Skip to content
How it works

Speakfold prepares everything. Your people pick and approve.

Behind every post is an agent that knows your company, each person’s expertise and voice, and what has worked on LinkedIn so far. Here is the week, and what makes it work.

Start free
The week

How one story travels through your team.

  1. What’s happening

    You say what’s happening

    A launch, an expansion, a hire, a customer story or a point of view, in one sentence, with a link if there is one. Speakfold takes the facts from the page, follows the recipe for that kind of story and picks the best person for each part, with the reasons. Add anyone you like and Speakfold suggests their part.

    Speakfold’s storyboard · Point of view

    Why AI pilots stall before the model

    1. MonThe shiftMarcus LindCardPoint of view posts get Marcus 1.3× the usual reach · CEO fits “the shift”
    2. TueThe approachAino VirtanenTextHead of Data fits “the approach”
    3. WedProof from the workJonas BergPhotoWrites about AI pilots (1.7× the usual reach) · Was in a company story in the last two weeks
    4. ThuWhat it means for customersSofia ReyesTextSales Director fits “what it means for customers” · Hasn’t been in a company story yet
    5. FriA real exampleLeena KoskiPhotoAdded by Marcus · Customer Success Lead fits “a real example”
    Ask them to joinAdd a part
  2. Their choice

    Each person is asked first

    They see their part, the angle and why Speakfold picked them. They say yes, ask for another angle or pass, and Speakfold asks the next best person. Nothing is written until they join.

    Speakfold asks · Point of view

    Join “Why AI pilots stall before the model”?

    Your part · Proof from the work · Wednesday41 definitions and how we got to oneSpeakfold will ask: What did you find in week three of the pilot?
    • Writes about AI pilots (1.7× the usual reach)
    With Marcus, Aino, Sofia and Leena
    JoinAnother angleSkip
  3. Sixty seconds each

    They answer one question

    Spoken or typed, in English or Finnish. One real detail from their own work is what makes a post worth reading, so that is the only thing Speakfold asks for.

    One question for Jonas · 60 seconds

    What did you find in week three of the pilot?

    0:42

    So we pulled every table that had a customer status in it, and there were forty-one definitions of active. Sales had one, finance had three

    Talk or type. English or Finnish.Use my answer
  4. Same day

    Speakfold drafts it in their voice

    Built from how they write, what they know and what the company stands for. Each post picks up where the previous one left off. Speakfold never invents a fact, and every draft passes the human check before anyone sees it.

    Speakfold drafted · in Jonas’s voice
    Jonas Berg
    Solutions Architect
    41 versions of “active customer”. That is what we found in week three of the last pilot. Aino’s point about owners is the reason. Here is how we got it down to one.

    Reads like Jonas.

    No AI tellsBuilds on Aino’s postOne real detail: 41 definitions
    Removed by the human checkIn today’s fast-paced data landscape, it’s important to note
  5. Their call

    They approve, and it goes out when they say

    Pick another opening, rewrite a line or approve as it is. Speakfold suggests the time and a visual: a card or carousel in your brand, their own image or a photo of them. Then they press Post, pick the time or post it themselves. It goes out from their own profile, so it is theirs in every way that matters.

    Ready when you are
    Jonas Berg
    Solutions Architect
    Wed 9:10
    41 versions of “active customer”. That is what we found in week three of the last pilot. Aino’s point about owners is the reason. Here is how we got it down to one.
    Approve for Wed 9:10Rewrite a line

    Jonas presses Post or picks the time. Colleagues approve their own comments.

  6. The first three hours

    Colleagues comment, in their own voices

    Early comments decide how far a post travels. Once a post is approved, each colleague gets a comment written in their own voice. They approve it (or edit it, or pass), and Speakfold posts it from their LinkedIn at its own moment in the first three hours, so the comments don’t all land at once.

    2:21
    First three hours · Jonas’s post is live

    Your comment, Sofia, in your voice.

