Knowledge
Docs, sites and policies agents search before they answer, with citations.
From first question to finished work
WisebotAI gives every team AI agents that answer from your company's knowledge, take action in your tools and deliver the finished result. Build them in plain language. Run them on web, WhatsApp, Slack, email and voice.
Hi, I returned order #4821 last week. Where is my refund?
Support agentYour return arrived on Oct 8 and qualifies under our 30-day policy. I've issued a refund of $84.00 to your Visa. It should land in 3–5 business days.
refund-policy.pdf §2Order #4821
Refund issuedauto-approved under $100
Help desk ticket resolvedtagged: refund
Proof log
- 10:42:01Question receivedWeb widget · returning customer
- 10:42:03Answer grounded2 sources cited
- 10:42:04Action takenRefund rule checked and passed
- 10:42:05Outcome recordedOwner: Support · no handoff needed
ResolutionResolved in 4s
Your data stays yours
You choose the models. Your content never trains third-party models without your consent.
People make the judgment calls
High-risk actions go to your team with full context and an approval record.
Encrypted and access-controlled
Encrypted in transit and at rest, with roles that decide who can change what.
Why WisebotAI
Company AI should run like infrastructure, not a pile of bots
Most companies run AI in silos: a bot per team, copied prompts and no shared memory. WisebotAI gives every agent the same knowledge, tools, guardrails and workflows, so support, sales and ops work from one source of truth.
Agents take the repeatable work. People keep the decisions that need judgment.
Tools
Integrations, bookings and your own APIs. One agent, many actions.
Workflows
Multi-step processes on a visual canvas, drafted for you by AI.
Guardrails
Blocked topics, response limits and approval rules the agent must follow.
Human handoff
Escalate to a person with the whole conversation attached.
Finished work
PDFs, slide decks, websites and code, built in a secure workspace.
Model choice
Pick GPT, Claude or Gemini per agent to balance cost and quality.
Analytics
Conversations, resolutions, handoffs and cost, by agent and source.
How it works
Live in five steps, not five tools
Go from an empty workspace to an agent in production in one guided path. No prompt engineering or servers required.
01Describe the job
Say what the agent should do. AI drafts the prompts, tools and workflow steps for you to review.
02Add your knowledge
Upload docs, web pages, policies and FAQs. Answers cite the source your team trusts.
03Connect channels and tools
Publish to web, email, WhatsApp, Slack or voice, then connect your CRM and help desk.
04Set the rules
Choose what needs approval, what’s off-limits and when a person takes over.
05Launch and measure
Test in the playground, go live, then track usage, results and cost by source.
Example: e-commerce store
An online store that answers, tracks and refunds on its own
Paste one snippet into your Shopify theme and the agent goes live on every product page. It answers sizing, shipping and return questions from your store policies, looks up orders, issues refunds under your limit and follows up on abandoned carts over WhatsApp.
- Before the saleSizing, stock and delivery-date questions answered from your catalog and policies
- After the sale“Where is my order?” answered with live tracking, without a ticket
- ReturnsRefunds under your limit run automatically; bigger ones wait for a teammate
- Every languageHindi and other Indian languages over chat and voice, powered by Sarvam AI
1 snippetadded to theme.liquid to go live on Shopify
24/7order, shipping and returns answers on web and WhatsApp
Linen shirt · SandSize M · ₹2,499 · in stock
Will this arrive in Pune before Friday? And can I return it if M is too big?
Store assistantYes. Order in the next 3 hours and it ships today, arriving Thursday by standard delivery. Returns are free within 15 days, and I can swap it for an L.
shipping-policy §1returns-policy §3
Order #A-2291 trackedout for delivery
Refund ₹1,299 issuedauto-approved under limit
Cart reminder sentWhatsApp · Hindi
Workflows
Automations your whole team can read
Describe what you need and AI builds the workflow. Change any step in plain words or on the drag-and-drop canvas. Every trigger, AI step, tool call and handoff stays visible.
StartNew request via webhooklive
AI stepClassify and check knowledgecited
IfRefund under $500
ElseEscalate to manager
ActUpdate CRM and replyv24
LogRun completed with output10:42
Build with AI
Describe it and AI drafts the whole graph.
Edit in plain words
Change a step and AI updates it.
Triggers and timers
Webhooks, API calls and scheduled runs.
Your business apps
Slack, Notion, your CRM and any REST API.
Code and web data
Optional code steps and site crawls.
Branches and approvals
If/else paths and manager sign-off.
Beyond the chatbot
A chatbot stops at the answer. WisebotAI keeps going.
