Blog

  • What exactly happens during and after the GPTpost free trial

    Free trials often hide their most important details: whether a card is charged, what happens to your data, which features are locked. Here is everything about the GPTpost trial in one place.

    During the trial

    • Length: 14 days from signup.
    • Card: not required.
    • Limits: 5 seats, 15 connected social accounts and 500 posts.
    • Features: all of them, including per-network versions, approvals, client review links, posting slots, recurring posts, CSV import, RSS to drafts, and the REST API and MCP server.

    When the trial ends

    Nothing is charged, because no card was given and the trial never converts to a paid plan on its own. Your workspace becomes read-only.

    What read-only means

    • Everyone can still sign in.
    • Every post, schedule, comment and file stays readable.
    • You can export your data.
    • Nothing is deleted.
    • New posts wait until a plan is chosen.

    Continuing

    Choose Solo (12 USD), Team (39 USD) or Agency (99 USD) per month in Settings, Billing. Checkout runs through Stripe. Your existing drafts and schedules are still there.

    Need more time?

    Evaluations with several stakeholders sometimes take longer than two weeks. Write to hello@oraglegpt.org and explain what you are testing.

    A planning suggestion

    To get the most out of 14 days: connect your real accounts on day one, set up posting slots, and run one full week of real posts through approval. Invite at least one reviewer, ideally the person who usually signs off, and send one client review link if you work with clients. By the second week you will know whether the workflow fits your team, which tells you more than any feature list. If you plan to use the API or MCP server, create a test key early and try a small scheduling script against a draft-only setup.

    Plan details are on the pricing page.

    https://oraglegpt.org/pricing/?utm_source=facebook&utm_medium=social&utm_campaign=organic90

  • Hashtags in 2026: none, a few, or a wall? A practical guide by network

    Every team has a hashtag habit. Some started with a guideline written years ago, some copied what a competitor does, a few actually tested it. Here is a practical way to think about hashtags network by network, and a question for you at the end.

    Start with the purpose

    Hashtags do two jobs: they help people discover posts on a topic, and they group related posts together, such as an event or a campaign. If a tag does neither, it is decoration. A long block of generic tags at the bottom of a post mostly adds clutter.

    Network by network

    • Mastodon: hashtags are one of the main ways people find posts, because there is no central algorithm pushing content. A few relevant tags help. Write them in CamelCase so screen readers pronounce the words correctly.
    • LinkedIn: one to three relevant tags is a common, reasonable habit. More rarely helps.
    • Instagram: tags are still used for discovery. Choose specific ones that match the post rather than the most popular ones.
    • Bluesky: many posts use none. Custom feeds and good text often matter more.
    • X: one or two at most, or none.

    Accessibility matters

    Whatever the network, CamelCase helps. A screen reader reading a lowercase multi-word tag often produces nonsense. Capitalising each word costs nothing.

    Make it repeatable

    If you manage several brands or clients, inconsistent tags creep in quickly. Saving approved sets as hashtag groups in your scheduler, as GPTpost allows, means the right tags go on the right client’s posts every time.

    Test your own rule

    Pick one network, alternate posts with and without tags for a month, and compare replies and clicks. Your own numbers beat any general advice, including this article.

    So, what is your rule? More from us: oraglegpt.org.

  • Post templates for recurring formats: consistency without the blank page

    Most social accounts that work well over time have recurring formats: a weekly tip, release notes, event reminders, a monthly roundup. Readers learn to recognise them, and the team knows what to write. The challenge is that each one still starts with a blank composer, and small inconsistencies creep in.

    What templates are in GPTpost

    Templates are a snippet library inside the composer. Each template has a name and a text. When you write a post, you insert the template and fill in the parts that change. It is deliberately simple: no special syntax to learn, just a reusable starting point.

    Formats that benefit most

    • Release notes: “What changed”, “who it helps”, “link to full notes”. The structure stays, the content changes.
    • Event reminders: what, when (with a timezone, always), how to join, and who to ask.
    • Weekly tip: a recognisable opening line, the tip, and a short sign-off.
    • Client sign-offs: the wording and closing line a client approved once, reused correctly every time.

    How to write a good template

    1. Start from a real post that worked well.
    2. Replace the specific details with clear placeholders, such as [what changed].
    3. Keep fixed text short. A template should guide, not dominate.
    4. Name it so anyone on the team can find it.

    Combine with other features

    Templates work well together with hashtag groups (saved hashtag sets), recurring posts for formats that do not change at all, and posting slots so the weekly tip always lands at the same time. Together they turn a recurring format into a routine rather than a task.

    Review them occasionally

    Formats get stale. Every few months, look at which templates are used and which are ignored, and update the wording.

    More about GPTpost: oraglegpt.org.

  • How to write a one-page brand voice guide your team will use

    Brand voice guides have a habit of growing until nobody reads them. A forty-page document with mood boards and archetypes may be useful once, but it is not what someone opens at 16:00 while writing a post. A one-page guide is.