    Approve commentEdit itSpeakfold posts it at 10:24
  7. Afterwards

    Results, and what it led to

    Impressions, comments, shares and saves for every post and the whole story. Each person’s own link shows who brought people in, and you add what it led to: applications, demo requests. Speakfold learns what worked for each person and plans the next story with it.

    Story results · 5 posts
    18,420impressions
    212comments
    96link clicks
    4demo requests
    • Mon2,940
    • Tue3,610
    • Wed5,880
    • Thu3,150
    • Fri2,840

    Posts that told a specific moment did 2.1× better. Next story leads with one.

Try it

Tell Speakfold what’s happening. It picks who tells each part.

Every kind of news has a recipe: an expansion starts with why now and ends with the roles you’re hiring for. Speakfold picks the best free person for each part, from their role, their topics, what has worked for them and how often they have carried the company lately.

  • The facts from your link, used exactly as written
  • Sign-off and a launch day when a story needs them
  • Add anyone, and Speakfold suggests their part

We’re opening in Stockholm in January, starting with a team of four.

Speakfold’s storyboard · 4 parts, each with a reason
  1. MonWhy nowSofia ReyesPhotoSales Director fits “why now” · Hasn’t been in a company story yetWhy this market, why this year, what the company saw
  2. TueWhat it means for customersLeena KoskiCardCustomer Success Lead fits “what it means for customers” · No company story in over five weeksWhat changes for customers there, from someone who talks to them
  3. WedWho builds it thereAino VirtanenPhotoHead of Data fits “who builds it there”The people and the first weeks on the ground
  4. ThuWe’re hiring thereMarcus LindCardExpansion posts get Marcus 1.5× the usual reach · CEO fits “we’re hiring there”The roles open there and who should apply
Expansion: Why now, Sofia Reyes; What it means for customers, Leena Koski; Who builds it there, Aino Virtanen; We’re hiring there, Marcus Lind

Speakfold’s real recipes and casting, run on an illustrative team. Everyone is asked first and can say yes, ask for another angle or pass.

Stories

Start from the news. Speakfold does the casting.

Pick the kind of story and say it in one sentence. Add a link to the press release, product page or job ad, and Speakfold takes the facts from it. Each kind of story has a recipe you can edit, and Speakfold picks the best free person for each part with the reasons.

  • Drafts use the source’s facts exactly as written
  • Sign-off for launches, or any kind of story you choose
  • A launch day: nobody can post their part before it
  • Add anyone to a story, and Speakfold suggests their part
Speakfold read the press release · 4 facts
  • Opening in Stockholm in January 2027
  • A team of four to start, growing to ten by the end of the year
  • Two logistics clients signed before the opening
  • Hiring two data engineers by March

Drafts use these facts exactly as written, and nothing else.

Launch day
Mon 11 Jan. Nobody can post their part before it.
Sign-off
Marcus gives the OK on each post first.
The plan

The next 13 weeks, at a glance.

Company stories, own posts and ideas, a week at a time. Speakfold notes when the mix tips, like three points of view in a row or an empty week. It looks at your website every day, and a new job ad or product page becomes a story ready to plan, with the cast already picked.

The next 13 weeks
  1. This weekPoint of viewWhy AI pilots stall before the model5 posts
  2. Week of 12 JanExpansionInto Stockholm4 posts
  3. Week of 19 JanCustomer storyReporting time cut in half
  4. Week of 26 JanNothing yet
  5. Week of 2 FebHiringTwo data engineers in Helsinki

Three points of view in a row last month. A customer story would balance it.

New on your website: Senior Data Engineer, Helsinki. A hiring story is ready to plan, with Marcus, Aino and Daniel.
What Speakfold knows

Four things Speakfold learns, so drafts start close to done.

The company brain

Add your website and any decks or one-pagers. Speakfold fills in what you do, who you sell to, key messages, proof, topics and how you sound. When it finds something new, it suggests a change and an admin picks.

