Agentic AI platform
Run one agent harness across your team's busywork in a no-code workspace, embed the same agent in your own product through the API, and scale on committed contracts. Want hands-on help? Our engineers can implement alongside you.
Paid trial at your plan rate. Cancel before day 30, owe nothing.
One agent harness, four very different jobs.
Assist in the workspace, Plan with approvals, headless sessions behind your API, and sandbox jobs that build deliverables - the same harness runs them all.
Ask, and watch it work.
“Brief me on Ferndale Robotics before the 10am call.”
Assist streams the answer live - every tool call visible in the loop, every claim cited back to a workspace file.
Streams tokens as it thinks; the tool loop stays visible
4 hits: past emails, a 2025 proposal, meeting notes
proposals/ferndale-2025.md - full text into context
News and funding via the web-search connector
One brief, 11 citations, delivered before the meeting
Ask, and watch it work.
“Brief me on Ferndale Robotics before the 10am call.”
Assist streams the answer live - every tool call visible in the loop, every claim cited back to a workspace file.
One workspace that carries the work.Agents handle the busywork.Your knowledge answers back. Every model, one door.
Agents that run your workspace.
Define an agent once and it works every request - routing questions, searching your data, and calling tools without being asked twice.
Runs on every new conversation in the workspace
Does the answer live in workspace knowledge?
Your agent
Running workflow
Your documents, ready to answer.
Drop in files, sites, and notes. The agent chunks, embeds, and indexes them so every reply is grounded in your sources - with citations.
PDFs, docs, pages - 240 files added
Split into passages, mapped to vectors
Searchable by meaning, not keywords
Every reply grounded in your sources
Every model, one door.
Chat and build against OpenAI, Anthropic, Google, Mistral, and Groq from the same workspace. Switch models per app - your data and billing stay put.
# open a session and stream replies curl -N -X POST /v1/managed-agents/sessions \ -H "Authorization: Bearer sk_live_…" \ -d '{ "agent_id": "agt_91xk", "app_id": "your-app" }' # → server-sent events stream back event: message.delta data: {"text": "Here's what I found…"}
Know what your AI is doing.
Conversations, token spend, and usage by app - live. Metered billing means the invoice always matches the chart.
- Usage by app
- Token spend over time
- Conversation reports
- Metered invoices
- Export to CSV
Real teams. Measured recovery.
“We started with three workflows. By week four it was running content, student support, and teacher prep across our whole platform.”
Gabriel T.
Founder & Lead Teacher, Acceleo Matemática · EdTech · Canada
“The agent handles 85% of inquiry routing without a human touch. Our team now gets the nuanced cases - not the repetitive ones.”
UC Baby
Healthcare · Prenatal imaging
SundayPyjamas runs measured workflows for teams across 4 industries - from startups to clinics.
Customer storiesOne agent. Every workflow.
Most platforms hand you a different bot for every job. Ours is a single agent harness - like a new co-worker - that connects to plugins, skills, and MCP servers. Teach it once, and every workflow runs through the same agent with the same context, memory, and guardrails.
No new agent to brief per task. No context lost between jobs. The agent that drafts your newsletter is the same one that triages your tickets - it already knows your business.
50 workflows it runs for teams like yours
Models, vectors, and cloud - wired for production.
The same integration surface we use in live customer deployments: frontier labs, hyperscalers, and data planes your security team already knows - not a toy sandbox.
Ecosystem you can vet
200+
models routed
AI models
Claude, GPT, Gemini, Grok, Groq-hosted Llama & Qwen, Cohere - unified routing, fallbacks, and metering.
6
database options
Vector data plane
Qdrant today; Pinecone, Weaviate, Milvus, Chroma, and Vald on the roadmap - pick what matches your ops model.
5
storage targets
Object storage
S3 today with GCS, Azure Blob, R2, and B2 in rollout - keep data where procurement and security already approved it.
AWSThree doors in. One platform underneath.
Operators run it as a no-code workspace. Product teams embed it through a full REST API. Platforms build on committed contracts with SLAs and priority routing. Same engine, same agent harness - the door you enter changes, not the product.
01 - For operators
Automate the work. Keep the judgment.
A no-code workspace where one agent harness runs your repetitive tasks against your real business data - in conversational assist mode or fully autonomous task runs. We equip your agent with its first workflows during a one-month trial.
Agent workspace
Plugins · Skills · MCP
Newsletter draft
RUN-1042
Support triage
RUN-1038
CRM enrichment
RUN-1036
Your agent
Running workflow
02 - For product teams
Your product. Our engine.
When the agents prove out, embed the exact same capability inside your own product or systems through a complete REST API - real-time streaming assistants and autonomous workflows, multi-tenant by design.
API Keys
spj_ai_••••••••••••••••03 - For platforms
Committed usage. Predictable economics.
Committed usage for production AI - Agents API, Image API, and managed agent runs - with SLAs, 50+ models, and spend controls that keep AI under 0.5% of revenue.
Committed Usage Contract
Commitment tiers
SLA guarantees
Cost you control to the credit.
Every turn is metered against a prepaid wallet. Hard budget caps stop spend at your ceiling, prompt and tool-schema caching skips repeat work, and a task-aware router sends each job to the cheapest capable model.
- Prepaid credit wallet
- Budget caps & alerts
- Prompt + result caching
- Task-aware model router
73% used · hard stop at cap
Start small. The platform scales with you.
Annual billing saves 2 months. All plans include a 1-month concierge trial.
Starter
Prove ROI with your first agent.
$6,000 billed annually · 2 months free
- 5 connectors included
- Usage dashboard and prepaid credit wallet
- Monthly strategy call (30 min)
- 48-hour support response
Growth
Reliability and parallel work for a team.
$15,000 billed annually · 2 months free
- 10 connectors included
- Priority model routing with fallback chains
- Bi-weekly strategy calls (45 min)
- Dedicated Slack channel · 24-hour support
Team
Full-team automation across every department.
$40,000 billed annually · 2 months free
- Custom integrations built on request
- Weekly strategy calls + dedicated success manager
- 4-hour support SLA
The first month is a paid trial at your plan rate. Cancel before day 30 and owe nothing - no questions asked. See full pricing.
Ready when you sign up.
Your workspace is live on signup. Optional: book a 60-minute kickoff to map your first three workflows. We ground workflows in your data while you run the agent, and measure hours recovered on day 30.