we capture your AI visibility data across supported engines, with each response traceable to a prompt, engine, capture method, available model or surface signal, and timestamp. Consumer-surface and retail-assistant validation are scoped where technically reliable.

How SolCrys Works

Corporate Context Is the New CMS

Corporate Context gives AI marketing agents the brand facts, claims, guardrails, and evidence they need to execute safely across AEO workflows.

Buyer Guides

AI Visibility Dashboard vs AEO Execution Engine

AEO platforms split into two architectures: dashboards measure and report; execution engines diagnose, fix, and verify. This guide compares 6 use cases, walks through 5 real scenarios, and ends with a decision tree.