Trigger
New Nexa Agents 2.0 — multi-agent handoffs
Your autonomous
Your autonomous
operations layer.
Nexa deploys AI agents that read your chats, inbox and CRM — then run the workflows your team used to do by hand. Observable, approvable, production-ready.
- Tasks automated / day
- 0
- Median step latency
- 0
- Human approvals saved
- 0
Trusted by operations teams at
- Orbitalis
- Kestrel
- Monograph
- Fieldnote
- Parallax
- Arcwell
How it works
From a message to a closed deal.
No one touched a keyboard.
Agents listen on any channel, reason with the model of your choice, run deterministic workflows for the parts that must be exact, and write the result back to your systems of record.
Agent · LLM
Intent & extraction
Workflow
Enrich & score
Action
CRM
Platform
Everything an agent needs
to be trusted with real work.
Multi-agent orchestration
Planner, researcher and executor agents hand work to each other with typed contracts, so a failure in one step never silently corrupts the next.
Human-in-the-loop
Set spend or risk thresholds. Above them, agents pause and ask in Slack or Telegram.
300+ integrations
Native connectors for chat, email, CRMs, billing, databases and any REST or webhook endpoint.
Traces for every step
Token-level traces, tool calls and diffs. Replay any run against a new prompt before you ship it.
Long-term memory
Per-customer vector memory that agents read before acting — and forget on request.
Bring any model. Keep your keys.
Route between hosted frontier models and self-hosted open weights per task, with automatic fallback and cost caps. Deploy in our cloud or your VPC.
- Region-pinned data residency
- SSO, SCIM & audit log
- PII redaction before the model
Developer experience
Ship an agent
in one file.
Define tools, guardrails and triggers in code. Run locally with full traces, then nexa deploy to production with versioning and instant rollback.
- Typed tools — schemas generated from your functions
- Guardrails — budgets, allowlists, approval rules
- Evals — regression suites on every deploy
Pricing
Pay for outcomes,
not seats.
Starter
For builders exploring their first agent.
$0/mo
Start free- 2 agents
- 1,000 runs / month
- Community integrations
- 7-day trace history
Most popular
Pro
For teams putting agents into production.
$49/mo
Start 14-day trial- Unlimited agents
- 50,000 runs / month
- Human-in-the-loop approvals
- Evals & replay
- 90-day trace history
Scale
For orgs with compliance and volume needs.
$199/mo
Talk to us- Everything in Pro
- 500,000 runs / month
- VPC deploy & SSO
- Dedicated support engineer
Billed monthly. Cancel anytime.
Loved by operators
The quiet part of the team
that never sleeps.
FAQ
Questions,
answered.
Can't find what you need? Our team replies in under two hours on weekdays.
Which models does Nexa support?
Any model with a chat-completions style API, plus self-hosted open-weight models via our runner. You can route different steps to different models and set per-run cost caps.
How do approvals work?
Define a rule — for example “refunds over $500” or “emails to new domains”. When an agent hits it, the run pauses and a reviewer gets a one-tap approve/reject card in Slack, Telegram or the Nexa console.
Is my data used for training?
No. Prompts, outputs and memory are never used to train models. Data is encrypted at rest and in transit, and Scale customers can pin storage to a region or their own VPC.
Can I self-host?
Yes. The runner ships as a single container. Use the hosted control plane, or run the full stack on your own infrastructure under the Scale plan.
What counts as a run?
One trigger to completion — however many steps and tool calls the agent takes in between. Retries of failed steps are free.
Give your team
its time back.
Deploy your first agent in under ten minutes. Free forever for small workloads.