A Palantir Foundry alternative that runs on your own warehouse
An ontology-style model of your business, actions back into your tools and governed AI on top, for mid-market teams. No multi-year platform contract, no team of forward-deployed engineers: Rig's FDE agent builds the context layer on the Snowflake, BigQuery or Databricks you already run.
What does a Foundry-style programme cost?
Palantir does not publish Foundry pricing. Its AWS Marketplace listing is private pricing only, and its own quarterly filing says contracts generally run one to five years. What is public is the shape of the business: in the 12 months to 30 June 2026 its top 20 customers averaged $124M of revenue each. That is not what a mid-market team would pay, and we will not guess what you would. It does tell you who the platform is built around.
The contract
Private pricing, negotiated per customer, on terms Palantir describes as generally one to five years. The commitment comes before you have seen it work on your own data at your own scale.
The people
Palantir's model embeds forward-deployed engineers with the customer, and your own team builds pipelines, the Ontology and Workshop apps beside them. That is people time, on both sides, that a licence line does not show.
The platform
Foundry is a platform your team works inside: Pipeline Builder, Code Repositories, the Ontology, Workshop, AIP. It can read your warehouse through virtual tables, but the model, the apps and the permissions live in Foundry.
Three public tiers, unlimited seats on all of them
A first warehouse and a handful of sources.
The whole team self-serving, daily syncs across the stack.
Agents running all day, writes going back into the tools.
Entry prices on a bring-your-own warehouse, from the pricing page. Start is $499 a month if Rig hosts the warehouse. Enterprise, inside your own cloud account with dedicated FDE hours, is quoted per engagement.
Who runs on Rig
Mid-market companies that wanted a governed model of their business and AI on top of it, and got there in weeks. Every figure is from a published case study.
40+ sources in one warehouse, every core metric certified, and in their words a solo data function outputting what would usually take a 10+ person team.
Read the case studyA 4,000-table warehouse opened to fraud, ops, support and marketing through one governed context layer. 16 weeks from a fraud-team pilot to org-wide rollout.
Read the case studyFrom a small early-adopter group to most of the commercial org querying the warehouse directly inside Claude in two months, with 4,000+ Rig MCP tool calls in the last 30 days.
Read the case studyA BigQuery estate the commercial team could not query, opened up to sales managers, country leads and revenue ops in two months.
Read the case studyIs there anything like the Ontology?
Yes, and it is the part of Rig that matters most. Foundry's Ontology maps your data onto the things your business runs on, and lets people and AI act on them. Rig's context layer does the same job over your warehouse. Here is how the concepts line up, and the difference is in who builds it.
| In Foundry | In Rig | What it does |
|---|---|---|
| Object types | Tables with meaning attached | Every table and column in your warehouse described: what it is, what a row means, which one is the source of truth. |
| Link types | Inferred joins | The relationships between tables, read from the schema and from the queries people actually run, then kept current as the schema drifts. |
| Properties and business definitions | Business terms and certified metrics | Revenue, active customer, churn: defined once, certified, and used by every agent, dashboard and report so two people get one number. |
| Action types | Actions and write-backs | Agents and workflows that update the CRM, post to Slack, raise a ticket or send a report, with a human sign-off where you want one. |
| Ontology permissions | Row, column and role-level access | Enforced per user on every query, whether the question came from a person, a dashboard or Claude over MCP. |
The honest difference: the Ontology is a richer object model, designed to be the backbone of operational applications. Rig's context layer is designed to make a warehouse answerable and actionable by people and agents, and to build itself. If you need the former, see where Foundry wins, below.
Who does the forward-deployed engineering?
Palantir describes its forward-deployed engineers as engineers embedded directly with customers to tackle their most pressing problems. A large part of that work is building the Ontology: working out what each dataset means, how it links to the next, and which actions people should be able to take on it.
