Pricing

Hosted, self-hosted, or enterprise.

Three paths. Pick the one your security team will sign off on. Switch later if it changes.

Hosted

Managed SaaS
$1,999 per month

+ LLM token usage at cost (passed through)

We run Inhouse Agents on isolated AWS infrastructure. Fastest path to a working AI workforce — sign up, OAuth into your tools, agents come online in 60 seconds.

Start hosted trial
  • ✓All 8 starter roles + 2 Team Packs
  • ✓All 18 integration packs
  • ✓BYO OAuth apps · BYO Webhook integrations
  • ✓Per-agent observability + cost tracking
  • ✓Logical tenant isolation on AWS
  • ✓OAuth tokens encrypted at rest (AES-256-GCM)
  • ✓Email support (24h response)
  • ✓Cancel anytime · monthly billing
Most popular

Self-Host

BYO infra or hardware
BYO LLM $4,900

We install on your existing infra. You bring the LLM (Ollama, OpenRouter, AWS Bedrock — all native providers).

Spark $11,900

NVIDIA DGX Spark — 128GB unified memory, GB10 Grace Blackwell. Runs Qwen3 235B, Llama 3.3 70B. Entry-level local AI.

Workstation $24,900

RTX 6000 Ada, 48GB VRAM, 64GB DDR5, 2TB NVMe, dual 10GbE. Balanced compute + memory for 8-12 agents.

Dual-GPU $44,900

2x RTX 6000 Ada, 96GB VRAM. 70B FP16 with headroom. Production-grade for mid-size teams.

Server $72,900

Single H100 80GB. 405B-class models at INT4. High-throughput inference for large deployments.

All Self-Host: + $1,200/yr support & updates after Y1

Install on infrastructure you own. Bring your own LLM endpoint, or take delivery of a turnkey GPU server.

Book a scoping call
  • ✓All 8 starter roles + 2 Team Packs
  • ✓All 18 integration packs
  • ✓Native providers: Ollama, OpenRouter, AWS Bedrock
  • ✓BYO OAuth apps (Google, MS, Salesforce, Slack, HubSpot)
  • ✓Webhook triggers · per-agent observability
  • ✓Source-available license — modify your copy
  • ✓Hardware tiers: models pre-installed, 3-yr warranty
  • ✓Air-gappable on Hardware tiers
Most complete

Enterprise

Regulated industries
Custom annual contract

Tailored to your scope

For HealthTech, FinTech, LegalTech, InsurTech, Government — any team that answers to auditors. Frontier hardware, native Bedrock with BAA, dedicated TAM.

Talk to us
  • ✓Everything in Hosted + Self-Host
  • ✓Frontier hardware: DGX H100/H200, $499,900+, multi-GPU clusters
  • ✓Native AWS Bedrock: Claude on Bedrock with HIPAA BAA
  • ✓FedRAMP path for Government workloads
  • ✓Dedicated TAM (Technical Account Manager)
  • ✓Custom SLAs: uptime, response-time, escalation
  • ✓Priority roadmap input — direct line to engineering
  • ✓Volume licensing: multi-seat, multi-agent discounts
  • ✓Compliance bundle: BAA paperwork, security questionnaires, vendor assessments
  • ✓Bring your own KMS keys
  • ✓Audit log export to your SIEM
  • ✓Custom roles + custom integrations
  • ✓Multi-node deployments (HA, regional failover)
  • ✓Quarterly business reviews

All tiers include all 8 roles, all integrations, and the full feature set · Custom integrations available on quote · Hosted token usage passed through at cost (no markup).

What you pay for, plainly

No hidden charges. No usage-based surprises on the platform. LLM tokens passed through at cost.

