Ask your business a question, and see how the answer was worked out
AtlasLinq Anchor is a desktop app for the questions that usually wait for whoever is good with spreadsheets. Which events produced the deals, who owes you past sixty days, what sold last quarter. Ask in plain English and get a table or a chart, with the query that produced it underneath.
Ask it something
Three questions, answered the way the app answers them: the result, a chart, and the query underneath, open by default
Which events produced the most qualified leads this quarter?
| event | qualified |
|---|---|
| Expo North | 18 |
| Retail Summit | 11 |
| Founders Meetup | 6 |
SQL
SELECT event, count(*) AS qualified
FROM tapcard_leads
WHERE status = 'qualified'
AND created_date >= date_trunc('quarter', current_date)
GROUP BY event
ORDER BY qualified DESC;3 names masked before sending
What you can ask
Your AtlasLinq leads and Trade Ledger, any CSV or Excel file including your CRM export, and your own notes and documents
Sales and pipeline
- Which lead source converts best?
- Which open deals have gone quiet?
- Win rate by team, on closed deals only
Money owed
- Which payments are overdue?
- How much has this customer been billed, and how much have they paid?
- Which accounts are past sixty days?
Anything in a spreadsheet
- What sold most last quarter, by revenue?
- Which customers ordered this year but not last?
- Monthly totals for the last year, as a line chart
Your notes and documents
- Which leads were worried about price?
- What discount do we give charities?
- How do we usually answer “it’s too expensive”?
When your data cannot answer a question, Anchor says so. Ask for the pipeline total on a file with no deal values and it tells you the column is missing, rather than inventing a figure.
Who it is for
The people who need the numbers, and the people who work the leads
Owners and managers
The weekly numbers without waiting for someone to build them. Sales by source and by team, which events paid for themselves, who owes you and for how long.
- Every Monday at 9, give me the pipeline review
- Which team has the most wins this quarter?
- How much is overdue, split by age?
Reps and outreach
The list of who to call next, and why. Anchor reads your leads and notes, so it can tell you who went quiet, who asked about pricing, and what they said at the stand.
- Who should I follow up with this week?
- Who did we meet at the expo, and what did they say?
- Draft a follow-up to the second one
Three ways to run it
You choose where the model runs. Every reply lists what was masked and what went as written, so the masking is something you can see
| Model | Where it runs | What leaves your laptop | What we charge per question |
|---|---|---|---|
| Built-in model | On your laptop, after a one-time download of about 3 GB. | Nothing. | Nothing. |
| Your company’s server | On a server your company already runs, such as Ollama, LM Studio or llama.cpp. Nothing large to download. | Your question and the rows it needs, to your own server and no one else. | Nothing. |
| A frontier model | With the provider you choose, such as Anthropic, OpenAI or Gemini, on your own key. | Your question and the rows it needs, with names, companies, emails, phone numbers, addresses and profile links replaced by placeholders first. | Nothing. You pay the provider directly, at their rate. |
Built for how a business actually runs
It reads what you tell it about your business before every answer, and does the repeat asking for you
A profile of your business
What you sell, who you sell to, what your stages mean and your name in the owner column, so “my deals” means yours. Metrics you define are calculated the same way every time.
Routines
Save a question with a schedule, or run it after every import, and the answer is waiting when you open the app.
One person, one record
The same customer in your leads, your CRM export and an event list is merged into one person. Anything uncertain is left for you to confirm.
Connects to your other tools
Stripe, Notion, Intercom and others plug in as connectors. Anything that would change data in them waits for your approval.
Works inside your AI tools
Anchor Connect makes your data available to Claude Desktop, Cursor and VS Code on the same machine, read-only.
Talk to it
Dictate questions and have answers read back. Speech is transcribed on the laptop.
How Anchor is different
Including where the alternatives are the better choice
| Alternative | Built around | Better if | Anchor |
|---|---|---|---|
| BI and dashboard tools | Dashboards someone builds in advance, over a data warehouse. | You have a data team and the same questions every week. | You ask the question you have today, about the files you already have. |
| CRM copilots | The records inside one CRM, usually on its top tier, often with usage credits. | Your whole team already works inside that CRM and has the budget for it. | Your CRM export, your ledger and your spreadsheets in one place you can ask, with no credits. |
| General AI chat | A conversation about anything, with your file uploaded to it. | You mostly need writing help and one-off questions. | Built for tables of business data, shows its query, and does not upload your file as it is. |
| Do-it-yourself local tools | People comfortable installing models and wiring up data themselves. | You are technical and want to assemble it yourself. | Packaged for people who are not, reads AtlasLinq data out of the box, and supported by us. |
Questions
When can I get it?
Anchor is being tested on real business data now. It will arrive through AtlasLinq accounts, starting with Enterprise teams, so the surest way to be early is to be on AtlasLinq when it does.
Is it a dashboard?
No. Nobody has to build anything in advance. You ask the question you have, and it answers from your data. Questions you ask every week can become routines.
Does it work on Mac?
Windows comes first. macOS is planned.
What does my laptop need?
With your company’s server or a frontier model, any recent laptop. The built-in model is about a 3 GB download and runs on a typical modern laptop, faster with a graphics card.
What about notes and other free text?
Anchor masks the names and patterns it recognizes inside notes, but free text can hide details no filter catches, which is why every reply shows what was sent. If your notes are sensitive, use the built-in model or your company’s server, where nothing reaches a third party. Amounts and dates are never masked, because the model needs them to answer.
How will it be priced?
Per person, as an add-on to your AtlasLinq plan, with no usage credits on top. We will publish the price when it launches.
Do I need AtlasLinq to use it?
Anchor reads any CSV or Excel file, so your data does not have to start in AtlasLinq. It is built to sit next to it, though, and it arrives through AtlasLinq accounts first.
Start with AtlasLinq today
Anchor arrives inside your AtlasLinq account. Sign in with it and your leads and card analytics come with you, limited to what your role can already see. Leads you capture from now on will already be there.