Slack assistant

Run your day from Slack. Alan does the typing.

Direct-message the TuringDock bot. Alan handles the same actions you'd take in the app: log counts, start, pause, mark done, query stock and schedules. Tool calls dispatch to your workspace state in real time.

Get started for freeTalk to the founder
B
Bob9:42 AM

done 47 PCB-100

A
Alan9:42 AM

Logged 47 PCB-100 completed. Stock movements posted: -47 PCB-RAW, -47 ATMEGA328P, -235 CAP-100, -470 RES-100, +47 PCB-100.

B
Bob9:44 AM

count PCB-RAW 24

A
Alan9:44 AM

Logged 24 PCB-RAW. Books had 28, variance −4. Want me to apply the adjustment, recount, or open an investigation?

How it works

Count from the bench

count PCB-RAW 47 in your DM with the bot. Alan finds the SKU, computes variance against current on-hand, writes the cycle-count event, and responds with the variance + an action question.

If books and counted match, you get a one-line confirmation. If they don't, Alan offers to apply the adjustment, recount, or open an investigation. The bench operator doesn't need to open the app.

Start, pause, mark done

start PCB-100 when you walk to the bench Monday morning.done 47 PCB-100 when you finish, with the actual quantity. Alan moves the task through its state machine and posts stock movements atomically against the BOM snapshot taken when the task was scheduled.

The Gantt and the My Tasks board update in the web app the moment Alan processes the message. Whoever's sitting at the desk sees the bar flip colour.

Ask, narrate, summarize

what's Bob doing this week? · what's our PCB-100 stock? · any open POs with Mouser? · forecast PCB-RAW for a 21-day lead time. Alan calls the right tool and replies in plain language with the concrete numbers.

Named after Alan Turing. First-person voice.

Alan doesn't refer to itself in the third person. Replies are first-person ("I logged 47 PCB-100 for you. Variance from books was -3 units.") and specific. Sign-offs on drafted emails attribute to Alan; in-thread messages don't need the byline.

Why this is the new age of MRP

Legacy MRPs are point-and-click only. Every action requires walking to a laptop, navigating menus, finding the right row, editing, saving. By the time the operator has logged the work, the next pallet is already in the way.

Alan moves the same workflow into Slack, where the operator already lives. The phone in their hand becomes the interface. The app becomes a place you go to look, not a place you go to do.

Build what's next. Leave the MRPing to our AI.

Build what's next.

Leave the MRPing to our AI. Sign up free, set your shop up in an afternoon.

Get started for free

Common questions

What is Alan?
Alan is TuringDock's AI assistant, named after Alan Turing. Alan lives in three places: the right-rail sidebar inside the TuringDock web app, as a DMable bot in your Slack workspace, and as a Model Context Protocol server you can connect Cursor or Claude Desktop to. The same eleven tools work across every surface: query inventory, log cycle counts, mark production tasks done, draft customer-order schedules, summarize reports, forecast demand.
How do I install the Alan Slack bot?
Sign in to TuringDock as a workspace admin, open Settings → Integrations, and click 'Add to Slack'. You'll be redirected to Slack to authorize Alan, then back to TuringDock. Takes about 10 seconds end-to-end. Alan is then DMable from any channel in your workspace.
What Slack commands does Alan understand?
Natural language. Examples: 'count PCB-100 47' (logs a cycle count and computes variance), 'done 50 PCB-100' (marks a production task complete), 'start PCB-100' (kicks off a task), 'what's our PCB-100 stock?' (queries inventory), 'what's on Bob's plate this week?' (operator schedule summary), 'forecast PCB-RAW for a 21-day lead time' (demand forecast with recommended PO quantity), 'summarize last month's revenue' (report summary). Alan figures out which tool to call from the message.
Does Alan work offline or only when I'm at the computer?
Alan runs server-side. Your Slack messages route to Alan over Slack's WebSocket connection any time of day, whether or not you have the TuringDock web app open. Operator at the bench can mark a task done at 6 AM from their phone; Alan posts the stock movements and updates the Gantt before you're at your desk.
Does Alan use my data to train models?
No. Alan uses Anthropic's Claude API in standard inference mode; per Anthropic's policy on API traffic, your data isn't used to train models. Your workspace data stays in your Postgres instance, scoped per workspace; Alan reads it per-request to answer your questions.
Can I connect TuringDock to my own AI tools (Cursor, Claude Desktop)?
Yes, via the Model Context Protocol (MCP) — premium tier. Paste one config block into your AI client's settings and the same eleven tools Alan uses become callable from inside Cursor, Claude Desktop, or any MCP-speaking client. See /features/mcp for the config block.

Related features

AI-first Gantt

What Alan updates when you mark a task done from Slack.

Cycle counts

The other side of Alan's count workflow.

Money page

Alan can also draft an overdue-customer follow-up from here.

Last updated 2026-05-24