An internal LLM platform that runs on AWS Bedrock inside our managed cloud. Your data stays in a secure boundary, encrypted, private, and never used to train a model. Choose how much help you want: do it yourself, do it with us, or have us run it for you.
Every level runs on the same architecture. The security posture never changes, only how much we do for you does.
Runs on AWS Bedrock inside our managed cloud. No infrastructure for your team to stand up or maintain.
Your prompts and company data are never used to improve any foundation model, ours or a provider's.
All data is encrypted in transit and at rest. Private connectivity, no exposure to the public internet.
Select one or many foundation models. Claude, Nova, Llama and more, and switch per use case.
Connect company knowledge so answers are based on your documents, with source citations and permissions preserved.
Built on Bedrock's compliance footprint. SOC 2, HIPAA-eligible, and FedRAMP-eligible deployments available.
The platform is free at every level. You only pay for what you use, plus services where you want a hand.
Private AI, out of the box. Set up in minutes, connect your company context, and start building, no help needed.
We build your first use cases alongside your team, connecting real data, configuring RAG and workspaces, then hand you the keys.
A fully managed AI function. We design, build, run, and continuously improve it, you set the priorities and review the outcomes.
Three simple parts. The first two are identical at every level, only the services change.
Flat monthly platform fee. No per-seat licensing, no ramp-up cost. Includes the workspace, admin, model selector, budget controls, and both the web and iOS apps for your whole team.
Every request runs on AWS Bedrock. We pass through the exact AWS token cost and add a transparent margin, visible on every invoice. No hidden markup, no surprise bills.
Spark: nothing extra, fully self-serve. Forge: one fixed-scope implementation fee from $5,000. Atlas: outcome-based pricing with a fixed monthly support model.
| SparkDo it yourself | ForgeDo it with you | AtlasHave us do it | |
|---|---|---|---|
| Getting started | |||
| Setup | Self-service | Guided with our team | Fully managed |
| Time to first use case | Hours–Days | 2–6 weeks | Ongoing partnership |
| Onboarding | Docs & guides | Scoping workshop | Dedicated kickoff |
| Platform | |||
| Choice of foundation models | ✓ | ✓ | ✓ |
| Ground responses in company data | Self-upload within chat | We implement RAG | We manage |
| Usage dashboards & cost controls | ✓ | ✓ | ✓ |
| Hosted in our AWS on Bedrock | ✓ | ✓ | ✓ |
| Build & implement | |||
| Data source integration | Self-serve | Done with you | Done for you |
| Custom prompts & RAG configurations | Self-serve | Configured with you | Built & owned by us |
| Team training & enablement | — | ✓ | ✓ |
| Production use cases delivered | You build | 1–3 at launch | Continuous |
| Operations | |||
| Monitoring & evals | Self-managed | We set up, you run | Fully managed |
| Model updates & optimization | Self-managed | Guidance provided | Fully managed |
| New use cases over time | You build | Add-on engagements | Included, continuous |
| Support | |||
| Channels | Docs, community, email | Priority support | Dedicated team, 24/7 |
| SLA, uptime & response | — | — | ✓ |
| Account team | — | Engagement contact | Dedicated architect |
| Security | |||
| Encryption in transit & at rest | ✓ | ✓ | ✓ |
| Data never used to train models | ✓ | ✓ | ✓ |
| Dedicated VPC | — | — | ✓ |
| HIPAA BAA / custom data residency | — | — | ✓ |
| Pricing | |||
| Platform fee | $200 / month | $200 / month | $200 / month |
| Usage | Bedrock cost + margin | Bedrock cost + margin | Bedrock cost + margin |
| Services | $0, self-serve | From $5,000, fixed-scope | Outcome-based + fixed support |
The platform runs on AWS Bedrock inside our managed AWS environment. Your data is processed within that secure boundary, encrypted in transit and at rest, and connected over private networking, never exposed to the public internet. For Atlas, we can deploy into a dedicated VPC with custom data residency.
Never. Your prompts, documents, and company context are not used to train or improve any foundation model, not ours, and not the model providers'. This is a hard architectural guarantee inherited from how Bedrock is built, and it applies identically at every level.
Two line items on every invoice: the $200/month platform fee (flat) and usage at AWS Bedrock cost plus a transparent margin. Both are visible line items. No per-seat fees, no minimums. A quiet month just costs the $200 base.
Yes, that's the design. Most teams start on Spark to pilot, move to Forge when they want help getting a real use case into production, and graduate to Atlas when they'd rather we run the whole function. You can also enter directly at any level.
A defined, priced package, not open-ended consulting hours. We agree the scope up front (discovery, secure data integration, configuration, training, handoff) and deliver 1–3 use cases live in production, typically in 2–6 weeks. You own and operate them afterward, with full documentation.
Atlas pricing is tied to the results we deliver rather than hours billed, paired with a fixed monthly support model so your spend is predictable. We define the outcomes and measures together during the kickoff. Platform and usage are billed the same as every other level.
Any model available in AWS Bedrock, including Anthropic Claude, Amazon Nova, Meta Llama, and others. You can use several at once and route each use case to the best-fit model. Switching models never requires rebuilding your setup.
Stand up a free, secure AI workspace today, or talk to us about doing it with you, or for you.