Jedify
Jedify · Jedify: Data Agents That Monitor Business Metrics and Investigate Why They Moved, Grounded in a Company Context Graph
Jedify is an enterprise analytics platform whose Data Agents are AI analysts you describe in plain language: they run on demand, on a schedule or when a metric crosses a threshold, investigate the likely causes, and deliver the analysis in Jedify, Slack or email or through MCP clients such as Claude. It suits data teams at mid-size and large companies that want finance, RevOps, operations and supply chain staff to get recurring, governed analysis without writing SQL.
Best for
Data teams at mid-size and large companies with a cloud warehouse who are tired of rebuilding the same reports, and who want finance, RevOps, operations or supply chain staff to set up their own monitoring and root-cause agents on governed metric definitions with per-user row-level permissions
Not ideal for
Small teams or individuals looking for a cheap self-serve AI analyst, companies without a warehouse or someone to own metric definitions, teams that need to self-host on AWS or Azure today, and buyers who want agents that change ERP or CRM records out of the box
Who it's for
Data teams and the finance, RevOps, operations and supply chain staff they support at mid-size and large companies with data in a cloud warehouse
Capabilities
- Data Agents created by describing the analysis in plain language (what to track, metric definitions, what counts as unusual, output format) and refined through conversation instead of a workflow builder
- Three ways to trigger an agent: on demand, on a schedule, or when a defined threshold is crossed
- Agent outputs are analytical reports with visualizations that stay conversational: you can ask a past run why something happened, break it down by region or compare it with the previous run
- Deep Research agent that plans a multi-step investigation, tests competing explanations across internal data and web search, and reports the evidence and what remains unanswered
- Semantic Fusion context graph built from connected sources, where each concept and metric carries a business definition, base SQL query and relationships, with a business glossary, synonyms, time-analysis rules and answer guidance
- Git-based versioning of the context graph with branch review, YAML diffs, restore of any version from the last 90 days, and optional hosting of the repository in your own GitHub
- Test sets that rerun a list of questions against the model and compare outputs to a golden standard before changes are published
- MCP server with Asker, Editor and Builder modes assigned by user role, available as a connector in Claude's directory or through a local proxy for Claude Desktop, Cursor and Claude Code
- REST API that takes natural-language questions and returns structured answers, plus an @jedify/sdk TypeScript SDK for calling saved semantic functions from apps
- Slack app for DMs and channel threads, email delivery of agent results, a Looker extension, and live homepage widgets that refresh with new data
- Per-user data access: Jedify forwards each user's identity to Snowflake so row access policies apply, and RBAC separates Users, Data Experts and Admins
- Self-hosted installation on Google Cloud (GKE Autopilot with Vertex AI by default, or your own OpenAI-compatible model gateway), including air-gapped setups
Limitations
- No pricing on jedify.com; the only published prices are Jedify's AWS Marketplace contracts, which start at $1,500 a month for up to 5 connected tables
- AWS Marketplace tiers cap connected tables (5, 10 or 25 below Enterprise) and set a starting prompt allowance; extra usage needs a new purchase order rather than automatic overage billing
- Answer quality depends on the context graph, which data experts are expected to review, tune and test, so it is not a drop-in tool for teams without data ownership
- Documented setup guides cover warehouses, BI tools, dbt and Athena; the ERP, MES, TMS and WMS connections described on Jedify's solution pages have no public setup docs
- Action-taking examples such as creating ERP stock transfers are shown on Jedify's site with human approval, but the self-serve Data Agents documentation describes analysis and delivery only
- Slack responses are text only, with no embedded charts, and Jedify must be @mentioned outside direct messages
- Over MCP, accounts with the User role get only Asker (question) tools; editing the model or building apps needs a Data Expert or Admin role, and a published app built on an API key shows every viewer data at that key's permission level
- Self-hosting is documented only on Google Cloud; AWS and Azure self-hosted installs are listed as coming soon, although Jedify's site advertises cloud, VPC or on-prem deployment
