llms.txt
# Caddi
> Caddi is the end-to-end AI agents platform for the professional services back office: law firms, wealth management and RIAs, accounting firms, private equity, and insurance. Caddi AI agents find the repetitive work across every tool a firm already uses, automate it, and keep looking for more. The platform runs as one loop: Discover finds the work worth automating, Automate builds the agent with the person who does the work, and Govern gives firm-wide visibility into every run. Common first workflows are client intake, email triage, and document filing; the use-case library below lists 100+ more.
## How Caddi works
- Discover (AI process mining): connects read only to the tools a firm already runs, reads activity in the background (usually 15 to 30 minutes), and ranks the repetitive work by hours it could give back, with the evidence behind each finding (counts of messages, envelopes, matters and steps observed). Customers find about 19,000 hours of repetitive work per customer in small first rollouts.
- Automate (agentic automation and AI workflow automation): the person who does the work shows Caddi the task on a screenshare and talks it through, or describes it in chat, like training a new hire. Caddi is a screenshare-to-API builder: it combines what is on screen, the network requests behind each click and the person's narration, so the agent it builds calls the tools' real APIs. This differs from record-and-replay RPA recorders, which play clicks back on the screen and break when the interface changes. It asks about edge cases and keeps suggesting improvements, because it is built with domain knowledge of law firm, wealth management and accounting operations. No IT project; the first agent takes about 45 minutes on average.
- Govern (AI agent governance): agents run on deterministic code, so the same input takes the same path every time, with AI only where a step needs judgment (hybrid agents). This uses about 80% fewer tokens than an agent reasoning on every run. Every run, decision and permission is logged and can be replayed. Caddi flags drift after a run and proposes the fix for a person to approve.
- Results across the platform: 99% of agent runs complete successfully, 1M+ back-office tasks completed, and 150 hours of work automated for every hour spent teaching Caddi.
- Runs inside 150+ tools (Clio, iManage, NetDocuments, Salesforce, DocuSign, Box, Outlook, QuickBooks, Orion, Wealthbox and more), with no rip and replace.
- Security: SOC 2 Type II. The contents of connected systems are never used to train models. Agents act with one scope per system under the access the person already had, with no standing admin.
## Plans
- Individual: free for one user with no time limit. Includes Caddi Automate and Discover, standard integrations, 200 credits and 10 active loops a month. A low-stakes way to try Caddi on real work. Sign up at https://www.trycaddi.com/pricing.
- Startup: $250 per user per month for companies with up to 10 employees. Adds Box, DocuSign, Clio and other integrations, firm-wide visibility, 1,000 credits per user.
- Business: $6,500 per month for firms standardizing operations on Caddi. Advanced integrations (Salesforce, iManage), unlimited users and loops, 20,000 credits, SSO, 99% uptime SLA. Implementation $5k to $15k one time.
- Enterprise: custom annual agreement with integration governance, reserved capacity, multi-instance integrations and audit logging.
- Business and Enterprise start with a two-week pilot: the firm gets full use of Discover to find where the hours go, Caddi automates the workflow worth the most to them, and the firm finishes with a working automation and the report behind it. Start at https://www.trycaddi.com/schedule-demo.
## When Caddi is a fit
- An operations, finance or practice-support team at a law firm, RIA, accounting firm, insurance agency or PE-backed company that moves the same details between tools by hand every day.
- A firm that wants automation built by the people who do the work, not an IT or RPA project, and that needs every run to be repeatable and auditable.
- A team that does not yet know what to automate first: Discover finds and ranks it.
- Caddi is compared with Zapier, Make, n8n, Power Automate, UiPath, Automation Anywhere, Workato and general-purpose AI agents; the comparisons below cover the differences.
