A new page on your site on schedule, aimed at a search people actually make.
The engine picks a phrase with measured demand, researches it, writes the piece, adds an image, files it as a Google Doc in your Drive and publishes it to your site on the cadence set with you. Approve every draft first, or let it run.
Will pages go out when nobody on my side has time?
Yes. The engine runs on a schedule whether anyone is at a desk. On the first of each month at 7am Eastern it queues the month's pieces for your account, the count set with you, and works through them: queued, researching, writing, draft, then approved, posting and posted, or rejected. A piece does not skip a stage. Research happens before writing, and a draft stays a draft until your standing approval or a person releases it.
Open the Content view and you see the pipeline as counts: researched, drafted, optimized, published this period, and the next piece on your cadence. Each piece opens to its type, its status and where it went, with a link to the live post and a control to request a rewrite. Below the pipeline sit your categories and tags, each with an add control, and the Drive folders where the finished Google Docs are filed.
Three kinds of piece run through the same stages: blog posts, research papers and press releases. The blog cadence is set per account. The default is two posts a month, accounts on the hands-off runner publish on the 1st and the 15th, and a blog targets about 1,000 words unless you set a different length.
How do you choose topics, so the page is actually read?
Every topic passes a demand gate before a word is written. The gate asks Google Ads for the phrase's real search volume through our own Ads access. Under 100 searches a month, the phrase is rejected: nobody is asking. Over 20,000 a month, it is rejected too: that is a head term a small firm will not take. If Google returns no figure at all, the phrase is rejected, because a missing answer is never a passing answer.
The second check reads the live top ten for the phrase. When more than six of the ten results belong to major publishers or funded software vendors, that page is already owned and the topic is dropped for one they are not defending. Each verdict is one sentence a person can act on, with the volume, the band and who holds the results page, and it is logged.
Take a phrase your site does not rank for. It clears the floor and the ceiling, two of the top ten belong to large publishers, so it passes. The engine then reads the titles already published on your site and the topics already queued this month, and leans toward the service your site covers least. The third near-duplicate is never written, and the quiet corners of your business get pages too.
We learned the gate the hard way. The first version of this engine wrote 26 well written answers on our own site to questions with zero searches a month, and the few phrases with volume were head terms owned by national software brands. Publishing is not earning either: in 90 days, 457 non-brand queries where our site sat in the top ten produced 4,719 impressions and 2 clicks. The gate exists so your pieces aim at a search someone makes and a results page you can take.
Is this the same generic AI content everyone else publishes?
No, and the research step is the reason. Before the draft, the engine runs up to four web searches on the topic and hands the writer up to twelve current sources, each with its address on the first line. The writer may cite only those addresses, copied character for character, three to five of them inline. It cannot construct a link, guess one, or point at a homepage as a stand-in.
The writing follows rules built for search and for answer engines alike. The piece opens with a direct answer to its title in 40 to 60 words, before any scene-setting. Headings are the questions a buyer asks out loud. The piece names your business and the area it serves in a plain sentence an engine can quote. Specifics beat adjectives: a size, a range, a timeframe, a standard, and never an invented one, so where the sources carry no figure the piece stays general. It closes with a FAQ of buyer questions, each answered in 40 to 60 words that stand alone, and it carries a meta description of 140 to 160 characters written for the search result. No em dashes, ever.
Then every external link is opened before publish. A dead link is unwrapped so the sentence still reads, and a piece with more than half its citations dead is held, not published. A post built on dead sources is not a post.
Our own site runs on this engine, which is how the failure modes were found before they reached a client. Source addresses that never reached the writer, links it invented in their place, topics nobody searched for: each became a check the pipeline now runs on every piece.
Does it publish without asking me?
Only if you tell it to. Auto-publish is a setting per account. With it on, a piece that passes its checks goes to your site on the cadence and our team is notified when it is live. With it off, the draft waits: it lands as a Google Doc in your Drive folder, filed under your business and the month, you read it there, and nothing is posted until you say so. A rejection carries your notes back into the pipeline. An account that approves every piece is never placed on the hands-off runner.
Publishing goes to WordPress through its API, or to a static site we host by re-render and redeploy. Our team connects your site during setup with a publishing credential you grant; there is no wizard and no consent screen for you to complete. A platform with no publishing API still gets the piece: a person publishes the approved Doc by hand and records the live address in the pipeline, so the record stays complete.
The archive is readable without our tool. Every piece is a Google Doc in a folder for your business, one subfolder per month, named by type and title, with the featured image saved beside it. It is month to month, and nothing in the archive needs our software to open.
