Lyncera

It never touches LinkedIn on your behalf.

Know who's worth your time.

Open any LinkedIn profile and Lyncera tells you whether this person is worth an hour — and writes the message if they are. No automation, no spam: you stay in control.

Start free Everything it does

14 days free. No credit card. Nothing is deleted when it ends.

The Lyncera dashboard, showing job changes, follow-ups due today and fit-score distribution.

Specifications, not performance claims

Every number a prospecting tool usually puts here is a number it cannot measure. So these are facts about the software instead — and each one is checked against the code by a test.

135
LinkedIn industries mapped: 100
Points in a fit score: 20
Pipeline stages: 12
LinkedIn languages it reads: 800+
Automated tests per release: 0
Actions taken as you

The moment

This is where it happens

Not in a dashboard. On the profile you already have open, in a side panel, in about twenty seconds. Watch it once and you have seen the product.

linkedin.com/in/aino-virtanen

Lyncera

You open a profile you were going to look at anyway.Example

One pass

From a profile to a sent message

One pass, about a minute.

Profile
You open a profile you found yourself.
Research
Public sources, with citations you can check.
Fit
A score out of 100, and the reasoning behind it.
Draft
Three angles, grounded in real facts, in your voice.
Follow-up
Scheduled before you close the tab.

The six parts

Everything it does

Six parts, one workflow. Pick any of them.

It works where you already are

A side panel on the profile you opened. Nothing is read until you press the button, and nothing is ever sent on your behalf.

Reads the visible page: name, headline, about, experience, skills
Recognises anyone already in your workspace, so you never pay twice
Alt+L opens it without reaching for the mouse
Saves a person without spending a credit when you only want the record

The extension

Everything in the box

The whole desk, not just the moment of discovery

Finding somebody worth an hour takes a minute. Everything after it is what decides whether the hour turns into a conversation — so all of it is in the product, not on a roadmap.

Capture

Getting a real record out of a page you are already reading.

A side panel on the profile: Reads the page you opened when you press the button, and not a moment before. Alt+L opens it without the mouse.
Companies from a page or a link: Name, industry, location, size and logo — read from the fact line rather than by position, so a tagline never lands in the industry field.
CSV import with a mapping step: Your columns, your delimiter, a preview of what will be written before a single row is.
Work history and skills, dated: Stored as structured rows with start and end months. That is what makes role fit computable later without spending anything.
Photos that keep working: Copied into private storage, because LinkedIn's own image links expire within days and a broken avatar is a broken record.
Duplicate detection and merge: The same person arriving twice from two sources becomes one record, with both timelines kept and nothing silently dropped.

Decide

Whether this one is worth the hour, and why.

A fit score out of 100: Six components, each shown next to the field it was read from. The model supplies signals; code does the addition.
Company fit, scored the same way: From stated facts and observed signals, with the same breakdown — not a vendor-wide model you are not allowed to inspect.
Research with citations: Public sources, each with a link you can open. When there is nothing to cite it fails loudly instead of writing something plausible.
Role fit against an open role: Computed in your browser from stored skills and tenure. It costs nothing, it cannot go stale, and it refuses to score a role that states no requirements.
Smart segments: Saved filters that live on the server, so everyone on the desk is looking at the same list and not their own copy of it.
Bulk analysis: Score a whole list in one pass, with the exact credit cost shown before it starts and refunded for anything that fails.

Reach out

A first message that reads as though a person wrote it.

Three drafts, three angles: Genuinely different openings, each built from something the profile actually says rather than three rewrites of one template.
Blueprints that hold your structure: You write the shape of the message once. The AI fills it from the person in front of it and leaves your structure alone.
Proof points you wrote: The claims in a draft come from a list you maintain. The model quotes them and never invents a new one.
Phrases to avoid: Words you never want in a first message are honoured on every draft, including regenerations.
Versions and outcomes: Every regeneration is kept, and the one you sent carries what happened to it. That is where a reply rate can honestly come from.
Reply capture: You paste back what came in. The timeline and every rate on the dashboard are built from that, and say so where they are shown.

Run the desk

The unglamorous half, which is the half that compounds.

Follow-ups on a real calendar: Month and week views, drag to reschedule, real durations, and an undo on every move you make.
Subscribe from Google or Outlook: A private iCal feed you can rotate at any time, so follow-ups appear beside your meetings instead of in a second tool.
Today, as one ranked list: Overdue first, then what closes on its own if you do nothing. It is a worklist, not a dashboard to admire.
Job-change alerts: A tracked contact moving to a new company reaches you the week it matters, when the message writes itself.
A data-health report: What is missing, what has gone stale, and what is worth fixing first — counted, not estimated.
API keys and webhooks: Scoped keys and outbound events, so what you build on top runs on your side of the LinkedIn boundary and stays there.