Speakfold read your website · 14 pages, 2 decks
What you do
Data and AI consulting for mid-size B2B companies.
Who you sell to
Heads of data, CFOs and COOs.
Topics
Data ownershipAI pilotsAnalytics teams
How you sound
Plain, specific, a little dry. Numbers over adjectives.
New case study on the site. Add “cut reporting time by half” to proof?Add

Each person’s expertise

Seven questions, answered out loud or typed. Speakfold builds an expertise profile with their beliefs, method, stories and proof, and every idea and draft builds on it.

Expertise interview · 7 questions
  1. What do you help people with? Say it the way you’d tell a friend.
  2. What do you believe about your field that a lot of people get wrong?
  3. Walk me through how you actually do the work. What are your steps?
  4. Tell me about one moment from work you still think about. What happened?
  5. What results can you point to? Numbers, before and after, anything you’ve seen.
  6. A year from now, what should people come to you for?
  7. What should your posts never sound like?

Each person’s voice

From their LinkedIn writing, when they allow it, and the way they talk in the interview. Sentence length, openings, the words they reach for and the ones they would never use.

How Aino writes
  • Short paragraphs, two lines at most
  • Opens with what happened, then why it matters
  • “We” for the team, “I” for opinions
  • Dry humour, never exclamation marks

From her LinkedIn writing and her interview

What works

A year of LinkedIn analytics, matched to every post. Speakfold finds what each person writes about and which topics and kinds of posts work for them, says where they could be known for more, and turns strong patterns into friendly checks on new drafts.

Speakfold’s read for Aino · 212 posts, 12 months
  • Telling a specific moment2.1×
  • Lessons learned about data ownership1.8×
  • Data ownership, any kind of post1.6×
  • Hiring postsearly1.4×
  • Industry news0.7×

Aino wants to be known for analytics teams and posted on it twice this year. Speakfold plans it as a lesson learned, the kind that works for her.

Each post is compared with her posts from around the same time, so a growing audience doesn’t flatter the newest ones. Under six posts is an early signal.

The agent

Every 15 minutes, Speakfold does the prep.

  • Prepares openings for upcoming posts
  • Writes each colleague’s comment in their voice once a post is approved
  • Posts approved comments and likes from each person’s LinkedIn, spread over the first three hours
  • Asks the whole team to like it, with one tap that opens the post
  • Refills story ideas and each person’s own-post ideas
  • Nudges the author when an approved post is due
  • Rebuilds what works when new numbers arrive
  • Finds each person’s topics in their posts
  • Prepares a visual for every approved post
  • Looks at your website once a day for news
  • Looks at who engaged with each live post, when you turn it on
Speakfold is workingevery 15 minutes
  1. 08:32New on your website: a job ad for a Senior Data Engineer. A hiring story is ready to plan
  2. 08:19Posted Sofia’s approved comment from her LinkedIn, 41 minutes after Jonas went live
  3. 08:06Asked Aino one question for Tuesday’s post
  4. 07:53Marcus’s post is approved. Wrote a comment for 4 colleagues, each in their own voice
  5. 07:40Drafted three openings for Jonas’s Wednesday post
The human check

Nothing reaches a person with an AI tell.

Generated text is cleaned before anyone sees it, and approval checks again. A red tell blocks approval. Amber notes are suggestions.

Blocked
Em dashes and dashes used as pauses, words like delve, leverage and seamless, stock phrases, “not X, it’s Y”, opening with “I”, more than two hashtags, and any word your company bans.
Noted
Emoji runs, every sentence the same length, the rule of three, details still in [brackets].
The human check · try it

Reads like AI. 5 things to fix.

Em dashAI vocabulary: delve, leverage, unlock, seamlessStock phrase: “in today's fast-paced”, “thoughts?”“Not X, it’s Y” framing4 hashtags

The same check runs on every Speakfold draft. A draft with a red tell can’t be approved. Nothing you paste here leaves your browser.