✓Included

  • All 8 starter roles + 2 Team Packs
  • All 18 integration packs (CRM, email, calendar, helpdesk, billing)
  • OAuth bring-your-own-app flows
  • Webhook triggers + per-agent observability
  • Email support (Hosted: 24h · Self-Host: Y1)
  • Quarterly software updates (Self-Host: with renewal)

×Not included

  • LLM inference cost (Hosted: passed through · Self-Host: BYO)
  • Cloud infrastructure cost on Self-Host BYO LLM (your AWS/GCP)
  • Custom roles or integrations beyond the 8/18 starters (quote)
  • Onsite training beyond Y1 (available on Enterprise)
  • HIPAA BAA / SOC 2 evidence package (Enterprise only)

Common questions

Which tier should I pick? ▾
Hosted ($1,999/mo) is the fastest path — sign up, OAuth your tools, you're running in an hour. Self-Host (one-time) is the right call if your security or compliance team requires data to stay on your infrastructure. Enterprise is for deployments that need BAA, SOC 2 evidence, or BYO KMS.
What does "LLM token usage at cost" mean on Hosted? ▾
You pay only what the upstream model provider charges (OpenAI, Anthropic, OpenRouter, etc.). We pass through wholesale pricing with no markup — your monthly Hosted bill is $1,999 plus whatever your agents actually spent on inference. Most ops teams see $200-800/mo of tokens depending on volume.
What hardware ships in the Self-Host tiers? ▾
Five options. Spark ($11,900): NVIDIA DGX Spark, 128GB unified memory, runs up to 200B-class models. Workstation ($24,900): RTX 6000 Ada, 48GB VRAM, 8-12 agents. Dual-GPU ($44,900): 2x RTX 6000, 96GB VRAM, 70B FP16. Server ($72,900): single H100 80GB, 405B-class. Enterprise ($499,900+): DGX H100/H200 multi-GPU clusters for frontier workloads.
What's the difference between Hosted and Self-Host with BYO LLM? ▾
Hosted: we run the orchestration platform on AWS, you OAuth in. Easiest. Self-Host: you run the orchestration platform on your AWS/GCP/your own server. More control, lower long-term cost, harder to set up — but we install it for you.
Can I switch tiers later? ▾
Yes — your config and integrations export cleanly. Hosted → Self-Host: we transfer the config, you take ownership. Self-Host → Hardware: we ship the box, restore your config, you continue. Hosted ↔ Self-Host transitions take ~2 weeks; Hardware delivery: 4-6 weeks.
Why no free tier? ▾
Two reasons. First, every install requires real human work to configure, even when the buyer is technical. Second, we want customers, not freeloaders — paid tiers signal you're serious about hiring AI agents that handle real work, and we can give you the install attention that deserves. If you need a true free starting point, the Hosted tier is monthly billing — cancel after 30 days for $1,999.
What happens if I stop paying the annual Self-Host renewal? ▾
Your installation keeps running. The $1,200/yr covers ongoing software updates and email support; without it you stop receiving updates and support tickets but your existing deployment continues to function indefinitely on the version you have.
Can I use Claude / Bedrock / my own model? ▾
Yes. Three native integration paths: (1) Any OpenAI-compatible endpoint (OpenRouter, Together, Fireworks) — set two env vars. (2) Local Ollama on your hardware — default install. (3) AWS Bedrock with HIPAA BAA — native provider, set PROVIDER=bedrock plus your AWS region. Your data never leaves your AWS account.
Can my agents talk to our internal tools? ▾
Yes. Paste your API docs (OpenAPI spec, Postman collection, or just markdown describing endpoints) and Inhouse generates the tool wrapper automatically — usually under a minute from spec to working agent integration. When the API drifts (renamed fields, new required headers, deprecated endpoints) the wrapper inspects the response and self-corrects. We've used this for HubSpot, Salesforce, Stripe, Zendesk, Linear, Asana, and several customers' homegrown billing/ticketing platforms. Auth methods supported: OAuth 2.0 + PKCE, bearer tokens, API keys, basic auth, mTLS.
Do you offer a pilot? ▾
Yes — book a fit call and we can scope a 30-day pilot before any commitment. Most customers prefer to evaluate against one specific workflow before sizing the full install.

Not sure which tier fits?

15-min call. We'll size it to your actual workload and recommend the cheapest tier that fits — or tell you when you should wait.

Book a 15-min demo →