## Guides
- [Workflow automation guide](https://www.trycaddi.com/blog/workflow-automation): What workflow automation is and the best workflow automation software for businesses
- [Best AI agent builders](https://www.trycaddi.com/blog/best-ai-agent-builders): How AI agent builders compare for operations teams
- [AI workflow automation](https://www.trycaddi.com/blog/ai-workflow-automation): How AI changes workflow automation
- [Why Caddi](https://www.trycaddi.com/why-caddi): Why firms choose Caddi
- [Technology](https://www.trycaddi.com/technology): How Caddi connects to and runs inside a firm's tools
## Core pages
- [Home](https://www.trycaddi.com/): What Caddi does and who it is for
- [Use Cases Library](https://www.trycaddi.com/use-cases): Browse 101 real workflows Caddi automates today
- [Integrations](https://www.trycaddi.com/integrations): Every tool Caddi works with, one page per tool below
- [Customers](https://www.trycaddi.com/customers): Law and finance firms running on Caddi
- [Law](https://www.trycaddi.com/law): Caddi for law firms
- [Finance](https://www.trycaddi.com/finance): Caddi for RIAs and finance firms
- [Accounting Firms](https://www.trycaddi.com/accounting-firms): Caddi for accounting and CPA firms: source-document intake, open-items chase, return assembly and 8879s, e-file rejects, CAS close, and WIP billing on autopilot
- [Private Equity](https://www.trycaddi.com/private-equity): Caddi for buy-and-build private equity: Day-1 add-on integration with no ERP migration, consolidated close, chart-of-accounts normalization, platform shared services, and portfolio reporting on autopilot
- [Insurance](https://www.trycaddi.com/insurance): Caddi for insurance brokerages and agencies: new business, COIs, renewals, claims FNOL, and commission reconciliation on autopilot
- [Pricing](https://www.trycaddi.com/pricing): Free Individual plan, Startup, Business and Enterprise; Business and Enterprise start with a pilot
- [Schedule a demo](https://www.trycaddi.com/schedule-demo)
- [Blog](https://www.trycaddi.com/blog)
- [Discover](https://www.trycaddi.com/product/discover): Caddi Discover reads the tools a firm already uses, read only, and ranks the repeated work worth automating with the evidence behind each finding
- [Automate](https://www.trycaddi.com/product/automate): Caddi Automate, where the ops person who does the work builds the automation, with no IT project
- [Govern](https://www.trycaddi.com/product/govern): Firm-wide visibility into work by AI and by people; every run and decision logged and reviewable
- [AI Agents](https://www.trycaddi.com/agents): AI agents that run the repetitive 80% of real back-office roles (billing specialist, accounts payable, accounts receivable, bookkeeper, data entry, document processor, intake specialist, loan processor, client service associate, compliance specialist, operations analyst, and records clerk) as deterministic, auditable code
## Categories
Guides to the software categories Caddi is compared against, including process mining, process intelligence and task mining.
- [Workflow Automation Trends 2026: What Is Changing](https://www.trycaddi.com/category/workflow-automation): The workflow automation trends reshaping 2026: from point-to-point triggers and brittle if/then rules to full workflows you show or describe, run as deterministic code. What changes, and what comes next.
- [How AI Workflow Automation Is Changing in 2026](https://www.trycaddi.com/category/ai-workflow-automation): AI workflow automation is moving past DIY AI canvases that get brittle as they grow. In 2026 the winner is easier to build, maintained by chat, and built for regulated work. Here's what's changing.
- [How AI Agents Are Changing in 2026](https://www.trycaddi.com/category/ai-agents): AI agents are best when they combine LLM reasoning with automation. In 2026 the winning pattern uses AI to craft the right workflow, then runs the actions as deterministic code.
- [How RPA Software Is Changing in 2026](https://www.trycaddi.com/category/rpa-software): RPA is moving from dumb, screen-scraping bots to AI-decisioning, API-first automation that handles unstructured data, runs in parallel, and is fully auditable. How robotic process automation is changing in 2026.
- [How Business Process Automation Is Changing in 2026](https://www.trycaddi.com/category/business-process-automation): Business process automation is moving from rigid rules engines and brittle UI connectors to AI-decisioning, API-first orchestration that handles unstructured data, runs in parallel, and is fully auditable.
- [How Document Automation Software Is Changing in 2026](https://www.trycaddi.com/category/document-automation-software): Document automation is moving from siloed template-and-labeling tools to AI-native automation that sees the whole workflow, handles many data types, integrates everywhere, and is set up by screen recording.
- [How Business Orchestration and Automation Technologies Are Changing in 2026](https://www.trycaddi.com/category/business-orchestration-automation): Business orchestration and automation is moving from heavyweight platforms that need a platform team and long rollouts to AI-decisioning, API-first orchestration that's built and maintained for you.
- [How Intelligent Document Processing Is Changing in 2026](https://www.trycaddi.com/category/intelligent-document-processing): What has changed in intelligent document processing in 2026: IDP is moving from OCR and template extraction that stops at structured data to AI that reads documents natively, decides, and writes to the systems of record. No templates, no model training.
- [How Enterprise Low-Code Application Platforms Are Changing in 2026](https://www.trycaddi.com/category/enterprise-low-code-application-platforms): Enterprise low-code lets you build custom apps with visual modeling instead of hand-coding. In 2026, for a huge class of back-office automation, you don't build an app at all, you record the workflow.
- [What Is Agentic Automation?](https://www.trycaddi.com/category/agentic-automation): Agentic automation is software that uses AI to make decisions and complete real work across your tools, not just follow fixed rules. What it is, how it differs from RPA and AI agents, and how to pick the best platform in 2026.
