SocialPower Setup in 10 Minutes: Step by Step

You can have SocialPower working in about ten minutes. Create the workspace, fill the Brand Brain, connect one account, generate a post in AI Studio, set the daily cap to one, and leave the approval gate on. The Brand Brain is the only step worth slowing down for. Everything else is a default you can change on day three.

Three things make this go quickly. Have them open before you begin.

• The login for one social account you actually post to. Not all eight. One.
• Five to ten sentences of your own writing you are happy with — an old caption, a product page, a founder note. You will lift the voice from it.
• Two facts you always say and two words you never say. Discount codes, a founder name, a claim you cannot legally make. These become rules, not suggestions.

You do not need a card. The Free plan gives you one brand, two accounts and 30 generations a month, which is more than enough to decide whether the output is worth paying for. Prices and limits are on the pricing page.

1. 01Create the workspace and your first brand (1 minute). Name, industry, target audience. You are the owner by default.
2. 02Fill the Brand Brain (4 minutes). Voice, audience, reading level, sentence length, always-say facts, banned words, emoji policy, hashtag ceiling, CTA style, claims policy.
3. 03Connect one account (1 minute). Pick the platform, connect the account, confirm it shows as connected.
4. 04Generate one caption in AI Studio (2 minutes). Read it. If it sounds wrong, fix the Brand Brain, not the caption.
5. 05Set the daily post cap to 1 in Autonomous Mode (30 seconds). Leave the enable toggle off for now.
6. 06Confirm the approval gate is on (30 seconds). It is on by default. Leave it there.

That is the whole setup. The rest of this article is what each step is actually doing, and what you can ignore until next week.

Most AI writing tools take a prompt and a chat history. SocialPower takes a stored, structured record and applies it before generation starts. That is the difference between reminding a model of your voice every time and having the constraint built into the request. The Brand Brain is not a note in a thread. It is a set of fields the generator reads first.

Five fields do most of the work. Spend your four minutes here:

• Voice and tone. Write it as instructions, not adjectives. "Short sentences. No hype. Say the price." beats "friendly and professional".
• Audience. "D2C founders in Bengaluru doing ₹5-50 lakh a month" produces different copy from "small businesses". Be narrow.
• Always-say facts. Free shipping over ₹999. Made in Tiruppur. GST invoice on request. These get pulled in instead of invented.
• Banned words. Add the ones that make you wince. "Unlock", "game-changer", "revolutionary". Also add competitor names and any claim your category regulates.
• Claims policy. Say what you are not allowed to promise. Skincare, supplements, fintech and edtech all have lines you cannot cross, and this is where you draw them once.

The remaining fields — reading level, sentence length, emoji policy, hashtag ceiling, CTA style — take thirty seconds each and are worth setting now because they are the ones you would otherwise fix by hand on every post.

If you run several brands, each one carries its own Brand Brain and its own connected accounts. Nothing crosses between them, which is the point for an agency: a client's banned words cannot leak into another client's caption.

SocialPower publishes to Instagram, Facebook, LinkedIn, X, Pinterest, Threads, TikTok and YouTube, each at its own native dimensions and character limit. That is a reason to connect more accounts eventually, not on day one.

Connect the platform where you already post most. One connected account gives you a real end-to-end test: generate, schedule, approve, publish, check it looks right in the feed. Eight connected accounts on day one gives you eight places to discover the same voice problem. See what each platform gets on the platforms page.

AI Studio has four modes. Use captions first.

• Captions — the post copy itself, conditioned by the Brand Brain.
• Hashtags — scored by competition, so you can see which tags are crowded before you use them.
• CTAs — written in the CTA style you set, not a generic "link in bio".
• Ideas — content angles when you are stuck, researched against your brand rather than pulled from a generic list.

Read the first caption critically. If it uses a word you hate, the fix is the banned words field. If it sounds too formal, the fix is the voice field. Editing the caption fixes one post; editing the Brand Brain fixes every post after it. This is the habit that makes the tool worth its price, and it takes two or three rounds to settle.

One honest note on providers. You can point a workspace at OpenAI, Claude, Gemini, Groq, Mistral, DeepSeek or HuggingFace. If you have not set a key yet, a deterministic mock provider answers instead of the app hard-failing on config. That is good for clicking through the flow. It is not good for judging caption quality — set a real provider before you decide the output is bad.

Autonomous Mode has two safety mechanisms, and they behave differently.

The daily post cap is enforced. The worker counts today's AI-generated posts for the workspace and stops when it hits your number. Set it to 1 while you are learning what the output looks like. You can raise it once you have approved a week of drafts without editing them.

The approval gate is on by default. With it on, autonomous generation fills a queue and nothing reaches a live account until a person releases it. Leave it on. There is no prize for turning it off in week one, and the cost of a bad caption going out under your brand name is higher than the ten seconds approving costs you.

There is also a monthly spend cap field. It is stored and displayed, but the worker does not enforce it. Treat it as a note to yourself, not a limit. The daily post cap is the control that actually stops work.

Payment is UPI, card or net banking. GST at 18% is added on top of the listed prices.

Every setup guide tells you to fill everything in. Most of it can wait, and some of it does not exist yet.

Open the queue and read the drafts before you approve them. For each thing that is wrong, there is one field that fixes it permanently:

1. 01Wrong words — add them to banned words. Do this every time you flinch, not once a month.
2. 02Wrong facts, or a number you did not recognise — add the real number to always-say facts. Anything missing from that field is something the model has to guess at.
3. 03Too long, or too dense — drop the sentence length and reading level. Indian audiences reading on a phone in their second or third language reward short sentences.
4. 04Weak ending — rewrite the CTA style field with an actual example of a CTA you would send. One example beats a description.
5. 05Too many hashtags — lower the hashtag ceiling. The scored list tells you which ones are crowded; the ceiling stops the generator reaching for all of them.

Three mornings of that and the queue stops needing edits. That is the point at which raising the daily cap from 1 to 3 is a decision rather than a gamble. If you want the mechanics behind each step, the how it works page walks through the generation pipeline in order.