Rig's CTO did that job at Palantir from 2022 to 2024. Rig exists because most of it can now be done by an agent. In his words:
"At Palantir, we built the ontology for major enterprises by manually constructing complex maps of their internal data. In the agent era, those maps should build and maintain themselves."
- 1
Connect the warehouse
Read-only access to Snowflake, BigQuery, Databricks or Redshift. Rig reads the schema, samples values and studies the queries your team already runs.
- 2
The FDE agent drafts the context layer
Table and column descriptions, joins, business terms and candidate metrics, written for you rather than modelled by hand in workshops.
- 3
Your people certify it
Data owners review and sign off the definitions that matter. Certified metrics become the one answer every agent and dashboard uses.
- 4
It keeps itself current
When the schema changes, the agent re-reads it and flags drift, so the model does not quietly go stale the way a hand-built one does.
Foundry and Rig, side by side
Foundry product names on the left, what Rig does for the same job on the right.
| Capability | Palantir Foundry | Rig |
|---|---|---|
| Where the data lives | Foundry datasets, or virtual tables that query Snowflake, BigQuery and Databricks in place, governed through Foundry | Your own Snowflake, BigQuery, Databricks or Redshift. Or Rig hosts one for you |
| Semantic model | The Ontology: objects, links and actions | Context layer: table meaning, joins, business terms and certified metrics, built by an agent |
| Who builds it | Your engineers, often alongside Palantir's forward-deployed engineers | Rig's FDE agent drafts and maintains it. Your team reviews and certifies |
| Pipelines | Pipeline Builder and Code Repositories | 300+ managed connectors on dlt, transformations in dbt Core |
| Operational apps | Workshop apps built on the Ontology | Data apps and dashboards generated from certified metrics. Simpler, and less of a UI toolkit |
| Writing back | Actions on Ontology objects | Actions into 300+ tools: CRM updates, alerts, tickets, reports |
| AI | AIP, working over the Ontology | Rig agents, plus one governed MCP endpoint for Claude, ChatGPT or any client you already use |
| Deployment | Cloud, on-prem and air-gapped, with FedRAMP High, IL5 and IL6 authorisations | Rig cloud, or inside your own cloud account on Enterprise. No air-gapped option |
| Pricing | Not published. Private pricing, contracts generally one to five years | Public tiers from $499 a month, unlimited seats |
Your warehouse stays the platform
Rig sits on the Snowflake, BigQuery, Databricks or Redshift you already pay for. Your models stay in dbt, your data stays in your account, and the tools you already use keep working.
The FDE work is automated
The part of a Foundry programme that takes people months, mapping what every table means and how it joins, is what Rig's FDE agent does on connection. It then keeps doing it as the schema changes.
Bring the AI you already use
Point Claude, ChatGPT or Cursor at one governed MCP endpoint. Permissions, sandboxed SQL and a full audit log come with it, so nobody needs raw warehouse credentials.
A price you can read before the first call
Public tiers from $499 a month, unlimited seats on every plan. No multi-year commitment to find out whether it works.
Where is Foundry the better choice?
Four places, stated plainly. If one of them is you, stay on Foundry.
Classified, government and air-gapped work. Palantir deploys into air-gapped environments and holds FedRAMP High, IL5 and IL6 authorisations. If you need that, Foundry is the answer and we are not.
Operational applications. Workshop exists to build interactive applications for operational users on top of the Ontology, and actions write back to Ontology objects as single transactions. If you are building that kind of operational software, Foundry is the platform for it. Rig's data apps and actions are real, but they are not a full application toolkit.
The very largest estates. Palantir reports 1,049 customers, and its top 20 averaged $124M of revenue each over the year to June 2026: the platform is built around organisations of that size. Our largest customer warehouses run to thousands of tables, which is a different order of problem.
Maturity and a reference list. Palantir is a public company with a long list of government and commercial customers. We are a younger company. If procurement weights vendor age heavily, that is fair to weigh.
Is there an open source Palantir alternative?
