Your code already has the answers.
Now your users do too.

Tacit reads your entire codebase — every workflow, rule, and feature — and turns it into an AI assistant your users can actually talk to. No documentation required.

Request Early Access See How It Works
.NET / ASP.NET Core Java / Spring Boot Node / NestJS / Express React / Angular / Vue
tacit · /ˈtæs.ɪt/

Knowledge that exists within a system but has never been written down. Understood by those who built it, invisible to everyone else.

Every product's codebase is full of tacit knowledge that never made it into a user guide. We surface it.

The problem

Your product has capabilities your users never discover.

Most product companies ship features faster than they can document them. There's no technical writer on the team, no user guide that keeps up. So users miss capabilities, underuse what they're paying for, and file tickets when a quick answer could have kept them moving.

12x
Cost of a human support interaction vs. an automated one
Dante AI
51%
Of users prefer a chatbot over waiting for a human
Zendesk CX Trends
$2M+
Annual cost of missing documentation — in support, delays, and churn
DECODE Research
80%
Of "how do I" questions are resolvable without a human
Gartner
The difference

Generic AI hallucinates.
Tacit knows your product.

An off-the-shelf LLM gives plausible-sounding guesses. Tacit gives answers grounded in how your product actually works — every claim traceable to real logic, not training data.

Generic AI Chatbot
How do I approve an invoice in the system?
To approve an invoice, navigate to the Invoices section in your dashboard. Look for the "Approve" button next to each pending invoice. Click it and confirm. You may need manager-level permissions.
Hallucinated — no source, generic guess
Tacit
How do I approve an invoice in the system?
Invoice approval is a 2-step process. First, an invoice must be in "Pending Review" status (set automatically after document import). Then, a user with the Approver role can approve it from the Processing Queue. Approval triggers an automatic 3-way match against the purchase order and shipping notice before posting.
Invoice Processing → Approval workflow
Required role: Invoice Approver
How it works

Set up in hours, not weeks.

Tacit understands your codebase structurally — features, rules, and data flow are indexed as relationships, not flat text.

Step 01

Connect your repository

Point Tacit at your codebase — a local copy for air-gapped environments, or a connected remote for cloud setups. It reads your project structure, framework conventions, and dependency graph.

$ tacit index --repo ./src
Step 02

Your codebase gets indexed

Controllers, services, DTOs, routes, middleware, validation rules, and business logic — indexed as structured relationships, not flat text.

✓ 847 endpoints · 234 services · 1,291 DTOs indexed
Step 03

Embed a user-facing assistant

Add a chat widget to your product. Users ask questions in plain language and get accurate, sourced answers about your product — no documentation needed on your end.

<script src="tacit.js" data-project="acme">
Use cases

Built for product teams without a docs team.

Deflect support tickets

Your users ask "how do I do X" dozens of times a day. Tacit answers them instantly, grounded in your actual product logic — so your support team handles fewer repetitive questions and your users don't wait.

Accelerate user onboarding

New users learn by asking, not by hunting through a help center that was last updated two quarters ago. They get to value faster, and you stop losing them in the first week.

Surface product blind spots

Every unanswered question is a signal. Tacit shows you what users are trying to do but struggling with — turning a support chatbot into a product discovery tool that reveals where your UX or docs fall short.

Deployment & Security

On-premise and air-gapped. Your code never leaves your network.

Built for regulated industries and security-conscious teams. Tacit runs entirely inside your infrastructure — air-gapped, on-prem, or private cloud. No data ever reaches our servers.

Air-gapped deployment

Tacit ships as a self-contained package — Docker images, local LLM inference, and embedded vector search. Deploy on your own hardware with zero external network calls. No API keys to an outside service. No telemetry. No phone-home.

Designed for environments where code cannot leave the building: defense contractors, financial institutions, healthcare platforms, and any organization operating under strict data residency requirements.

Full air-gap supported
Indexing
Your codebase → Tacit → Knowledge graph
↓
At query time
User question → Tacit → Grounded answer
✓ Everything runs inside your perimeter. Nothing leaves.

On-prem & private cloud

Deploy on bare metal, VMware, or your private Kubernetes cluster. Supports Azure Private, AWS VPC, and any Docker-compatible environment.

Runs on your hardware — GPU or CPU-only configurations
No cloud dependency, no SaaS data pipeline
Integrates with your existing SSO and access controls

Data sovereignty & compliance

Your source code, embeddings, and user queries never leave your jurisdiction. Built for teams operating under GDPR, SOC 2, ISO 27001, and industry-specific regulations.

EU, UK, IL, and other jurisdictional deployments
Full audit trail on every query and answer
No training on your data — ever
FAQ

Frequently asked questions

How does Tacit work without documentation?
Tacit indexes your codebase — controllers, services, routes, validation rules, and business logic — and builds a knowledge graph of how your product works. When users ask questions, Tacit retrieves accurate answers grounded in your actual code, not generic training data.
Can Tacit be deployed on-premise or in air-gapped environments?
Yes. Tacit ships as a self-contained package with Docker images, local LLM inference, and embedded vector search. It runs entirely inside your infrastructure with zero external network calls — suitable for defense, financial, and healthcare environments with strict data residency requirements.
What tech stacks does Tacit support?
.NET (ASP.NET Core), Java (Spring Boot), and Node.js (NestJS, Express) backends, along with React, Angular, and Vue frontends. Additional framework support is on the roadmap.
How does Tacit reduce support tickets?
Tacit answers routine "how do I do X" questions instantly and accurately, grounded in your actual product logic. This deflects repetitive tickets from your support team while giving users faster, more reliable answers than a generic chatbot or outdated help center.
Is Tacit a replacement for writing documentation?
Tacit works best for teams that don't have documentation or whose docs are perpetually out of date. It fills the gap by deriving user-facing answers directly from your codebase. If you do have documentation, Tacit can work alongside it — but the core value is that it doesn't require docs to exist at all.
How is Tacit different from ChatGPT, Claude, Gemini, or Perplexity?
General-purpose LLMs — ChatGPT, Claude, Gemini, Copilot, Perplexity, Mistral, Llama — generate plausible-sounding answers from training data. They don't know your product. Tacit is grounded in your actual codebase, so every answer reflects how your product really works. It cites specific workflows and features, not guesses.
Early Access

Your codebase is already the best documentation you have.

We're onboarding a small group of .NET, Java, and Node product teams. Get in early, get direct support, and help shape what we build.

No credit card required. We'll reach out within 48 hours.

Built by a CTO with 15+ years building enterprise .NET and logistics systems — born from the same pain you feel: shipping features faster than you can document them.