    1. Three adjectives that rule something out

    Pick three words that describe how you sound. The trick is to choose words that exclude something. “Friendly” and “professional” describe almost every brand, so they guide nothing. “Plain” rules out jargon. “Calm” rules out exclamation marks and urgency. “Specific” rules out vague claims. For each adjective, add one sentence explaining what it means in practice.

    2. Words you use

    List your product names exactly as they should be written, how you refer to customers (members, clients, teams), and any terms you prefer. This section prevents small, repeated inconsistencies.

    3. Words you avoid

    Hype words, internal jargon, competitor names in certain contexts, anything legal has flagged. Short and concrete. Ours lists the usual hype words and a ban on em dashes, which forces shorter sentences.

    4. Two example posts

    Examples teach faster than rules. Include one post that is clearly on voice. Then include one that was rewritten, with the original and the edited version side by side, and a line on what changed and why.

    The five-minute test

    Give the page to someone new, such as a freelancer or a new teammate, and ask them to write a post. If they cannot apply the guide in about five minutes, it is still too long or too abstract. Cut until they can.

    Keep it near the work

    Put the guide where posts are written. In GPTpost, saved templates can hold standard openings and sign-offs, so the voice is built into the starting point rather than checked at the end.

    More about GPTpost: oraglegpt.org.

  • “Can you just post this real quick?” and other Friday evening classics

    It arrives on a Friday, usually between 18:30 and 19:00. “Can you just post this real quick?” The attachment is a screenshot of a document. There is no link. The logo is forty pixels wide. The only direction is “make it pop”. On approval, the note says the boss is fine with it, probably.

    Every social media manager has a version of this story. Here is a slightly less stressful way to handle the next one.

    Why “real quick” is never quick

    A single post touches several small decisions: the right version for each network, a working link, an image with alt text, and someone with authority saying yes. Each of those takes a minute when planned and twenty when rushed. The person asking sees one post; you see five tasks and a risk.

    Agree the rules on a calm day

    • A cutoff time. For example, requests after 16:00 on Friday are published on Monday unless they are urgent.
    • A definition of urgent. A real incident or a time-bound announcement, not “the CEO likes it”.
    • Who approves. A named person, not “probably fine”.
    • What you need. Text, link, image, approver. Missing pieces mean it waits.

    Let the process say no for you

    When a rule exists, “first thing Monday” is not a personal refusal, it is just how things work. A shared calendar helps too: the requester can see that Monday at 9:30 is a perfectly good slot, and the approval stage shows who still needs to say yes.

    And when it really is urgent

    Sometimes it is. Then do the minimum safely: one network, one version, a second pair of eyes, and a note to review it on Monday.

    GPTpost has approval stages and a shared calendar for exactly these conversations: oraglegpt.org.

  • Sharing next month’s social plan with clients through their calendar

    Every agency has a client who approves everything by email, never logs in to the scheduling tool, and then asks in the middle of the month what is going out next week. You can fight that habit, or you can work with it.

    Put the plan where they already look

    Most clients check their calendar many times a day. GPTpost can export the planned schedule as an .ics file, the standard calendar format that Outlook, Google Calendar and Apple Calendar all open. You choose a date window, for example next month, and download the file. Each scheduled post appears as an event at its planned time.

    For a client, that turns an abstract plan into something concrete: the product post sits on the same day as their launch meeting, and a gap in the week is easy to spot.

    Keep approvals separate

    A calendar is good for an overview, not for sign-off. For that, send a client review link. The client opens it, reads the exact version of each post for each network, and approves or leaves a comment. They need no account and no seat. Posts that are not approved simply wait.

    A monthly routine

    1. Finish next month’s drafts and run them through your internal review.
    2. Export the schedule for that month as an .ics file and send it with a short note.
    3. Send the review link with a clear deadline for approvals.
    4. If dates change after feedback, export again and send the updated file.

    One honest limitation

    The export is a one-time download, not a live subscription feed. If the plan changes often, re-send it or point the client to the review link, which always shows the current versions.

    Learn more about GPTpost for agencies: oraglegpt.org.

  • When news breaks: a pause plan for scheduled social posts

    Scheduling lets you write posts when you have time and publish them when your audience is around. The catch is that a scheduled post was written in a different moment. When serious news breaks between writing and publishing, a cheerful promotion or a joke can land very badly.

    Decide before you need it

    The worst time to design a process is during the event it is for. Write down three things now:

    • Who can pause. In a small team, anyone who notices. In an agency, the account lead, with the client informed.
    • What gets paused. Usually everything scheduled for the next 24 to 48 hours.
    • Who decides to resume. One named person, so the queue does not restart by accident.

