Hire an SEO agent
Know what to publish next, and why.
Turn your site research into a topical map, a realistic calendar, and briefs and drafts your team can review. Keep the evidence behind each page decision and control what happens next.
Suggested action · refresh
Should we build or buy an AI workflow?
Why this page: readers need a practical way to compare ownership, integration effort and ongoing maintenance. Keep the cost model and security checklist on their own pages, and link them in context.
Research
Review real search questions, related concepts and relevant competing pages. Open the source and capture date behind each finding.
Plan
Assign one owning URL and a keyword cluster. Fit the refresh into the team's existing calendar.
Prepare
Brief the comparison, draft a worked example, then review its evidence and final presentation.
How priority is explained
Topsoil's PPR rubric combines prominence, popularity and relevance signals to help prioritise pages. Inspect the component evidence and missing measurements. It is an internal rubric, not a Google metric or ranking prediction.
- Use your site history and research evidence
- A reason on every page decision
- A calendar sized to the hours you actually have
- Inspect sources, dates and missing measurements
The gap
A useful map explains why each page belongs.
Start with the content you already have. See which questions it answers, which pages overlap, and which useful answers are missing. Each proposal carries its sources, intended reader and ownership boundary.
Bring together what people search, how the topics relate, and what relevant authority pages cover. Inspect capture dates and limitations before accepting a new page or refreshing an existing one.
Refresh evidence when you need it. Review resulting opportunities before changing your map or calendar.
Three evidence buckets
Understand the reason behind the recommendation.
What people search
Questions, queries and comparable result sets. Use Search Console when available, then inspect bounded discovery and its market scope.
How the topic fits together
Entities, attributes and relationships supported by source material. Use them to decide what the article should explain.
What useful pages already cover
Relevant authority pages, existing coverage and gaps worth investigating. A competitor heading is a research lead, not proof that your site needs another page.
One article, from evidence to review
Connected work. Visible reasons.
Described at the level of what each part does. The method is written down and credited; the implementation is ours.
Survey
Collects the evidence with its provenance: Search Console query and page rows, analytics landing context, bounded keyword discovery, comparable search result snapshots, entity relations checked against primary documents, and the page families of the sites that already rank. Every provider call has a ceiling, a cache and a receipt.
Map
Turns evidence into decisions. Source context and central entity. Core section and outer section. Entity relations with their direction verified. Search-result clustering as a candidate signal, then a human intent review before anything merges. A prominence, popularity and relevance rubric where an unmeasured component stays blank. On every page: keep, refresh, expand, create, consolidate or defer, with a boundary and a reason.
Calendar
The refresh schedule for what exists and the publishing schedule for what is new, in order, sized to the capacity you declare. Set separate limits for new articles and refreshes. Review scheduling changes before applying them.
Brief
For each scheduled page: the primary entity, the buyer's question, the attributes it must cover, what it must leave to its neighbours, the proof it needs, the links it makes, and how it will be measured.
Draft
Prepare a draft against its brief and attached brand standards. Keep revisions, visuals and review findings together. Your editor checks factual accuracy, usefulness and the final rendered article.
Record
Every observation, every cost, every decision and what superseded it, every release, every measurement window. When someone asks why a page is on the map, the answer is one query, not a reconstruction.
The toolkit
The whole SEO toolkit is in the same login.
Use the research toolkit for focused keyword, competitor and search-result work alongside your map. Results retain their source and measurement status. Provider-backed runs need configured access and a spending limit; unavailable measurements stay visibly unavailable.
Keyword research
Bounded discovery with a ceiling, a cache and a receipt on every batch.
Rank tracking
Agreed queries after a release, on your key, never a recurring surprise.
Competitor insights
Who actually ranks for a cluster, their page families and sitemaps.
Backlinks
Link data on demand, priced per check.
Site audit
Indexability, links and page checks, starting from the priority pages.
