Most AI knowledge tools ask you to ship your documents to their cloud. nohow.ai does the opposite: the entire stack deploys into an AWS account you own and control. Your documents, your embeddings, your conversation logs, your region, your retention policy. We never hold your data, because it never leaves.
Procurement, security review, data residency, audit. The things that make enterprise deployments slow are the things we have already solved.
Deployed via CloudFormation into your own account, in your own region. Documents, vectors and logs never transit our infrastructure. Delete the stack and everything is gone.
SAML / OIDC single sign-on, with document visibility driven by your existing groups. Public, customer and internal audiences see different answers from the same knowledge base.
Separate deployments per region, business unit or legal entity, each with its own data boundary — managed from one place, without duplicating the knowledge base by hand.
No question caps and no per-resolution billing. Because inference runs on your AWS account, cost scales with your usage at AWS rates — not at a vendor's markup.
Audit-grade conversation logging, PII redaction, citation-required answering, and configurable retention. Every answer traceable to the source document it came from.
Data processing agreement, custom service levels, and a completed security questionnaire. We have been through vendor onboarding before and will not slow yours down.
The parts other vendors hard-code — the model, the prompt, the retrieval behaviour, the branding, the access rules — are configuration here. If something still does not fit, we build it.
Claude Haiku, Sonnet or Opus, Amazon Nova, Llama, Mistral, Cohere — selected per bot and per answer-quality level. Trade cost against accuracy where it matters, and change your mind later without redeploying.
Custom system prompt, tone, and response language. Control how the bot refuses, when it must cite, how much it hedges, and what it does when the answer is not in your documents.
Your logo, colours, fonts, and welcome message, with all nohow attribution removed. The widget can be embedded in your site, your product, or your intranet — or replaced entirely by our REST API behind your own interface.
PDF, DOCX, PPTX, XLSX, CSV, HTML and Markdown upload; automatic website and sitemap crawling on your schedule; Jira for live ticket context. Custom connectors to your internal systems are built as part of the engagement.
Public, customer and internal tiers over one knowledge base, mapped to your directory groups. The same question returns different answers depending on who is asking — without maintaining three separate bots.
Independent bots for support, internal docs, sales enablement or per-region content — each with its own brand, prompt, language, knowledge scope, API key and integrations, all from a single deployment.
Chat widget, REST API, command-line client, MCP server for IDEs and AI assistants, and a browser extension for right-click lookups on any page. One knowledge base, reachable from wherever your people already work.
PII redaction, profanity filtering, off-topic refusal, citation-required answering, and a hallucination-critic pass. Thresholds and blocked topics configured to your compliance requirements, not to a generic default.
The agent can call your own systems — look up an order, check entitlement, file a ticket, query an internal API. We build and wire these tools as bespoke work inside the engagement.
Enterprise engagements include a direct line to the people who build the product. Onboarding, integration work, and change requests go to someone who knows your deployment.
Enterprise agreements are priced on what you actually need: volume, regions, integrations, and support level. We would rather understand the deployment first and quote something accurate than publish a number that fits nobody.
Most enterprise deployments are live within a few weeks, not a few quarters.
We map your use case, document estate, audiences and compliance constraints. Around 45 minutes, no commitment.
We stand up a working bot on a sample of your real documents so you can judge answer quality on your own content, not on a demo dataset.
Once we know the shape of the deployment, you get a written proposal covering scope, integrations, service levels and commercials.
We complete your questionnaire, sign the DPA, and walk your team through the architecture and data boundaries.
The stack deploys into your AWS account. We handle ingestion, tuning and integration, then train your team on the admin console.
A couple of lines about your documents, your audiences and your constraints. We will come back with whether nohow.ai fits — and say so plainly if it does not.
We reply to enterprise enquiries within one business day.