    The playbook

    1. Pause. Stop the next day or two of posts. It is easier to resume a post than to delete one that already went out.
    2. Read the room. Look at what your audience and similar accounts are posting. Silence is sometimes the right call.
    3. Review the queue. Promotional posts, jokes and anything time-sensitive are the usual candidates to edit, move or drop.
    4. Resume deliberately. When you restart, check the next week again with fresh eyes.

    What your tools should make easy

    In the moment, you want rescheduling and editing to take seconds. A shared calendar where you can see everything queued, move posts by date and leave a comment for the team is worth more than any automation. If posts need approval, a change of plan can also go back through the same approval stage, which keeps everyone aligned.

    After it passes

    Take five minutes to write down what happened: what you paused, how long, and what you would do differently. Next time the decision will be faster.

    GPTpost is a scheduler with a shared calendar and approvals: oraglegpt.org.

  • A link-in-bio page that people actually use: keep it short

    Link-in-bio pages tend to grow the way junk drawers do. Something gets added for a launch, a webinar, a seasonal offer, and nothing ever leaves. A year later the page has fifteen buttons and the one you care about is somewhere in the middle.

    Why short pages work better

    Visitors arrive from a profile with very little context and very little patience. They are looking for one thing: the product, the latest article, a way to get in touch. Every additional link is another decision before they find it. Fewer options make the important one easier to see.

    A simple structure

    1. Three to five links. If you cannot choose, that is the real problem to solve.
    2. Most important first. Usually your newest piece of content or your main offer.
    3. Labels that say where the link goes. “Pricing” or “Latest article” beats a clever slogan.
    4. One evergreen link. A home page or an about page for people who just want to know who you are.

    The monthly sweep

    Put a recurring reminder in your calendar: once a month, open the page and remove anything expired. Finished events, ended promotions, old campaign landing pages. It takes five minutes and keeps the page honest.

    Measure clicks, not just presence

    If you use short links in your posts and your bio page, you can see which links people actually click. Over a few months that tells you which spots on the page earn their place.

    Where GPTpost fits

    GPTpost includes a simple link-in-bio page and short links with click counts, so you can update the page during the same weekly session where you plan your posts. It is not a design tool, just a tidy page you control.

    Learn more about GPTpost: oraglegpt.org.

  • Thread or single post? A practical rule of thumb

    Threads, a series of connected posts, are a useful format on networks with short limits such as X, Bluesky and Threads. They are also overused. Many threads would work better as a single post, and some would work better as an article. Here is how to decide.

    When a thread earns its length

    • The order matters. Setup guides, checklists where each step depends on the previous one, and timelines.
    • There is a payoff. A story that builds toward a lesson or a surprise keeps readers moving.
    • Each part adds something. If you can delete a post from the thread without losing anything, it should go.

    When a single post is better

    • There is one idea.
    • The point fits in the first sentence.
    • The extra posts would only repeat or pad the first.

    A single, sharp post is easier to share and quote, and readers do not need to commit to reading ten parts.

    When it is really an article

    If a thread grows beyond five or six parts, or needs headings, examples and links, it is an article in disguise. Write it on your blog, then post a short summary with the link. The article is easier to find later, and you can quote it in future posts.

    Adapting by network

    Mastodon allows 500 characters by default, so an idea that needs a three-part thread on X may fit in one post there. LinkedIn allows long posts, so threads are uncommon. When you schedule across networks, write the thread where it is needed and a single post where it fits.

    Planning posts across networks with different limits is what GPTpost’s per-network composer is for. More at oraglegpt.org.

  • Would you let an AI agent schedule your social posts? The case for and against

    With MCP servers and agent-friendly APIs, it is now practical for an AI agent to draft and schedule social posts directly. GPTpost offers this, with approvals enforced for agents. But the technical possibility raises a fair question: should you let an agent near your accounts at all?

    The case for

    • Drafting is tedious. Turning a changelog into five network versions is repetitive work an agent handles well.
    • Approval is where judgment lives. If a person still signs off each post, the agent only removes busywork.
    • Consistency. An agent following a clear brief does not skip the alt text or forget the hashtags on a busy day.

    The case against

    • Tone. Models produce plausible but generic text. Reviewers may start rubber-stamping.
    • Accuracy. Invented features or numbers are a real risk, especially in product announcements.
    • Credentials. Any key given to an agent is a key that can leak.

    Guardrails that make it reasonable

    1. Approvals on. Agents should not be able to skip review. In GPTpost an agent’s post that needs approval waits for it, exactly like a person’s.
    2. Scoped keys. Give the agent only the scopes it needs, with an expiry date. A key should never have more rights than its creator.
    3. Start with routine posts. Release notes from a changelog are a safer starting point than opinion pieces.
    4. Review like it matters. Treat agent drafts like a junior writer’s work: check facts and claims.

    Where do you stand?

    We are interested in honest objections as much as enthusiasm. Tell us on any of our accounts. More on how we approach it at oraglegpt.org.