One AI toolkit formanagement consultants
Interview notes into findings. Slide prose into 2×2s and pyramids as native shapes. One local memory per engagement, so client A never leaks into client B. One menu-bar app, one shortcut.
Your deliverable is a structured argument,
your raw material is forty pages of notes
An engagement is the same climb every time: interviews and workshops pile up as raw notes, the notes must become findings, the findings must become a framework a steering committee can absorb in one slide, and the whole thing must read like your firm wrote it. A browser chat tab doesn't know which client you're on, can't draw the 2×2, and pasting interview transcripts into it is exactly what your NDA says not to do.
Khint puts that climb behind one shortcut, inside the documents you already work in. Select text anywhere and run an AI Action on it in place: interview notes become structured findings, rough prose becomes deliverable prose, and on a Mac a slide's text becomes a consulting framework in native editable shapes. One Memory session per engagement keeps context local and per-client, and an opt-in switch redacts identifiers before anything reaches the model.
One engagement,
kickoff to delivered memo
Every phase below is a shipped feature, not a roadmap slide.
- KickoffOpen a pack and a session for the engagementNew diagnostic at a client? Create a pack for the engagement (your synthesis, findings, and memo Actions live there) and start a Memory session with the client's name. From now on everything you pull, capture, and run is compacted into one engagement context, kept local, and isolated from every other client.
- DiscoveryInterview notes → structured findingsAfter each stakeholder interview, select your raw notes and run a synthesis Action in place: themes, verbatims worth quoting, pain points, open questions. Because the session context rides along, Thursday's synthesis knows what Monday's interviews already established: recurring themes stop being rediscovered.
- DiscoveryCapture the client's reality as textThe as-is process lives in screenshots: the legacy tool nobody can export from, the whiteboard after the workshop, a scanned org chart. Run Extract text from the palette, draw a region, and the text lands in your clipboard, and in the engagement session. Capture & ask lets you question a screenshot directly when you just need one number out of a dashboard.
- RecommendationSlide text → consulting framework, natively editableIn PowerPoint on your Mac, select the prose on a slide and run Transform to diagram. A prioritization argument comes back as a 2×2 matrix, a findings-to-conclusion argument as a pyramid, an option analysis as a comparison table, a target-state story as a before/after. Preview, switch layout or palette, then save native editable .pptx shapes and keep building the deck.
- DeliveryClient-ready prose, in your own documentSelect the rough executive summary and run a polish Action: tighten, keep the meaning, match the register a partner will sign off on. The Doc Polish starter pack ships prompts for exactly this; edit them once to your firm's tone and reuse them on every deliverable.
- DeliverySend the memo where the client reads itFrom the same palette: draft the send-out email in Gmail with a preview before anything goes out, create the page in Confluence or Notion if that's where the client's program lives, or log follow-ups as Jira or Linear issues once the recommendations turn into a delivery plan.
Built for work you signed
an NDA about
Client work needs sharper boundaries than a general chat app gives you. In Khint the boundaries are structural. An Action only ever sees the text you selected, plus the active session's compacted summary if you opted that Action in, and nothing else on your screen or disk. Engagement sessions live in a local database on your machine and are never synced to a server. Each client's Actions live in their own pack, and the palette only shows the active one. And for material where even the selection is sensitive, a per-Action switch replaces emails, phone numbers, SSNs, and IBANs with placeholders before the request leaves your machine: the model works on the shape of the text, not the identifiers. What leaves your machine is a choice you make per action, visible every time.
The four pieces of the toolkit
Engagement workflows, one page each
The parts of the loop, in depth: how each one works with real Khint features.
Common questions from consultants
What can a consultant actually do with Khint?
Select any text (raw interview notes, a workshop dump, a paragraph of your draft memo), hit Cmd+Shift+K, and run a saved AI Action on it in place: notes become structured findings, a rough argument becomes client-ready prose, a list of options becomes a comparison. On a Mac, select the text of a PowerPoint slide and Transform to diagram lays it out as a consulting framework (a 2×2 matrix, a findings pyramid, a before/after, a trade-off balance), exported as native editable shapes. You stay in the document or the deck; nothing moves through a chat tab.
Which slide frameworks does the diagram generator know?
Ten layouts are live: process, 2×2 matrix, pyramid, waterfall, timeline, comparison table, risk-to-mitigation cards, before/after, a two-sided trade-off balance, and a free-form flow diagram. Khint reads the slide text you selected, picks the framework that fits the content, and shows a preview where you can override the framework, switch layout variations, and pick a color palette before saving. Exports are native PowerPoint shapes (.pptx) or a transparent PNG: editable, not a screenshot. The PowerPoint companion is macOS-only.
I'm under NDA. What does the AI actually see?
Only what you explicitly act on: an Action sees the text you selected, plus (only if you tick the option on that Action) the compacted summary of your active engagement session. Engagement sessions live in a local database on your machine and are never synced to a server. For sensitive material you can also flip a per-Action redaction switch that replaces email addresses, phone numbers, SSNs, and IBANs with placeholders before the model sees the text. It is a pattern-based safety net, not a DLP product (read what you select), but it keeps the obvious identifiers out of the request.
How do I keep two client engagements from bleeding into each other?
Two boundaries, both explicit. Keep one pack of Actions per engagement: the palette only shows the active pack, so client A's ticket formats and terminology never appear while you work for client B. And keep one Memory session per engagement: only the active session's compacted context is injected into your Actions, so a findings draft for one client is never informed by another client's interviews. Switching engagement means switching the active pack and session, two clicks.
Is there a free plan?
Yes. The free tier includes 10 AI actions per day and 5 screen captures per day, no credit card. Paid plans start at €7/month: Pro at €29/month gives 100 AI actions a day, and every plan includes all the work integrations. Khint runs on macOS 13+ and Windows 10 22H2+; the PowerPoint diagram companion is macOS-only.
Bring it to your next engagement
Free with 10 AI actions and 5 captures per day. No credit card, no client-side install to clear with IT; it runs on your own machine.