LotsSocial
No. Not by default. But yes — after you teach it.
Nothing goes live without you.
The agent doesn't know your voice until you give it something to work with. Four inputs shape what it writes: a one-page brand brief, 5–10 past posts as samples, a list of words you never use, and tone rules per platform. That's it. No 20-page style guide required.
The 5-minute reality check
Open the agent. Send a one-line brand brief: "We're a direct-to-consumer coffee brand. Warm, specific, no fluff. Think Blue Bottle meets your favorite barista."
The first draft comes back 60–70% there. The last 30–40% comes from correction loops. You don't need to be a prompt engineer. You just need to say what to keep and what to cut in plain words.
What "voice" actually means for a social agent
The agent reads four concrete inputs:
- Brand brief — who you are, who you sound like, words you use, words you ban, default tone by platform
- Sample posts — 5–10 past posts that actually sound like you
- Banned words — the clichés, buzzwords, and phrases that make you cringe
- Platform tone rules — "less formal on X, more formal on LinkedIn, never use the word game-changing"
This ties directly to the platform-specific captions feature. The agent adapts the core message for each platform instead of copying the same caption everywhere.
The voice-training loop, 3 steps
Step 1: Drop 5–10 past posts into the brand brief as samples.
Pick posts that performed well and actually sound like you. Not the ones a freelancer wrote. Not the ones you rushed. The ones where a customer replied "this is exactly why I follow you."
Step 2: Tell the agent what to keep and what to cut in plain words.
"Keep the conversational opener. Cut the hashtag stack. On LinkedIn, lead with the insight. On Instagram, lead with the feeling. Never use 'game-changing,' 'revolutionary,' or 'unlock.'"
Step 3: Review the first 5 drafts. Leave inline notes. Approve the ones that match.
Each cycle narrows the gap. By draft 5, the agent is usually at 90%+. You're not rewriting — you're correcting course.
Why the first drafts won't sound like you (and why that's fine)
The agent has no memory of past conversations by default. It relies on the brand brief you wrote, not on vibes. It doesn't "learn" from your Slack messages or your email tone. It learns from what you explicitly put in the brief and what you correct in the drafts.
The fix is a one-page brand brief. Not a 20-page style guide.
What to put in your one-page brand brief
Six blocks. Fill each in 10 minutes total.
| Block | What to write | Example |
|---|---|---|
| Who we are | 2 sentences | We roast specialty coffee in small batches. Shipped fresh every Monday. |
| Who we sound like | 1–2 reference brands | Blue Bottle's clarity. Your favorite barista's warmth. |
| Words we use | 5–10 words/phrases | Small-batch. Monday roast. First sip. Morning ritual. |
| Words we never use | 5–10 words/phrases | Game-changing. Revolutionary. Unlock. Leverage. Seamless. |
| Default tone by platform | One line each | X: conversational, short. LinkedIn: insight-led, professional. Instagram: feeling-first, visual. Threads: casual, behind-the-scenes. |
| One sentence that should never appear | Your hard line | "We're the best coffee in the world." |
Copy this template. Fill it. Paste it into the agent. Done.
How agencies handle multi-client voice
Per-workspace brand brief. Each client lives in its own workspace with its own one-pager. No cross-contamination. Workspaces keep clients separated — the agent in Workspace A never sees Workspace B's brief, samples, or corrections.
You set up the client's brand brief once. The agent drafts in their voice. You review. You approve. Nothing goes live without you.
What to do when a draft still sounds off
Four specific corrections that work better than vague notes:
| Vague note | Specific correction | Before | After |
|---|---|---|---|
| "Too formal" | "Replace 'We are pleased to announce' with 'Here's what's new'" | We are pleased to announce our new seasonal blend. | Here's what's new: our autumn blend drops Monday. |
| "Too salesy" | "Kill the cliché. Say what it actually is." | Unlock the ultimate coffee experience. | Our autumn blend: maple, cinnamon, and a hint of smoke. |
| "Too long" | "Shorten by 30%. Keep the hook." | [3-sentence paragraph about roasting philosophy] | We roast Monday. You sip Tuesday. That's the whole system. |
| "Wrong platform tone" | "Switch platform tone per the brief." | [LinkedIn post written like an X thread] | Most teams overthink their content calendar. Here's the 4-line brief that plans a month. |
When the agent should not draft at all
Off-limits topics. Sensitive announcements. Anything requiring your lived experience — the founder story, the customer crisis you navigated, the opinion only you can give.
Reinforce: nothing goes live without you. Manual Approval stays the right permission level until you trust the voice. One week at Manual Approval. Then decide.
FAQ
How long until the agent sounds like my brand?
Most people hit 90%+ match by draft 5. That's one week at 1 post per day, or two weeks at 3 posts per week. The variable is how specific your corrections are.
Can I give it more than a one-page brief?
You can. But the agent pays most attention to the brand brief, sample posts, and your inline corrections. A 20-page PDF usually gets ignored. Keep it to one page.
What if my brand voice changes?
Update the brand brief. Add new samples. The agent reads the current brief every time it drafts. It doesn't remember the old one.
Does the agent remember past corrections between sessions?
Only if you keep the same workspace and don't clear the brief. Corrections live in the draft history. The agent references them when you ask for a new draft on the same topic.
Meet your agent → https://agent.lots.social. Default first-week permission level is Manual Approval, so every post waits for your review. Tell your agent what your brand sounds like and watch the drafts come back in your voice.