- [What Is Agentic AI?](https://www.trycaddi.com/category/agentic-ai): Agentic AI means AI that decides and acts on its own. But autonomous agents are unpredictable for real work. What agentic AI is, where it breaks, and why what most teams actually want is agentic automation.
- [What Is Agentic Process Automation (APA)?](https://www.trycaddi.com/category/agentic-process-automation): Agentic process automation (APA) automates entire business processes with AI that makes decisions, not just bots that follow recorded steps. What APA is, how it differs from RPA, and how to choose a platform.
- [What Is Process Mining?](https://www.trycaddi.com/category/process-mining): Process mining maps how work really runs from event logs. What it is, how it works, what it costs, and what happens after it finds a process worth automating.
- [What Is Process Intelligence?](https://www.trycaddi.com/category/process-intelligence): Process intelligence adds context and action to process mining. What it is, how platforms compare, and why intelligence without automation is still a report.
- [What Is Task Mining?](https://www.trycaddi.com/category/task-mining): Task mining records clicks and screens on desktops to map how work gets done. How it works, how it differs from process mining, and a no-recorder route.
## Comparisons
- [Caddi vs. Harvey AI: Automation vs. Legal AI Assistant](https://www.trycaddi.com/compare/caddi-vs-harvey): Harvey is legal-grade AI for research and drafting. Caddi automates the back-office ops work it doesn't touch, and delivers a reliable, deterministic win immediately while Harvey rolls out over months.
- [Caddi vs. ChatGPT: Done-for-You Automation vs. AI Assistant](https://www.trycaddi.com/compare/caddi-vs-chatgpt): ChatGPT is a powerful assistant a person drives. Caddi does the multi-step work across your tools as deterministic code: cheaper per run, repeatable, and fast, with no copy-paste.
- [Caddi vs. Microsoft 365 Copilot: Assist vs. Automate](https://www.trycaddi.com/compare/caddi-vs-copilot): Copilot helps a person inside Office. Caddi runs the cross-tool workflow end to end, moving documents and data between your DMS, CRM, and custodians, unattended and audit-logged.
- [Caddi vs. Claude: AI Assistant vs. Deterministic Automation](https://www.trycaddi.com/compare/caddi-vs-claude): Claude is a strong reasoning assistant for drafting and long documents. Caddi turns a single recording into a maintained, deterministic automation that runs your back office, cheap, reliable, and fast.
- [Caddi vs. Agentforce: In-Salesforce Agents vs. Cross-System Automation](https://www.trycaddi.com/compare/caddi-vs-agentforce): Agentforce runs conversational agents inside Salesforce, priced per action and grounded in Data Cloud. Caddi runs the recurring back-office work that spans Salesforce, the DMS, billing and the inbox, as deterministic code.
- [Caddi vs. Mili: Back-Office Operations vs. an Advisor's AI Assistant](https://www.trycaddi.com/compare/caddi-vs-mili): Mili helps wealth advisors prepare for, document and follow up on client meetings. Caddi finds the repetitive back-office work, automates it end to end, and governs every run.
- [Caddi vs. PitCrew: Find the Work First vs. an Embedded Build](https://www.trycaddi.com/compare/caddi-vs-pitcrew): Both automate back-office operations for financial firms. PitCrew embeds experts to configure prebuilt agents, with a person approving every action. Caddi finds the work, has your ops team build it by showing it once, and governs every run.
- [Caddi vs. Subatomic: Running Processes vs. an AI Chief of Staff](https://www.trycaddi.com/compare/caddi-vs-subatomic): Subatomic's AI Chief of Staff reasons over a firm knowledge graph and hands advisors briefs and drafts to approve. Caddi finds the repetitive back-office processes and runs them end to end, the same way every time.
- [Caddi vs. Humanity Labs: Hybrid Agents vs. a Pure AI Workforce](https://www.trycaddi.com/compare/caddi-vs-humanity-labs): Humanity Labs sells RIAs a managed AI workforce that reasons on every run, which suits work where even you don't know the right way yet. Caddi finds the repetitive work in your systems and runs it the way you want it done, as deterministic code, the same way every time, fully logged.
- [Caddi vs. Akai by Deel: Independent Agents vs. an Agent Network](https://www.trycaddi.com/compare/caddi-vs-akai): Both let ops teams automate their own work without an IT project. Akai runs a shared-context agent network on connectors built from recorded traffic. Caddi runs independent, deterministic agents on verified connections, finds the work in your tools and governs every run.