Every request runs through the same four steps, so your team can always see what happened and why.
Customers ask on web, email, WhatsApp, Slack, SMS or voice. One thread per customer means nobody repeats themselves.
Answers come from the docs, policies and FAQs you approve, with citations your team can check.
The agent updates tickets, CRM records and calendars in your tools. Sensitive actions wait for a teammate to approve.
Each run records its sources, the tools it called, who approved what and the outcome.
Finished work
Agents that hand you the file, not just the answer
A secure built-in workspace lets agents run code, browse the web and produce real deliverables. There are no servers to set up, and every run keeps a trace your team can review.
PDF reports
Summaries, proposals and policy documents.
Slide decks
Board updates and pitch decks from your data.
Websites
Landing pages and microsites with a live preview.
Code
Scripts and data jobs in Python, Node or shell.
Agents can also generate images and structured data, and export results to the tools you use.
RequestTurn last quarter's support metrics into a 6-slide deck for the leadership meeting.
Read analytics exportdone
Ran chart code in sandboxdone
Built 6 slidesdone
Saved q3-support-review.pptxready
App and SDK
Business teams build in the app. Engineers build in code.
Support, sales and ops set up channels, knowledge and tone in the dashboard. Engineers version agents, tools and guardrails in TypeScript or Python through the same API. Both work from one source of truth.
- Support and CX tune agents, escalation paths and channels
- Sales reuses the same knowledge and handoff threads
- Ops publishes playbooks and guardrails every agent follows
- Engineers ship updates through the API or CI, with no forked prompts
Choose the model for each agent
Read the docs →GPT-5.6Claude Sonnet 5Gemini 3.8Sarvam AI · Indic LLM and voiceOpen models · rolling out
import { WisebotAI, Tools } from '@wisebotai/sdk'; const client = new WisebotAI({ apiKey: process.env.WISEBOTAI_API_KEY, }); // Upload a policy, then attach it by filename await client.knowledge.upload({ file: policyPdf, filename: 'refund-policy.pdf', }); await client.agents.create({ name: 'Release-Ops-Agent', model: 'gpt-5.6-terra', tools: [ Tools.KnowledgeBase({ label: 'refund-policy.pdf' }), Tools.Connector({ provider: 'github' }), ], });
Every channel
Wherever customers ask, WisebotAI answers
Publish the same agent everywhere. Customers move between channels and the conversation follows them, so sales and support see one timeline.
One customer, one thread
Every reply on every channel comes from the same knowledge, tools and guardrails. Add voice and live video avatars when your team is ready.
Web
Asked about Startup plan limitsMon 09:14 · answered from the pricing doc
Email
Sent a security questionnaireMon 11:02 · routed to sales, owner: Priya
WhatsApp
Moved the demo to ThursdayTue 16:40 · calendar updated
Voice
Called about SSO setupWed 10:05 · handed off with full context
Video
Onboarding walkthrough with avatarThu 14:30 · transcript saved
Works with the tools you already use
Ready-made connectors, anything that supports MCP, and your own APIs.
Shopify
Salesforce
HubSpot
Zendesk
Intercom
ServiceNow
Jira
Linear
Asana
GitLab
Notion
Google Workspace
Microsoft 365
Snowflake
BigQuery
Postgres
Sarvam AI
Your APIs & MCP
Use cases
Six jobs teams hire WisebotAI for
Most teams start with one high-volume flow, prove it in two weeks, then add the next.
Support leadB2B SaaS team
One thread, one policy
Web, email and messaging in one thread. Agents answer from the same policy and hand off with full context.
Revenue operationsSales-led team
Qualify while it’s warm
Qualify inbound conversations, write the CRM notes and route enterprise prospects to a rep before interest cools.
Operations managerServices team
Repeat work, on rails
Turn approvals, intake triage and exceptions into workflows anyone can read, with a review trail.
FounderEarly-stage company
Start small, grow into it
Launch one agent and add channels as you grow, without stitching together separate chat, knowledge and automation tools.
Engineering teamInternal tooling group
Code where it helps
Connect APIs and MCP tools while model choice, citations and handoff rules stay visible to the business.
LeadershipMulti-team organization
See it all in one place
Track agent performance, conversation volume and where people still step in, from one shared view.
ROI calculator
What would agents save you in year one?
Move the sliders to match your support volume. You get an estimate you can take to finance, with the assumptions written out.