Twenty-four things, all shipped. None of this is behind a waitlist and none of it is an integration you have to buy separately.

The score tells you when to walk away.

Most profiles are not worth an hour. Lyncera reads the one on your screen, weighs it against what you sell and who you sell to, and shows a fit score with the reasoning laid out. The time it saves is mostly in the people it tells you to skip.

Fit scoring is hybrid: the model produces signals, deterministic code computes the number. It never returns an unchecked total.

The number

A score you can take apart

The model produces signals. Deterministic code turns them into the number, and the number is the sum of the parts you can see. Nothing rounds up in the dark.

0Fit score

Example breakdown

ICP match26/30
Role relevance17/20
Company fit15/20
Buying signals9/15
Timing7/10
Data confidence4/5

26 + 17 + 15 + 9 + 7 + 4 = 78. Always.

Messages that sound like you

Three genuinely different drafts, built from what the profile actually says and the way you told Lyncera you write.

Nothing gets dropped

Follow-ups land in your calendar. A tracked contact changing jobs reaches you the week it matters.

Numbers you can trust

Every figure is counted from your own workspace. Lyncera never shows an open rate it cannot measure.

Recruiting mode

If you place people, the whole thing changes shape

One answer at signup swaps the vocabulary. A profile stops being measured against an ideal customer and starts being measured against an open role — and the pipeline gets the stages a search actually has.

Roles carry their own requirements

Must-have and nice-to-have skills, a minimum number of years, seniority and location. Paste in a job description and the form fills itself.

Role fit is deterministic

Computed in your browser from stored skills and dated history. No credit, no cache, and it cannot be stale — because it is recalculated every time you look.

Three kinds of record, three filters

A candidate, a client and a hiring contact are not the same thing, do not belong in the same list, and are never counted together.

A timer for the ones you never contacted

Twelve months, then the record goes — unless somebody is actually working it, in which case nothing expires under them.

Two pipelines, one product

Sales

New
Researching
Ready
Contacted
Replied
Qualified
Meeting
Won
Lost

Recruiting

New
Researching
Ready
Contacted
Replied
Screening
Submitted
Interviewing
Offer
Placed
Lost

The first 5 stages are the same in both. These stages are read from the code that runs your workspace, not typed into a slide. Which set you see follows the answer you gave at signup.

The alternatives

Three ways to prospect on LinkedIn

Two of them are common, well funded, and carry a cost that is easy to miss until it lands. Here is the honest version of all three.

Comparison	Automation tools	Data platforms	Lyncera
How it finds people	Search filters, then a queue that runs on its own	A database someone else assembled and resold	The profile you opened, because you decided it was interesting
What touches LinkedIn	It does, as you: connects, messages, views at scale	Enrichment calls and scraped snapshots	Nothing. It reads the page already rendered in your browser
What the score means	Usually none: volume is the strategy	A vendor-wide model you cannot inspect	Signals from this profile, totalled by code you can read the breakdown of
Whose account is at risk	Yours. Automation is against the platform's terms	Not yours, but the data provenance may be	Nobody's. There is no action attributed to your account
What it says when it does not know	Sends anyway	Returns a confident-looking blank	Refuses, names what is missing, and does not charge you
Who sends the message	The tool, on a schedule	Whatever you connect it to	You, from your own account, after reading it

How it finds people

Automation tools: Search filters, then a queue that runs on its own
Data platforms: A database someone else assembled and resold
Lyncera: The profile you opened, because you decided it was interesting

What touches LinkedIn

Automation tools: It does, as you: connects, messages, views at scale
Data platforms: Enrichment calls and scraped snapshots
Lyncera: Nothing. It reads the page already rendered in your browser

What the score means

Automation tools: Usually none: volume is the strategy
Data platforms: A vendor-wide model you cannot inspect
Lyncera: Signals from this profile, totalled by code you can read the breakdown of

Whose account is at risk

Automation tools: Yours. Automation is against the platform's terms
Data platforms: Not yours, but the data provenance may be
Lyncera: Nobody's. There is no action attributed to your account

What it says when it does not know

Automation tools: Sends anyway
Data platforms: Returns a confident-looking blank
Lyncera: Refuses, names what is missing, and does not charge you

Who sends the message

Automation tools: The tool, on a schedule
Data platforms: Whatever you connect it to
Lyncera: You, from your own account, after reading it

Described as categories rather than named products: a competitor's feature list changes weekly, and a claim you can rebut is worth less than no claim. Every line in the Lyncera column is something you can check in the product.

