Solutions

One platform. Three levels of help.

The Palm Valley AI platform is the same for everyone. The difference is how much we set up with you, from DIY to fully managed. Today the product is a private LLM with all the bits you need around it. Custom workflows and agents are services we layer on top.

The platform

Everything everyone gets.

Whether you're on Spark, Forge, or Atlas, the platform itself is identical. Same capabilities, same security, same web and iOS apps.

Web & iOS apps

Chat from any browser at work or jump to the iOS app on the go. Same workspace, same conversations, same context grounded in your knowledge.

Budget controls

Set monthly token and dollar caps per user. We notify before the limit and stop spend cold at it. No surprise invoices, no runaway costs.

Any model, routed

Use Claude, Amazon Nova, Llama, and more. Pick per use case or let us route to the best fit for each task, balancing quality, speed, and cost.

Self-serve RAG

Drop in your docs, PDFs, and slides. The platform grounds answers in your content with source citations, no engineering required.

Enterprise security

Hosted in our AWS on AWS Bedrock. Your data never trains a model. Encrypted in transit and at rest. SOC 2 inherited from the Bedrock platform.

Admin & audit

Manage users, configure spend caps, see live usage dashboards, and review a full audit log of usage and admin actions. Built for ops teams.

BUDGET CONTROLS

AI spend that can't surprise you.

Most AI tools bill you after the fact and hope you trust them. Palm Valley AI enforces budgets up front, every prompt is metered against a per-user monthly cap, and spend stops cold at the limit.

  • Per-user monthly token and dollar caps
  • Warning thresholds before users hit their limit
  • Hard stop at the cap, no overage billing
  • Tiers and per-user overrides for power users
  • Live admin dashboard + monthly usage exports
Talk to us →
alex@acme.com
growth · 42%
$84.30 / $200.00 · 8 days remaining
jamie@acme.com
growth · 88%
$176.40 / $200.00 · warning sent
priya@acme.com · override
$500/mo
$108.00 / $500.00 · custom cap
MODEL CHOICE

Pick any model. Or let us pick.

The right model depends on the task. Code generation isn't the same workload as document summarization. Palm Valley AI gives you access to the best models on AWS Bedrock and routes to the right one for each task, so you get quality where it matters and savings where it doesn't.

  • Anthropic Claude, Amazon Nova, Meta Llama, and more
  • Switch models per use case or per conversation
  • Optional auto-routing for speed and cost efficiency
  • Add or restrict models per workspace
  • Bring your own model on Atlas
Claude Sonnet
reasoning · drafting
recommended balanced quality & cost
Amazon Nova Pro
analysis · long context
cost-efficient fast on long docs
Llama 3.3 70B
open · routine tasks
cheapest good enough for most queries
+ 9 more available · auto-routed
SELF-SERVE RAG

Your company knowledge, grounded.

Drop in your docs and the platform grounds every answer in your content. Every response shows its sources so people can trust, and verify, the answer. No data leaves your security boundary.

  • Upload PDFs, Word docs, slides, Markdown, plain text
  • Per-document permissions preserved (Forge / Atlas)
  • Citations on every answer, click through to the source
  • Re-index on demand or on schedule
  • Custom embeddings & chunking strategies on Atlas
See how it works →
Q: What's our refund SLA for enterprise?
Net-15 from date of approved request. Approvals require a CSM signoff per refund-policy.pdf · pg 4.
// SOURCES
📄 refund-policy.pdf · uploaded by Alex M.
📊 enterprise-contracts.xlsx · uploaded by Sara L.
📝 cs-runbook.md · uploaded by Marcus J.
Professional services

What scales with the tier.

The platform is the same for everyone. Where Forge and Atlas differ is the work we do with you, from setup to ongoing custom builds.

FORGE · ONBOARDING

Workspace & data setup

We run a scoping workshop, configure your workspace, set up budgets and tiers for your team, connect your documents to the knowledge base, and train your team on how to get the most out of it.

FORGE · FUTURE

Custom domain

Host the platform on your domain, e.g. ai.yourcompany.com, so it feels like your internal tool. Available in a future release.

ATLAS · CUSTOM BUILD

Custom workflows & insights

We design and build deeper workflows on top of the platform, custom prompts and RAG configurations tuned to specific tasks, plus reports and insights stitched into your existing tools.

ATLAS · ONGOING

Ongoing optimization

A dedicated solution architect who knows your stack. Continuous improvements to your configurations, new use-case rollouts, and quarterly business reviews to keep things moving.

See Private AI pricing →
Built on Bedrock

Secure by architecture, not by policy.

Every tier runs on the same foundation. AWS Bedrock in our managed cloud, with security guarantees inherited from the platform itself.

Hosted in our AWS

Nothing for your team to stand up or maintain.

Never trains a model

Your data isn't used to improve any foundation model.

Encrypted end-to-end

In transit and at rest. Private connectivity.

Compliance inherited

SOC 2 from Bedrock; HIPAA & residency on Atlas.

See it on your data.

Book a 30-minute working session. We'll connect a sandbox to your docs and show you what's possible, live.