Use cases
- Scheduling a weekly sales performance agent that flags any metric moving more than 15% week over week, investigates the top drivers and posts the result to Slack
- Asking a Deep Research agent why enterprise churn rose last quarter and getting the matched accounts, the evidence and the SQL behind it
- Letting a finance or RevOps lead build their own recurring revenue or pipeline report instead of filing a request with the data team
- Monitoring supply chain exceptions such as at-risk orders or stockouts and getting a ranked list with the likely root cause attached
- Asking governed business questions from Claude or Cursor through Jedify's MCP server, with answers resolved against the company's metric definitions
- Building an internal dashboard from Claude Code in Builder mode using saved semantic functions and the Jedify SDK
Our take
Jedify is worth a look if your bottleneck is not access to data but trust in the answers. Its Data Agents are only as good as the context graph underneath, and that is where Jedify has invested most: metric definitions with their SQL, versioned like code, tested against expected answers, and enforced per user down to Snowflake row access policies. That makes it a better fit for a data team that wants to hand recurring reporting and why-did-this-move investigations to business users than for a business user shopping for an AI analyst on their own. Read the marketing carefully. The ERP-writing supply chain agents on the homepage are presented as approved, human-reviewed actions, and the public docs back analysis and delivery rather than self-serve write-backs, and the context graph needs ongoing ownership. Pricing is also a real gate: Jedify's AWS Marketplace contracts start at $1,500 a month for 5 tables, so run the trial against one recurring report your team already rebuilds by hand before committing.
Who should use it
Heads of data and analytics engineers at companies with a cloud warehouse and a dbt or BI layer who want business teams to set up their own scheduled or threshold-triggered analysis on governed definitions, and organizations that want those answers inside Slack, Claude or their own apps, or self-hosted on Google Cloud.
Who should skip it
Small teams wanting a low-cost self-serve analyst, companies whose data is mostly in spreadsheets or SaaS apps without a warehouse, organizations that need a self-hosted install on AWS or Azure today, and teams looking for agents that autonomously write back to ERP or CRM systems without a scoped project.
Strengths
- Agents answer from the company's own metric definitions and relationships rather than guessing joins from raw schemas
- Business users build agents by describing the analysis, with on-demand, scheduled and threshold triggers
- Agent outputs stay conversational, so a scheduled report can be questioned and broken down after it arrives
- Strong data-team controls: Git-based model versioning, test sets against a golden standard, RBAC and per-user Snowflake row-level security
- Reachable outside Jedify through Slack, email, a Claude connector, an MCP server, a REST API and a TypeScript SDK
- Documented self-hosted install on Google Cloud, including air-gapped environments
Weaknesses
- No pricing on Jedify's site; AWS Marketplace contracts start at $1,500 a month for 5 tables
- Needs a data team to connect sources and curate the context graph before agents are trustworthy
- ERP, MES and supply chain system connections are marketed but not covered by public setup docs
- Slack answers are text only, without charts
- Self-hosting on AWS or Azure is not yet available
Jedify pricing
Starter
$1,500/month
Billed monthly
- Up to 5 tables
- Starts at 1,750 prompts
- 1-month contract price on AWS Marketplace
Standard
$2,500/month
Billed monthly
- Up to 10 tables
- Starts at 5,000 prompts
Pro
$3,500/month
Billed monthly
- Up to 25 tables
- Starts at 8,000 prompts
Enterprise
$5,500/month
Billed monthly
- Unlimited tables
- Starts at 15,000 prompts
Custom
$100,000/month
Billed monthly
- Table and prompt limits set independently
- Listed price on AWS Marketplace; arrange terms with Jedify
Note: Jedify does not list prices on its website. These are the 1-month contract prices on Jedify's AWS Marketplace listing (sold by Jedify), which also offers a free trial; 12, 24 and 36-month contracts save up to 33%, 40% and 50%. A prompt is one query run against your data and a table is a connected data source; going past a tier's limits requires a new purchase order for the extra quantity. SiliconANGLE reported at launch that companies can start on Data Agents for free, but Jedify publishes no free plan limits. Direct contracts and self-hosted deployments are arranged with Jedify's sales team.
