AI inside the work, under your rules.
Xifdesk uses AI where work actually slows down: writing proposals, answering clients and shortlisting candidates. It drafts from your own data, follows clear guardrails, and leaves every decision to a person.
Five rules the AI always follows.
These guardrails are built into proposal writing and client replies. They are what lets a team move faster without losing control of what it promises.
-
It never commits to prices or timelines
Commercial terms stay with your people. Drafts describe the work, never what it costs or when it lands.
-
It never invents numbers
No made-up results, percentages or figures in a proposal or a reply.
-
It only cites work you have verified
Portfolio answers use projects and links your team has confirmed.
-
A person reviews before anything is sent
Every draft is copied, shortened, regenerated or edited by someone on your team, then sent by them.
-
Every draft is accounted for
Proposals are saved with their matched projects, rating, feedback and token cost.
Proposals matched to the work you have delivered.
Paste the job description with any attachments and client questions, then choose the tone and length and add your brand voice. Xifdesk ranks your best-matching past projects with a match score, writes three versions and answers the client’s questions in the same tone. Improve or regenerate on request.
- Logistics driver app92Offline trip logging, background sync
- Fleet maintenance portal84Vehicle records, service schedules
- Courier dispatch dashboard77Live map, route history
Hi Jonah, we would build the app in React Native with a local-first data store, so trip logging works fully offline.
A background sync queue sends trips to your API as soon as a connection returns. We used the same approach in our logistics driver app.
Happy to walk your team through the architecture on a short call.
- Only verified portfolio links
- No prices or timelines committed
- No invented numbers
- Reviewed by a person before sending
Try it: switch versions, or press Draft again. Fictional sample data.
The next client reply, ready when your team is.
Most deals are won or lost in the replies after the first proposal. Intelligent Client Management reads the conversation so far, the lead’s details and your own portfolio, then drafts the next reply for your team to review and send.
- Conversation memory
- Every earlier message, your proposal and the client’s questions are kept, so each reply follows on from the last.
- Portfolio-aware answers
- When a client asks whether you have done this before, Xifdesk searches your past projects and cites the relevant ones with verified links.
- Ask AI on any card
- A chat panel on every board card answers questions about that lead from its details, comments and attachments, including documents and Loom transcripts.
- Reply logging and response tracking
- Each client reply is logged on the card with its date and time, so managers can see response times and follow-ups.
Hi Jonah, yes to both. In the logistics driver app we delivered, trips are stored on the device and sync when the connection returns, and the app talked to the client’s existing Node.js API.
Happy to show both on a short call this week.
A shortlist with its reasons, and a first call already made.
AI scores each resume against the opening’s must-have skills and shows the evidence behind every score. HR approves the shortlist. An AI voice agent then calls approved candidates, asks the screening questions and returns a transcript, scores and an outcome.
Resume scores
Against the opening’s must-have skills- Nadia Petrov885 years React NativeShipped offline-first appsLed a team of threeApproved for screening call
- James Okoro814 years React NativePayments integrationsAwaiting HR review
- Lena Wolff642 years FlutterNo React Native yetAwaiting HR review
Every score shows its evidence. HR approves the shortlist before any call is made.
Try it: press Replay. Fictional sample data.
AI you can administer like any other system.
Your choice of model
Administrators choose the AI model provider the platform uses.
Versioned prompts
The prompts behind each AI feature are versioned.
Cost you can see
Token use and cost are tracked per organization and per user.
See AI work on your own pipeline.
A walkthrough shaped around your teams: how sales, client conversations, people, payroll and delivery would work together on Xifdesk.