Nisa AI for Sales
The follow-up you meant to send on Tuesday.
Nisa lives in your sales channel. She reads the pipeline every morning, names the deals that have gone quiet, and drafts the next message. You approve it, or you do not.
Nisa lives in Slack, WhatsApp and iMessage.
#sales
You
@Nisa which deals have gone quiet?
Nisa
- crm-pipeline · 34 open deals
- gmail · last reply per thread
Four, and one of them is the reason I would start here. Meridian has been silent 19 days after asking for pricing, which is the longest gap on a deal that had already asked to buy. The other three went quiet before a first call, so they are a different problem. Draft for Meridian is below.
You
yes, send it
Nisa
Ready to send to 1 recipient from your mailbox. Subject: "Following up on the pricing you asked for". This is the text that will arrive, unchanged.
What changes
Same day
from "went quiet" to a drafted reply
The sweep runs before you sit down, so the deal is named on the morning it goes quiet rather than the week someone happens to scroll the pipeline.
0
messages sent without a person
Not a setting and not a default. There is no always-allow for a tool or a contact, so there is nothing to switch off in a hurry and leave off.
No addresses stored
replies are matched on the company domain
Nisa attributes an inbound reply to a deal by the sender’s company domain, so an individual email address never has to be kept to make the loop close.
These are the targets we hold Nisa to, not an average across customers. Each one below says how to check it against your own numbers.
Check it yourself. Take the last ten deals you lost to silence and find the date each one went quiet. Then run the sweep for a month and compare how long the next ten stayed unnoticed.
Three jobs to hand over first
Every weekday, 08:00
The quiet-deal sweep
Every open deal with no activity past your threshold, named in the channel with what was last said and who said it.
When a reply lands
Triage the inbound
A reply to an outreach thread is matched to its deal, summarised, and put in front of the person who owns it, instead of sitting in one inbox.
Friday
The pipeline read
What moved, what did not, and which deals need a decision from a person rather than another touch.
It reaches the tools you already run on
- HubSpot Deals
- Attio CRM
- Gmail Threads
- Google Calendar Calls
- Google Sheets Trackers
- Slack Where it lands
Plus 3,000+ more. Nisa only reaches what you connect, and you can revoke any of it.
Autonomous, not unsupervised
Nothing leaves without you, character for character
Every outbound message stops at a card in Slack showing the full text, with approve or edit-and-approve. What you approve is what arrives. Nothing is appended, trimmed or re-generated after you press the button.
A private source answers you privately
A connection or a part of your brain can be marked sensitive. Nisa still reads it, but the answer goes to you as a card in a DM instead of into the shared thread. The rule is enforced by the answer never being staged there, not by asking a model to keep a secret.
Built in the EU, by a German company
Practible AI UG (haftungsbeschränkt), Hamburg. GDPR is the floor we build to, and your data never trains a model. You can read, correct and delete anything Nisa knows about your company, because it is a set of files and not a set of weights.