Channels: Slack, SMS & voice

This page covers two different ideas that both involve channels beyond the main Nash chat. Part 1 — Nash Agent in Slack is about a place you talk to Nash. Mention or DM the Nash bot in your workspace and it answers in-thread, with the same data and permissions you’d have in the web app. Part 2 — Reaching customers and drivers is the opposite direction: actions Nash can take on your behalf, sending a text message or placing a voice call to someone outside your team and bringing their answer back into the conversation. These are not channels you chat with Nash through — they’re things Nash does for you, and they always wait for your approval first.

---

​

Part 1 — Nash Agent in Slack

If your workspace has the Nash bot installed, you can ask Nash questions without leaving Slack. It’s the same assistant you’d use in the web app: it sees your organization’s data, runs under your own permissions, and remembers the conversation thread.

​

Talk to Nash in Slack

There are two ways to reach the bot:

• Mention it in a channel — @Nash how many deliveries failed today? The bot replies in a thread off your message.
• Direct-message it — open a DM or the bot’s assistant view and type your question. A fresh assistant conversation suggests a few starter prompts (like delivery status, provider performance, or tracking an order) to get you going.

Either way, the reply streams in live: first a “thinking” indicator, then a running timeline of the steps Nash takes as it works, and finally the written answer. Each step shows whether it’s in progress, finished, or hit a problem, so you can follow along. Keep the conversation going in the same thread. Replies in a thread continue the same conversation — Nash remembers what you’ve already discussed there, just like a chat session in the web app.

​

Signing in is automatic

You don’t have to link your account by hand. The first time you mention or DM the bot, Nash matches your Slack profile email to your Nash account, and from then on you’re recognized automatically. Every question you ask runs under your Nash permissions — you see only what your role allows.

​

Working across multiple organizations

If your account belongs to more than one organization, Nash needs to know which one you’re asking about. The first time it isn’t sure, it asks you to pick your organization from a searchable dropdown, then remembers your choice for next time and answers your original question automatically. You can also switch organizations right in the conversation — ask Nash to switch to a different org, or to show you the list of organizations you can pick from, and it confirms the change.

​

Approving actions from Slack

Today, Slack is best for asking questions and read-only work — checking statuses, pulling analytics, tracking orders, and the like. Those run and return their answer directly.

​

Setting up the bot

Adding the Nash bot to a Slack workspace is a one-time setup handled by your Nash administrator together with Nash, not something an individual user self-installs. Once it’s in place, everyone on your team uses it the way described above — no per-person setup required. If the bot isn’t responding in your workspace, ask your admin whether it’s been installed.

---

​

Part 2 — Reaching customers and drivers

Sometimes the fastest way to resolve a delivery is to ask the person involved directly: text a customer to confirm a drop-off spot, or call a driver to find out where they are. Nash can do both on your behalf — and because these reach people outside your team, they come with firm guardrails. There are two of these capabilities: In both cases, the person’s response comes back into the same conversation as a new turn, so Nash can react to it in context — retry, escalate, or follow up.

​

These always wait for your approval

This is the most important thing to understand: Nash never sends a text or places a call on its own. Every single time, before anything goes out, Nash shows you a confirmation:

• For a text, you see the recipient and the message before it sends.
• For a call, you see the goal and the questions Nash plans to ask.

Nothing happens until you approve. There is no “auto-send” setting for these — they are confirm-every-time by design, in both interactive chat and scheduled agent runs. If you decline, nothing goes out.

​

When a reply comes back

Texts are a one-shot capture, not a chat thread. When Nash sends a text, it waits up to 15 minutes for a single reply. When a reply lands (or the window closes without one), the outcome comes back into your conversation so Nash can decide what to do next — for example, suggesting a follow-up or escalating to a call. If the customer texts back several times in quick succession, Nash takes those together as one reply. A few things to set expectations:

• It’s one message and one reply, not a back-and-forth text conversation. To send another message, Nash sends a fresh text — which means another confirmation from you.
• If someone replies after the 15-minute window, that late reply isn’t pulled back into this conversation.

Calls return a structured summary. When Nash places a voice call, the AI voice agent works toward the goal you set and asks your questions. A typical call runs a few minutes. When it ends, Nash brings back a summary of what was discussed, a recording, any specific details you asked it to collect, and a note if it reached voicemail — all delivered into your conversation as a new turn so you can act on it.

​

Turning these on

These capabilities aren’t on for everyone — they have to be enabled, and they’re limited by your permissions.

​

FAQ

​

Related