Know the ceiling. The engine publishes; it does not measure. Whether a page earns shows in the search view from Search Console, 2 to 3 days behind, and in the analytics view from GA4, hours behind intraday. Positions there are Search Console average positions. Google's AI Overviews are measured through the search view, and the four answer engines are scored at baseline and re-run as the program moves.
Where do the images come from?
A real photograph first. For a service business a photo of the subject in good condition sells better than an illustration, so the engine searches stock photography for the subject of the post, not the action, and a model vets each candidate against the piece: the subject must be visible and fit this post, a photograph rather than a render, with no text, watermark or logo. If no candidate fits, the whole set is rejected and the next query runs. Only when stock cannot serve the topic is an image generated, and that image carries no text.
The chosen image is sized to 1200 by 800, set as the featured image on the post and saved beside the Google Doc in your Drive folder. A photographer whose work already covers your recent posts is skipped, so the same two people do not appear on a dozen covers. On a site we host, no post goes live without its image: a missing image holds the piece and alerts the team rather than shipping a bare page.
Have your own photography? Drop it into the photo folders in the Website view and it lands with your team.
Can I steer what it writes about?
Yes, from the screen, and each change is a logged request the next pieces follow. Add a category and new content files under it. Add a tag and it applies across content and answer-engine targeting. Add a target keyword and the engine starts aiming at it. Edit the brand voice and every future piece follows it. Change the cadence from the Schedule control. Ask for a rewrite from any piece; the screen tells you the rewrite usually goes live within one business day. Negative keywords, set with you, keep the engine off subjects that are not your business.
None of these write to your site directly. A request lands with the team, is logged, and shows up in the next pieces, and the Settings view shows the current voice, keywords and cadence, so what the engine is aiming at is never a guess.
Press releases are a la carte. They are written the same way, distributed on request, and listed in the view with the date distributed, the syndicated reprints tracked and the wire reach.
Check it in the demo. Open the Content view on demo.trulata.com. The company is fictional and every number is illustrative, but it is the same engine code every client runs. Look at four things: the pipeline counts, one piece opened to its detail, the categories and tags with their add controls, and the settings cards for platform, cadence, content types and auto-publish. Those four are what your own screen shows in weeks three and four of setup, when the content cadence starts.
What the content engine reads and writes.
| What | Where it comes from | How fresh |
|---|---|---|
| Topics | Google Ads search volume and the live top ten for the phrase | Before each piece is written |
| Research | Up to twelve current web sources, each with its address, cited inline | At writing time; every link opened again before publish |
| Dedupe | Titles already published on your site and topics queued this month | At topic selection |
| Publish | WordPress through its API, or a static site we host by re-render and redeploy | On your cadence; approval optional |
| Archive | A Google Doc in your Drive folder, one subfolder per month | As each piece is written |
| Images | A vetted real photograph first, generated only as the fallback, 1200 by 800 | At draft; set on the post at publish |
A topic nobody searches for, a claim its sources do not carry, a title already on your site, or a page that links to a source it could not open.
Questions, answered.
Will it publish without my approval?
Only if auto-publish is on for your account. With it off, every draft waits as a Google Doc in your Drive folder until you say yes.
Where do the topics come from?
From measured gaps. A phrase needs between 100 and 20,000 real searches a month and a results page not already owned by major publishers, and it must not repeat a title on your site. Anything else is not written.
Is the content researched or generated?
Researched first, then written. Up to twelve current sources with their addresses go to the writer, it may cite only those, and every link is opened again before publish.
Which platforms can it publish to?
WordPress through its API, and static sites we host by re-render and redeploy. A platform without a publishing API gets the approved Doc published by hand, with the live address recorded.
Do I get the content outside your tool?
Yes. Every piece is a Google Doc in your Drive folder, filed by month, with its image saved beside it.
How many pieces a month, and when does it start?
The cadence is set with you; the default is two blog posts a month, with research papers and press releases counted separately. The content cadence starts in weeks three and four of setup, after access in week one and the first pull, baselines and keyword set in week two.
Which tier includes it?
On Launch, Growth and Command. The full list of what each tier includes is on the pricing page.
See it running
before you decide.
The demo is the real product on a fictional company, with no form in front of it. Pricing is three published tiers. Setup on your own accounts takes two to four weeks.
Open the live demo See pricing
New here? See what the Command Center is. Prefer to write? Send us the one thing you want to know.
- Refreshed when you open itLive data on page load, never a monthly PDF
- Counted from your own formsLeads recorded server-side, compared daily with Ads and GA4
- Four answer engines, measuredChatGPT, Gemini, Claude and Grok, with a published denominator
- A review-first queue on every sendEvery draft waits in a review-first queue and every send is logged.