Where Jedify excels
Monday performance report that investigates itself
Instead of an analyst rebuilding the weekly report, a Data Agent can run on a schedule, flag metrics that moved past a threshold, investigate the likely drivers and send the result to Slack or email, where leaders can keep asking follow-up questions.
Root-cause analysis on a KPI drop
The Deep Research agent breaks a question like why churn or revenue fell into steps, checks several explanations against CRM, billing and web data, and shows the evidence and SQL it used, which makes the answer reviewable by the data team.
Governed answers inside Claude
With the Jedify connector in Claude, employees can ask business questions where they already work, and answers resolve through the company's metric definitions and their own data permissions instead of a model reading raw tables.
Jedify vs. competitors
Jedify vs. AgentSpot
AgentSpot is ThoughtSpot's no-code workspace for building and sharing agents, scheduled workflows and data apps, grounded in ThoughtSpot data models through Spotter. It is hosted SaaS with a free plan and per-outcome credits, and its workflows act in tools such as Slack, Google Workspace, Microsoft 365 and Jira, but it has no event triggers yet. Jedify builds its own versioned context graph from your warehouses, BI tools and dbt, adds threshold-triggered agents and a Deep Research agent for root-cause work, offers a documented Google Cloud self-hosted install, and is sold on contracts that start at $1,500 a month on AWS Marketplace.
Frequently asked questions
What are Jedify Data Agents?
Data Agents are AI analysts inside Jedify that you create by describing the analysis you want in plain language: what to track, how to define it, what counts as unusual and how to format the result. You refine the agent by chatting with it, then run it on demand, on a schedule or when a defined threshold is crossed. Each agent works from Jedify's context graph of your company's metrics and data, and its results can be read in Jedify, delivered to Slack or email, or reached from MCP clients such as Claude and ChatGPT.
How much does Jedify cost?
Jedify does not publish prices on its website. Its AWS Marketplace listing, sold by Jedify, offers a free trial and monthly contracts: Starter at $1,500 a month (up to 5 tables, starting at 1,750 prompts), Standard at $2,500 (10 tables, 5,000 prompts), Pro at $3,500 (25 tables, 8,000 prompts), Enterprise at $5,500 (unlimited tables, 15,000 prompts) and a Custom tier, listed at $100,000 a month, whose table and prompt limits are set independently. 12, 24 and 36-month contracts save up to 33%, 40% and 50%. SiliconANGLE reported at launch that companies can start on Data Agents for free; Jedify's site offers a sign-up and a demo booking, but no free plan limits are published.
What data sources does Jedify connect to?
Jedify's docs have setup guides for Snowflake (including row-level security), BigQuery, Databricks, Redshift, AWS Athena, Looker, Tableau, Sigma, Redash, Preset and dbt. Its connectors page also lists ClickHouse, PostgreSQL, MongoDB and Salesforce, and knowledge sources such as Notion, Jira, Google Docs, SharePoint, S3 and uploaded documents. Jedify's solution pages describe connections to ERP, TMS, WMS and MES systems, but those have no public setup guides, so confirm them with Jedify.
Can Jedify agents take actions in other systems?
Jedify's homepage shows supply chain agents that, once a person approves, open a stock transfer in an ERP or issue an expedite PO, and its supply chain page says nothing acts on your systems without review. The self-serve Data Agents documentation describes analysis, recommendations and delivery to Jedify, Slack and email rather than writes to other systems, and Jedify's engineers and AI Studio build custom agentic solutions, so check which actions are included in your deployment.
Does Jedify work with Claude and ChatGPT?
Yes. Jedify runs an MCP server and is listed as a connector in Claude's directory, which works in the Claude web app, Claude Desktop and Claude Code. Local clients such as Cursor connect through the @jedify/mcp-auth proxy, and hosted agents use the server URL with an API key. Jedify says MCP clients like Claude and ChatGPT can list your active Data Agents and ask them questions. Your role decides the tools you get: Users get Asker mode for questions, while Data Experts and Admins also get Editor mode for changing the semantic model and Builder mode for building dashboards and apps.
