RFP & tender responses
Past wins become this week's answers. The bid manager pastes the question column; Knowledge fills the answer, confidence, and source columns next to it.
Knowledge turns your team's documents into instant answers — in Word, Excel and chat. Fill 200-question RFPs and security questionnaires in an afternoon. Draft design docs grounded in past decisions. Let employees ask the handbook anything. Your data never leaves your tenant.
In a typical mid-market company, the answer to almost every internal question already exists — in a past RFP, a SOC 2 report, an architecture decision record, a contract, a policy. But finding it, copy-pasting it and rewriting it for the new context burns 3–10 senior-IC hours every week — visible only as missed deadlines and weekend work.
At fully-loaded senior salaries, the knowledge tax costs a typical mid-market company between €12,000 and €60,000 per month. A 200-question security questionnaire takes 3 weeks. A 100-question RFP takes a person-week. New engineers ping principal engineers for answers buried in last year's ADRs.
Knowledge gives that time back — without sending a single document out of your tenant.
One private corpus. Three native surfaces. No retraining, no model fine-tune, no infrastructure team.
Ask questions, get cited answers. Conversations remember the thread so follow-ups land in context.
Five one-click skills that operate on the document you have open, not just a chat box on the side.
Bulk-answer a whole column at once. The killer feature for questionnaires, audits, and clause review.
Every team that owns a folder of "the way we do it here" gets a 5–10× speed-up on the work that uses it. Bulk-fill questionnaires in Excel; audit, fill, summarize and translate documents directly in Word.
Past wins become this week's answers. The bid manager pastes the question column; Knowledge fills the answer, confidence, and source columns next to it.
SIG, CAIQ, custom DDQs answered from your SOC 2, ISO documentation and policies. The agent says "I don't know" when the answer isn't in your corpus — never invented.
Drafts grounded in past decisions. New engineers stop pinging principal engineers for last quarter's trade-offs.
Each row is a clause. Output columns: deviation from playbook, severity, recommended response. The agent stays inside your playbook — never invents positions.
Auditor's control list goes in column A. Knowledge fills the existing evidence summary, document name, page numbers, and control owner in B–E. Gaps are flagged so you know what's missing.
Point the agent at your rulebook (ISO 13485, GDPR, internal SOP, quality manual). Open the document being audited, click Audit document. Findings appear as Word comments anchored to the offending paragraphs — severity, issue, suggestion, rule reference. Ready for a human reviewer.
Open a proposal template with empty H1/H2 sections. The agent walks each heading and drafts the body grounded in your past wins, capability statements, and project documentation. Stays consistent with surrounding sections instead of restarting from a blank page.
Employees ask the handbook anything. Support agents ask the runbooks. High volume, low cost, instant answers.
Most RAG products give you a chat box. Knowledge gives you Excel. Paste questions in column A, run the agent, and rows fill themselves with the answer, your confidence score, and the source document reference — ready to send back to the prospect.
Every claim on this page is enforced in code, not policy. Procurement and Security teams can audit the implementation, not just trust the marketing. Full security overview, DPA and sub-processor list are published.
Standards-based Entra ID single sign-on (OIDC + PKCE). Onboarding and offboarding follow your existing identity policies — no password lists, no shared accounts.
Each customer's data lives in its own scope at the database level. Cross-tenant access is rejected — even the existence of another tenant's documents stays invisible.
Each integration has its own rate limit, monthly budget, IP allowlist, and origin lock. A runaway plugin can never drain the whole quota.
Every call records who ran it, against which agent, with what tokens, and the outcome. Exportable for compliance reviews.
Documents and embeddings stored in European infrastructure. Custom data-residency arrangements available on Business plans.
Workspace admins, project owners, editors, and viewers — and documents that follow the projects they belong to. Only the right people see the right docs.
All prices in EUR per user per month, exclusive of VAT.
Documents, embeddings, and conversation history live in an EU-27 cloud host in Germany, isolated by tenant at the database row level. Cross-tenant access returns HTTP 404, not 403 — so the very existence of another customer's data isn't discoverable. Deleting your workspace cascade-deletes every record. Full breakdown in the Security overview and the DPA §5.
Mistral, OpenAI, Anthropic and Google. Free runs on Mistral Small only. Starter unlocks the full Standard tier (Mistral Small, Gemini Flash, GPT-4o mini, Claude Haiku). Pro adds the Premium tier (GPT-4o, Claude Sonnet, Gemini Pro, Mistral Large). Enterprise adds the Frontier tier (Claude Opus, GPT-4 Turbo). Pick the model per agent — use cheap ones for high-volume helpdesk Q&A and premium ones for high-stakes RFPs.
PDF, Word (.docx), Excel (.xlsx), PowerPoint (.pptx), CSV, plain text, Markdown, and images (PNG / JPG / WEBP) — including OCR via Tesseract for scanned PDFs and images.
Every answer is grounded in passages retrieved from your own documents, with citations. For spreadsheet workflows, the agent returns a confidence score next to each answer so reviewers can sort and only edit the weak rows. When the corpus doesn't cover a question, the agent says so — it never invents an answer. This is especially important for security questionnaires.
The fastest path is Office for the web: Insert → Add-ins → Upload My Add-in, then pick the manifest file. Two minutes per user. For organization-wide rollouts, use the Microsoft 365 admin center — supported on Pro and Business plans.
Your data stays in your workspace, is deletable on request, and is never used to train AI models — by us or by our LLM sub-processors. The Data Processing Agreement is published and counter-signable; the sub-processor list is version-dated; the Privacy Policy covers GDPR rights including 30-day deletion. Custom DPAs (with your company name + counter-signature) are available on every paid plan.
Projects are folders of documents with their own membership rules — they decide who sees what. Templates (also called agents in the app) are saved presets that combine a system instruction, a project scope, an AI model, and an output format. The Word and Excel plugins call these templates by name.
Workflows never hard-stop mid-questionnaire. Beyond the included usage, your plan charges metered overage (€4 / €3 / €2 per 100,000 generations depending on tier). You can also set spending caps per individual integration so a single workload can't drain the whole quota.
Spin up a free workspace in 60 seconds. Upload one PDF. Ask a question. If it doesn't earn its place this week, walk away — no card needed.