Straight answer: Rig is not open source. Neither is Foundry: Palantir publishes some open-source libraries, such as the Blueprint UI toolkit, but the platform itself is commercial subscription software. We are not aware of an open-source project that covers what Foundry does end to end.
What is open is everything underneath Rig that holds your data. Ingestion runs on dlt and transformations on dbt Core, both open source and both runnable without us. The warehouse is your own account, and the models are SQL in a git repository you own. What is Rig's is the context layer, the agents and the interface on top.
If you are searching for "open source" because you want to avoid lock-in rather than to self-host everything, that split is usually what you are after: the expensive, hard-to-move parts are open, and the part you would be paying for is the part that saves your team the work.
What am I locked into?
Very little, by design. Your data is in your warehouse, which you pay for directly. Your pipelines are dlt, your transformations are dbt Core, and both live in a repository you control. Your AI clients talk to Rig over MCP, an open protocol, so the tools people use every day are not ours either.
If you left Rig, or if Rig disappeared, what you would lose is the context layer and the interface. Not the data, not the models and not the plumbing. That is the difference between a platform you live inside and a layer you put on top of what you already own.
Everything else
For mid-market and growth-stage companies, yes. Rig gives you the parts of Foundry most teams actually buy it for: a semantic model of your business over your data, actions that write back into your tools, and governed AI on top. It does it on your own warehouse, with public pricing, and without a team of forward-deployed engineers. It is not a replacement for Foundry in defence, classified or air-gapped environments.
Not a complete one that we know of. Foundry is commercial software, and Rig is not open source either. What is open source is the plumbing underneath Rig: ingestion runs on dlt and transformations on dbt Core, both of which you can run without us. The warehouse is yours and the models are SQL in your own git repository. The context layer and the agents are Rig's product.
Not from us. Rig's public tiers start at $499 a month, which is on the pricing page, and you can try it before you talk to anyone. The free parts of the stack are the open-source ones: dbt Core and dlt, plus whatever warehouse free tier you are on.
The context layer. It describes what every table and column means, how tables join, which business terms and metrics are certified, and who can see what. Foundry's Ontology is a richer object model built to drive operational applications. Rig's context layer is built to make your warehouse answerable and actionable by people and AI agents, and it is generated for you rather than modelled by hand.
Rig's FDE agent. On connection it reads the schema, samples values, studies the queries your team already runs and drafts descriptions, joins and metric definitions. Your data owners review and certify them. As the warehouse changes, it re-reads and flags drift. On Enterprise there are dedicated FDE hours from our team as well, for the parts that need a person.
No. Rig runs on the Snowflake, BigQuery, Databricks or Redshift you already have. If you do not have a warehouse yet, Rig can host one, and it is still a standard warehouse you can take with you.
Rig runs in our cloud, or inside your own cloud account on the Enterprise plan. It does not run air-gapped or in classified environments. If that is a requirement, Foundry is built for it and Rig is not.
Connect a warehouse and the FDE agent starts building context the same day. Real customers have gone from first connection to a company-wide rollout in weeks: Suri centralised 95% of its sources in under two months, and Cleo went from a fraud-team pilot to org-wide in 16 weeks.
Yes, if your data also lands in a warehouse Rig can read. Some teams keep Foundry for the operational programme it was bought for and use Rig to give the wider company governed, self-serve access to the warehouse without adding Foundry seats or builders.
Start with the problem you would buy Foundry for
Twenty minutes. Tell us the decisions you want your data to drive and the warehouse it lives in, and we will show you the context layer Rig would build on it. If the honest answer is Foundry, we will say so.
Twenty-minute call about the problem and the warehouse
Warehouse connected, the FDE agent starts building context
Your team certifies the metrics that matter and starts asking questions
Palantir product names and public facts checked 28 September 2026 against Palantir's own documentation and filings. Palantir, Foundry, AIP, Ontology and Workshop are Palantir Technologies' names, used here only to describe their products. Rig prices are the public tiers on rig.so/pricing. Customer figures are from Rig's published case studies.