Can Jedify be self-hosted?
Yes, on Google Cloud. Jedify documents a self-hosted install into your own Google Cloud project on GKE Autopilot using OpenTofu and Helm, with sign-in through your own identity provider and model calls going to Vertex AI by default or to your own OpenAI-compatible gateway; air-gapped installs are covered too. Self-hosted installs on AWS and Azure are listed as coming soon. Jedify's website also mentions cloud, VPC and on-prem options and SOC 2 compliance.
Jedify vs ThoughtSpot AgentSpot: what is the difference?
Both let business users describe agents in plain language and ground answers in a governed semantic layer. AgentSpot is ThoughtSpot-hosted, builds on ThoughtSpot data models, has a free plan and per-outcome credit pricing, and its scheduled workflows also act in tools like Google Workspace, Microsoft 365 and Jira. Jedify builds its own context graph from your warehouses, BI tools and dbt, supports threshold-triggered agents and root-cause research, can be self-hosted on Google Cloud, and is sold through contracts that start at $1,500 a month on AWS Marketplace.
Integrations & fit
About Jedify
Jedify's core product is a context graph it calls Semantic Fusion: it connects to a company's warehouses, BI tools and dbt project, plus documents and other knowledge sources, and builds a model in which each business concept or metric carries its definition, its underlying SQL and its relationships to other entities. Data experts review and edit that model in natural language or SQL, with Git-based versioning, branch review, rollback and test sets that compare answers against a golden standard. Everything else sits on top of it: Ask Jedify for conversational questions, a Deep Research agent that plans a multi-step investigation and reports what the evidence does and does not support, live dashboard widgets, and Data Agents, which Jedify introduced in August 2026 and formally launched through its Agents Hub on October 28, 2026. A Data Agent starts as a conversation rather than a workflow builder: you describe what to track, how to define it, what counts as unusual and how the output should look, refine it by chatting, then run it on demand, schedule it or have it fire on a defined threshold. Results are analytical reports with charts that you can keep questioning, and they can be delivered to Slack or email; MCP clients like Claude and ChatGPT can list your active agents and ask them questions. Jedify also sells AI Studio, where its own team builds custom agentic solutions such as revenue forecasting or meeting prep, and its homepage shows supply chain agents that, after a human approves, open stock transfers in an ERP or issue expedite POs. Treat those action-taking agents as part of a scoped deployment rather than the self-serve Data Agents, whose documented outputs are analysis and recommendations. The distinctive bet is grounding: agents answer from the company's own metric definitions and inherit source permissions, including Snowflake row-level security per user, instead of guessing joins from raw schemas. The tradeoffs are cost and setup. The only published prices are Jedify's AWS Marketplace contracts, from $1,500 a month for up to 5 tables, the value depends on a data team curating the context graph, Slack answers are text only, and self-hosting is documented only on Google Cloud, with AWS and Azure listed as coming soon.
Updates from Jedify
Jedify formally launched Data Agents, autonomous AI analysts that business teams set up in plain language and run on demand, on a schedule or when a condition is met, available through its Agents Hub. Jedify first introduced Data Agents, Agents Hub and AI Studio custom solutions in an August 2026 blog post.
Jedify's MCP server now returns tools based on the user's role: Tenant Admins and Data Experts get Asker, Editor and Builder modes, while Users get Asker. This applies to every connection method, including the Claude connector.
The Semantic Fusion model became Git-managed, with edits on branches, side-by-side YAML diffs, merge and decline, restore of any version from the last 90 days, and an option to host the repository in your own GitHub.
Jedify became a connector in Claude's directory, so users can connect and sign in without a local install; on Claude Team and Enterprise an admin can enable it for the whole organization. At launch the connector offered Asker (question) tools only.
Are you the founder? Claim this listing →