50050,000
$1$50
Estimated savings in year one
$273,600
36,000tickets handled by agents a year
$480,000current annual support cost
Assumes agents handle 60% of repetitive tier-one tickets and about 5% of the savings goes to AI processing. Your results will depend on ticket mix and setup.
Security
Controls your security team can check
Guardrails, approvals and access rules ship in the product, not on a slide. Everything else is clearly marked as available on request.
Guardrails
Block topics and set response limits. Policy is enforced, not just written into a prompt.
Approvals
Sensitive actions pause until someone on your team approves.
Approve refunds over $500
Block legal and medical advice
Restrict widget to allowed IPs
Hand off when confidence is low
Cite a source on every answer
IP access
Allow or block access to widgets and the API by IP address.
Review trail
Conversation history, approval records and run traces for gated actions.
Sign-in: Google and GitHub today. SAML SSO, audit-log exports and DPAs on request.
Your data: never sold, and never used to train third-party models without consent.
Security docs →Pricing
Start free. Pay as conversations grow.
Price depends on AI replies, seats, channels and model usage. Pick your monthly volume to see the plan that fits.
How many AI replies do you need a month?
501,0001,7507,500
StartupMost popular
Growing teams on several channels
$49
Per month, billed yearly
- 10 agents, 4 channels, 5 seats
- 1,750 AI replies a month
- 500 knowledge documents
- 175 build & automate actions
- 500 premium AI and 150 build credits
- Integrations and priority support
Every plan includes monthly premium AI and sandbox credits. Need more? $9 packs add 45 premium AI credits or 30 build credits.
Corporate
For regulated teams and private deployments, sized to your systems and security needs.
Custom pricing
- Unlimited agents and any model, including your own
- Private or on-premise deployment, full white labeling
- SSO, DPA, audit program and 24/7 phone support
FAQ
Questions buyers ask before rolling out
Agents, workflows, security and pricing, answered plainly.
Ask us anything else →A digital teammate trained on your company’s knowledge. It answers questions, uses the tools you approve, follows your guardrails and hands off to a person when judgment is needed. Support, sales, ops and internal teams can each have their own.
Create an agent by describing its job, ground it in your approved knowledge, and connect the channels and tools your team already uses. Add workflows for multi-step tasks. People stay in control of sensitive decisions.
Optional automations drawn on a canvas: what starts them, what the AI does, any branches (for example, escalate refunds over $500), which apps to update and what to send back. Repeat cases get faster and every run leaves a trail you can review.
Knowledge is the docs, sites and sources agents search before they answer. Tools are the actions they can take: integrations, bookings, your own APIs, and a secure workspace for running code, browsing the web and producing sites, PDFs and decks.
Yes. Use ready-made connectors for Salesforce, HubSpot, Zendesk, Notion, Google Workspace and more, connect anything that supports MCP, or add your own HTTP tools and use our API. Some connectors need your credentials set up in agent settings.
Yes. Agents can answer phone calls (via Vapi) and run live video-avatar sessions (via Tavus or HeyGen) with your own provider key. Voice and video use the same knowledge, tools and guardrails as chat, and transcripts are saved to the conversation. Sarvam AI voice and LLMs cover Hindi and other Indian languages.
OpenAI GPT, Anthropic Claude, Google Gemini and Sarvam AI models for Indian languages, chosen per agent, with low-cost defaults for high-volume replies. Premium AI credits cover flagship models. Open models on dedicated GPUs are rolling out; the docs list what’s available today.
Data is encrypted in transit and at rest, with access controls designed for business data. Your conversations and knowledge are not used to train third-party models without your written consent. We share data only with the infrastructure and model providers needed to run the features you turn on.
No. We don’t sell personal information or share it for marketing. Sharing is limited to what’s needed to run the service, such as hosting, email delivery and the AI or integration providers you connect. Our Privacy policy lists every category.
Google and GitHub sign-in ship today. SAML SSO, formal audit-log exports and DPAs are available on request for enterprise customers. In-product controls include guardrails, approvals, IP allow and block lists, conversation history and approval records.
Three monthly allowances (AI replies, knowledge documents, and build & automate actions) plus premium AI and sandbox credits on every plan. Included allowances reset each billing period; purchased credit packs roll over.
Yes: 50 AI replies, 10 knowledge documents, 15 build & automate actions, 20 premium AI credits and 15 sandbox credits a month, with one channel and one seat. No credit card needed.
Yes. $9 packs add 45 premium AI credits or 30 build credits. Larger packs and Enterprise overage are available as you grow.
Put your first agent to work today
Pick one high-volume task. Go live in an afternoon. Measure it for two weeks.