- [Caddi vs. Worktrace: Reading the Tools vs. Recording the Screen](https://www.trycaddi.com/compare/caddi-vs-worktrace): Both start with discovery, then build the agents. Worktrace installs a desktop app that records screens from install day on and builds computer-use agents on a custom quote. Caddi reads your tools and their history on day one, lets ops build the agent in chat, and runs the fixed steps as code, every run logged.
- [Caddi vs. Nevis: A Full Rebuild vs. Verified Priorities and Verified Runs](https://www.trycaddi.com/compare/caddi-vs-nevis): Nevis is a full rebuild with nothing like Discover, so you automate on a guess and learn the outcome after the move. With Caddi you verify where to start, the hours before you build, then verify the agents handle the work, every run logged.
- [Caddi vs. Power Automate: Done-for-You Automation vs. DIY Flows](https://www.trycaddi.com/compare/caddi-vs-power-automate): Record-to-code AI automation vs. Microsoft's DIY cloud and desktop flows. How each handles UI changes, documents, premium-connector costs, and which fits law & finance back offices.
- [Caddi vs. Zapier: Which Should You Use?](https://www.trycaddi.com/compare/caddi-vs-zapier): Done-for-you record-to-code automation vs. DIY app-to-app Zaps. Build model, document handling, pricing at volume, and which fits law & finance back offices.
- [Caddi vs. UiPath: Which Automation Platform Should You Use?](https://www.trycaddi.com/compare/caddi-vs-uipath): Record-to-code AI automation vs. screen-scraping RPA. How each handles UI changes, documents, and maintenance, and which fits law & finance ops teams.
- [Caddi vs. Gumloop: AI Automation Compared](https://www.trycaddi.com/compare/caddi-vs-gumloop): Done-for-you record-to-code automation vs. a DIY AI canvas. Build model, vertical fit, production reliability, governance, and which fits law & finance teams.
- [Caddi vs. n8n: Build It Yourself or Have It Built](https://www.trycaddi.com/compare/caddi-vs-n8n): A developer's workflow toolkit you host and run vs. automation built and maintained for the operations team. Who builds it, hosting, pricing at volume, and governance.
- [Caddi vs. Make: Visual Scenarios or Automation Built for You](https://www.trycaddi.com/compare/caddi-vs-make): A low-cost visual scenario builder vs. automation the ops person shows once and Caddi builds. Build model, documents and inboxes, pricing, and which fits law and RIA teams.
- [Caddi vs. Automation Anywhere: AI Automation vs. Enterprise RPA](https://www.trycaddi.com/compare/caddi-vs-automation-anywhere): Record-to-code AI automation vs. a Control Room full of screen-scraping bots. How each handles UI changes, documents, and maintenance, and which fits law & finance ops teams.
- [Caddi vs. Certinia: Automation Layer vs. Salesforce-Native Suite](https://www.trycaddi.com/compare/caddi-vs-certinia): A thin record-to-code automation layer over the tools you already run vs. a heavy, Salesforce-locked PSA/ERP you migrate onto. Time to value, ownership, and which fits services ops.
- [Caddi vs. ServiceNow: Record-to-Code Automation vs. an Enterprise Platform](https://www.trycaddi.com/compare/caddi-vs-servicenow): Done-for-you automation live immediately vs. a heavyweight orchestration platform that needs admins, implementation partners, and a quarter to stand up. Time to value, ownership, and fit for law & finance ops.
- [Caddi vs. Appian: Record-to-Code Automation vs. Low-Code Process Apps](https://www.trycaddi.com/compare/caddi-vs-appian): Record a workflow and have it built for you vs. modeling processes and building apps on a low-code BPM platform. Setup effort, who owns it, and which fits regulated back-office teams.
- [Caddi vs. Pega: AI-Native Automation vs. Enterprise BPM Suite](https://www.trycaddi.com/compare/caddi-vs-pega): A lightweight automation layer over the tools you already run vs. Pegasystems' heavy case-management and BPM suite. Implementation time, developer dependence, and fit for law & finance ops.
- [Caddi vs. Workato: Done-for-You Automation vs. DIY Integration Recipes](https://www.trycaddi.com/compare/caddi-vs-workato): Record-to-code automation built and maintained for you vs. building and owning iPaaS recipes yourself. Document handling, maintenance, and which fits non-technical law & finance teams.
- [Caddi vs. Tungsten Automation (Kofax): Whole-Workflow Automation vs. Document Capture](https://www.trycaddi.com/compare/caddi-vs-tungsten-automation): Caddi automates the whole workflow the document lives in vs. Tungsten (formerly Kofax) capturing and extracting data, then handing it to a person. Setup