Your existing stack

It fits where you already work

Four surfaces, all of them ones you already have open. Nothing here asks you to move your work into a new place and live there.

Chrome, on the page you opened

A Manifest V3 extension with a side panel. Pairing is a short code you approve in the web app — never a password you hand to an extension.

Your calendar, by subscription

A private iCal URL, rotatable on demand. Google Calendar, Outlook and Apple Calendar all show your follow-ups with the right duration.

Excel and Sheets, correctly

UTF-8 with a byte-order mark, CRLF endings, the delimiter your locale expects and a guard against formula injection. Finnish Excel opens it without a wizard.

Your own automation

Scoped API keys and outbound webhooks. Whatever you build sits on your side of the LinkedIn boundary, which is the only side there is.

Storage and access

Where your data lives, in plain terms

This is the section most tools write in legal language. Here is the short version, in the words the code would use.

One workspace cannot see another: Row-level security in Postgres, and a workspace filter in every handler on top of it. Two independent locks, because one lock is a single point of failure.
Extracted fields, never raw HTML: Name, headline, history, skills, location. The page itself is read in memory and discarded — there is no archive of LinkedIn on our disks.
A retention window you set: Records nobody ever acted on expire on your schedule. Anything actively being worked is exempt, so a timer never deletes live work.
Every write is in the audit log: Who changed what, when, and from which surface — written in the same database transaction as the change, so the two cannot disagree.
Export and deletion, on request: A full export of your workspace whenever you want it, and a deletion request that actually deletes rather than hiding a row.
Secrets never reach a browser: The database service key, the payment keys and the AI key are server-only, and the build fails if one is imported into client or extension code.

The receipts

Built to be checked

Fit scores are computed in code, not returned by a model
Research fails loudly when it finds nothing to cite
Every AI run is recorded with its real token cost
Exports open correctly in Finnish and English Excel

The line

It never touches LinkedIn on your behalf.

No background crawling, no automated connection requests, no scraping queues. Lyncera reads the page you opened, when you press the button. You copy the message and send it yourself — which is why your account is never the thing at risk.

Four things it will never do

Send a connection request or a message for you
Run a queue of profiles while you are asleep
Store the raw HTML of a page it read
Show you a number it cannot compute from your own data

Before you sign up

Questions worth asking first

Does Lyncera automate anything on LinkedIn?

No. No background crawling, no automated connection requests, no scraping queues, no message sending. It reads the page you opened when you press the button, and you copy and send the message yourself. That boundary is the design, not a limitation waiting to be lifted.

Is my LinkedIn account at risk?

Nothing acts as you. Lyncera never signs in to LinkedIn, never holds your LinkedIn credentials, and never performs an action attributed to your account. The extension reads the page already rendered in your browser.

What does it store about the people I look at?

Extracted fields: name, headline, role history, skills, location, plus a copy of the profile photo in private storage because the original link expires. Never raw page HTML. You can set a retention window and delete a record at any time.

Does it work for recruiting as well as sales?

Yes, and it is asked at signup. Turn recruiting on and you get open roles, candidate scoring from stored skills and tenure, recruiting pipeline stages, and a retention timer for candidates you never contacted.

Do I have to install the extension?

The web app works on its own: import contacts from CSV, add companies from a website link, research, score and draft. The extension is what turns a profile you are already reading into a record in one click.

What does the trial include?

Fourteen days, no credit card, and nothing is deleted when it ends. You keep every contact, message and note; the AI actions pause until you choose a plan.

What does an AI action cost me?

Credits, reserved before the run and refunded automatically if it fails. Research, fit scoring and drafting spend them; role fit, segments, imports, follow-ups and the calendar never do, because none of those needs a model. The cost of a bulk run is shown before it starts.

Can a team share one workspace?

Yes. Shared lists, shared segments, shared templates and analytics rolled up across the desk. Every record belongs to the workspace, and row-level security plus a filter in every handler is what stops one workspace seeing another.

Does it work properly in Finnish?

End to end: the app, the extension, the notification emails and the CSV exports. A Finnish LinkedIn also fills an English workspace correctly, because the industry vocabulary is mapped both ways rather than stored as whatever the page happened to say.

What happens if the AI is unavailable?

It says so plainly and charges you nothing. Everything deterministic keeps working — role fit, segments, imports, duplicate detection, follow-ups, the calendar feed and every export — because none of it ever needed a model to run.

Start with the next profile you open.

Free plan, no credit card. The Chrome extension takes a minute to set up.

Start free

14 days free. No credit card. Nothing is deleted when it ends.