Platform · Dashly Access
Your own AI on numbers that add up.
With Dashly Access you ask Claude, ChatGPT, Copilot or your own agents about your organisation, with the source behind every answer.
The problem
Your AI doesn't know your organisation by itself.
With loose exports, AI gives a quick answer that sounds convincing, but is often wrong.
Last week's exports
You collect the numbers yourself and keep them up to date. Via Dashly your AI works with today's data.
The same client three times
The same client is in your CRM, your scheduling and your accounting, under different names. Dashly counts it once.
Guessed definitions
'Placed' means something different to recruitment than to planning. Dashly calculates with one definition per metric.
Example
Just ask.
In plain language. The answer comes from the model, with the calculation included.
You ask
Which clients deliver the lowest margin per hour?
Your AI answers
The lowest margin per hour is at clients where the rate lags behind the cost price, after the latest increase in allowances. At one of them, not all worked hours are invoiced either. Proposal: schedule a rate conversation and invoice the missing hours.
Example answer, without real numbers.
Why, and what next
Questions across your systems.
Your AI can dig deeper, because clients, people, recruitment and margin are in one model.
Which recruitment channel delivers people who keep working?
- What does each channel cost per worked hour, not per application?
- What share of placed candidates is still active after 90 days, per channel?
- What share of applications has a known source?
What you do: Shift budget to the channel with the lowest cost per worked hour.
Which clients are requesting less, and why?
- Has their demand dropped, or are we filling their requests less often?
- Do placements at this client end early more often, without a complaint?
- Is a contract ending soon, or is volume falling short of the contract?
What you do: Prepare an account review, with the numbers at hand.
Which vacancies and requests are at risk of staying open?
- How many suitable and placement-ready candidates does each vacancy or request have?
- Which skills or certificates are missing from your pool?
- On average, how long do similar vacancies and requests stay open with you?
What you do: Prepare a shortlist based on skills and availability, or recruit specifically for the role.
Where you work
In the AI you already use.
Via MCP, on every layer. You choose which AI.
- ClaudeFor long questions about your whole organisation, with the formula included.
- ChatGPTAsk in plain language, with your real numbers underneath.
- CopilotFor those who work in Microsoft 365, alongside email and documents.
- Your own agentsAny agent that speaks MCP, with its own access and the same formulas.
On every layer
The higher the layer, the more consistent the answer.
Your AI can connect to any layer. What it reads differs.
On Connect
Your AI reads the raw sources. You decide what belongs together.
On Data
Your AI reads linked data: every client, candidate and employee once.
On Knowledge
Your AI calculates with the model's fixed formulas. This is what we recommend.
Secure
You decide what your AI sees.
- Permissions per user and per agent, down to row level.
- Your AI sees no personal data, such as BSNs (Dutch citizen service numbers).
- Reading is different from changing. Your AI reads; actions come with Dashly Agent.
- Every question is logged: who, when and on which data.
- Your data is isolated in a dedicated Google Cloud environment per customer.
Questions
Questions about your own AI.
Want to see it with your own questions? We do that in a demo.
01Why not just put Claude or ChatGPT on my systems?
Because your AI then doesn't know which client is the same one, which definition you use or whether last week's export is still correct. The answer looks good, but it's different every time. Via Dashly your AI calculates with today's data and fixed formulas, with the source included.
02What is MCP?
An open standard that lets AI talk to other systems. Via MCP your AI requests numbers from Dashly, within the permissions you give.
03Is Dashly Access available?
Yes. Dashly Access works with Claude, ChatGPT, Copilot and your own agents, on every layer. It works best on Knowledge, because your AI then calculates with the model's fixed formulas.
04Does my AI see personal data?
No. Personal data doesn't belong in a standalone chat. Your AI calculates with hours, placements and margin, and you decide per user and per agent which data it sees.
05Can my AI also change things in my systems?
Not yet. That comes with Dashly Agent, which is coming soon. For each type of action you then decide whether someone approves it first.