AI visibility
Inspect recorded AI-related observations and their provenance. Missing measurement is not a visibility score.
Open source, credited, pinned. You can read every line of what runs your data.
Your expertise, your standards
Prepare the work. Keep the decisions.
A brand profile belongs to one site. Review its voice, examples, factual boundaries and calibration before making it the default. Existing articles retain their attached standards until you explicitly create a revised version.
Useful visuals and downloads
Upload or reuse artwork, or render a simple process diagram from your verified copy. Retain originals, sources, rights, alt text, captions and article references. A generated file still needs a visual review.
A WordPress draft for review
The Topsoil companion connects through a dedicated WordPress account. Preview an exact package, map its media, then send a draft. Refreshes are staged separately from the published page. Site-specific theme and metadata behaviour needs verification during setup.
Ask about your next step
Ask Topsoil uses saved project context and links you back to the evidence. A local guide works without an AI account; AI explanations require configured model access. Answers and proposals do not approve articles or change your calendar.
Checks with honest limits
Inspect repetition, unfinished text and brief coverage. Record factual, voice, comprehension and rendered-page reviews separately. The optional Sapling diagnostic needs an account; its benchmark is pending and no score proves authorship or grants approval.
A bounded preparation queue
Queue exact saved article versions for package preparation. Set daily and retry limits, pause work and inspect durable outcomes. Changed articles stop for review. Paid calls, WordPress delivery and publication are separate actions.
You decide what goes live
Approve real expertise and claims, useful coverage, brand fit and asset rights. Review the final CMS preview. Topsoil's connector delivers drafts only; publication and actual completion remain separate records.
Pricing
Start free. Choose the access your work needs.
Agency
Run access for up to five sites, each with its own brand and context.
Review AgencyPreview pricing. Checkout is currently in test mode; no live charge. Subscription access and provider usage are separate. Research, AI and media services require their own configured access and budgets. Scheduled preparation currently packages selected saved articles while the server is running; it does not automatically refresh research or publish.
Who built it
We ran it on ourselves first.
Services Ground builds agentic systems and runs them in production. Topsoil started as our own SEO lane. In the first week of September 2026 our agent surveyed our Search Console rows, the search result sets for our topics and every live URL we had, made the page decisions, and scheduled the next seven weeks of new pages and refreshes for a team of one. Then we wrote the process down, stage by stage, and built the product on it.
Figures are from the Services Ground pilot, September 2026, servicesground.com, US and English. They describe work done, not results obtained.
Topsoil is built on OpenSEO, the open-source SEO tool, and is inspired by the topical authority method of Koray Tugberk Gubur. We credit both. We claim neither.
Get started
Start with your site, then inspect the evidence.
- Your query and page rows loaded, page by page, with the scope and capture time on screen.
- Your captured URLs, with exact matches and candidates that still need review.
- Any detected ownership overlaps, with the evidence needed to decide whether they matter.
- The decision Topsoil proposes for them, with the evidence behind it and the alternatives it rejected.
- The research scope and configured spending ceiling before a paid run. Actual provider usage is recorded separately.
Before you start
A few practical questions.
Can a new site start without Search Console history?
Yes. Define its audience and central topic, inventory any existing pages and gather research. First-party performance stays unavailable until there is real data; it is not filled with estimates.
Does Topsoil publish automatically?
No. The WordPress companion sends drafts and stages refreshes. Your team reviews the CMS result and handles publication separately. The preparation queue only packages selected saved versions.
Does a detector score approve an article?
No. Sapling is an optional diagnostic with false positives and false negatives. Its benchmark is pending. Factual accuracy, usefulness, voice, rights and release approval remain separate checks.
What happens to existing work?
Article versions, dates, evidence and export history are retained. Changing the default brand does not rewrite earlier articles. A deliberate standards migration creates a new article version.
Start
Start free.
Connect your sources, inspect your inventory and choose the plan for your work. Your team reviews the evidence, article quality and release decisions.