Visuals

A visual for every post, in your brand.

Speakfold reads your logo, colours and fonts from your website. After approval it suggests what to add: a card or carousel it designs, the person’s own image, or a photo of them, which they add in LinkedIn. People edit the words and download a PNG or a PDF.

Stat card · for Jonas’s post
Quote card · for Sofia’s post
Carousel · for Marcus’s post

Drawn by Speakfold’s own designer with an illustrative brand. Numbers on a card only ever come from the post.

Results

See how it’s going, and what it led to.

The team page shows the team brand index, weekly reach split into own posts and company stories, how each story did and who is set up. Everyone posting a story gets their own tracked link for the first comment, so you see who brought people in, then the applications or demo requests that followed.

Team reach · 12 weeksOwn postsCompany stories
Team brand index 6811 of 14 people posting
Where people went · clicks per person
  • Marcus31
  • Aino18
  • Jonas27
  • Sofia14
  • Leena6
What it led to4 demo requestsAdded by Marcus

Each person’s own link, with UTM tags. Crawlers and link previews don’t count.

Who engaged

See who your posts reach, and who fits your buyers.

Turn it on, and Speakfold looks at the reactions and comments on every live post after 3 hours, a day, 3 days and a week. It leaves your colleagues out, reads each person’s headline against who you sell to and says why they fit. The author hears in Slack, with a reply to the comment written in their voice.

  • Fit your buyers, maybe, or everyone, with the reasons
  • Not a fit teaches Speakfold who you don’t sell to
  • Send to your CRM, or download a CSV
  • Ideas say which topics bring people who fit
Who engaged · Jonas’s post

14 people engaged, 4 fit your buyers

  • Maria LindqvistFits your buyersHead of Sales, Polar LedgerSales: a function you sell to · Director or headCommented on Jonas’s post
    The part about agreeing on one definition first is exactly where our last rollout stalled.
    Write my reply
  • Henrik AalborgFits your buyersFounder and CEO, Fjordline SoftwareLeadership: a function you sell to · C-levelReacted to Jonas’s and Sofia’s posts
  • Sanna HeikkiläMaybeHead of Growth, Tuuli HealthDirector or head: a level you sell toLiked Jonas’s post

Illustrative people. Only public reactions, comments and headlines, kept 90 days.

Getting started

Up and running in an afternoon.

  1. 1

    Sign up with your website

    Speakfold reads it while you check your email and drafts the company brain and your brand kit: what you do, who you sell to, proof, your logo, colours and fonts. Then it looks there for news every day.

  2. 2

    Paste your team

    Names and work emails, straight from a spreadsheet or a chat. Speakfold emails each person their own link, sends one reminder and tells you who hasn’t started.

  3. 3

    Three minutes each

    Everyone picks what they want to be known for and leaves with their first post written. Their LinkedIn files and seven questions about their work come after, when they have a minute, and make every post sound more like them.

Ground rules

What Speakfold will never do.

  • Post or comment without a yes. A post goes out only when its author presses Post or picks the time. A comment or like goes out only when that colleague approved it, one by one or once in Settings, and they can take it down.
  • Invent a fact. Missing details become a question in [brackets].
  • Rank colleagues. The brand index compares each person with themselves. The team sees shared wins.
  • Turn a feed into an ad. By default, company stories are at most one in three of a person’s posts.
  • Let a launch out early. With a launch day set, nobody can post their part before it.
  • Dig up the people who react. Who engaged keeps only public reactions, comments and headlines, never looks up an email or phone number, and forgets them after 90 days.
  • Keep photos of your people. A photo added to a scheduled post is deleted once it goes out.

See your own week planned by Speakfold.

Book a demo and we’ll walk through a week of Speakfold with your team and your story in mind.

Start free