The model your revenue team was waiting for.

A 120B-parameter mixture-of-experts post-trained on real outbound, real replies, and our live B2B knowledge graph — paired with live retrieval at inference. Nico writes outreach that gets replies, reasons about accounts with citations, and runs agentic revenue workflows end-to-end.

Request API access Book a demo

Nico 2.5 · account brief

> Brief me on Ramp. Who do I call this week and why?

Account: Ramp (ramp.com, Series D, ~1,100 employees)
Signal: Posted 14 Senior Platform Eng roles in last 30 days (LinkedIn)
         + Announced Treasury 2.0 product Mar 11 (TechCrunch)

Who to call:
  1. Diego Zaks — VP Platform Engineering. Owns the hiring surge.
     Angle: "scaling the platform team" — we 3x'd $COMPARABLE's
     interview throughput in 6 weeks.
  2. Eric Glyman — CEO. Treasury 2.0 launch narrative.
     Angle: GTM motion for financial-services enterprise accounts.

Open with: "Saw the Treasury 2.0 launch — curious how you're
sequencing the mid-market rollout vs. enterprise."

Citations: 10-K §1A, LinkedIn (2026-04-02), TechCrunch (2026-03-11)

120B

parameters — mixture-of-experts, single-GPU inference

context window — a full deal history per prompt

Reply-labeled

outbound corpus — real emails, real outcomes

Private

Salestools can deploy Nico privately for your organisation on demand

How it works

Three steps from first prompt to production.

Ingest

Nico connects to your CRM, enrichment stack, inbox, call recorder, and data warehouse. It reads accounts, contacts, signals, and rep activity in real time.

Reason

Extended thinking traces produce a grounded plan per account — what to say, to whom, when, on which channel, with which proof point. Every step is auditable.

Execute

Drafts outreach, logs CRM activity, books meetings, and hands off to reps where trust thresholds require human review. Humans stay in the loop by design.

Built for every seat

However you go to market, the agent adapts.

AE / Full-cycle seller

Morning briefing on your top 20 accounts — what changed, who to call, what to open with. Nico drafts the emails; you approve and send.

SDR / BDR team

Triple prospecting throughput without hiring. Nico sources lookalike contacts, researches each one, personalizes every touch, and self-routes replies.

RevOps

Replace brittle rule-based routing and scoring with a reasoning model that cites its sources. Ship playbook changes as prompts, not Zapier graphs.

Founders / Hands-on GTM

Run an AI-first sales motion before you hire your first rep. Nico handles research, writing, and follow-up; you close.

Plugs into your stack

Native integrations across the tools your team already uses.

OpenAI-compatible APISalestools GTMGrain agent fleetSalesforceHubSpotOutreachSalesloftSnowflake CortexAWS Bedrock (private)Azure OpenAI Service (private)Private VPC deployment

Frequently asked

How does Nico 2.5 compare to GPT-4 or Claude on sales tasks?

On head-to-head GTM benchmarks — account brief quality, first-touch reply rate, stakeholder identification, multi-step pipeline workflows — Nico 2.5 outperforms GPT-4o and Claude Sonnet at a fraction of their inference cost. Methodology and results are in the model card. The short version: a generalist model, no matter how big, doesn't know your buyer the way a model post-trained on a reply-labeled outbound corpus does.

What does Nico 2.5 cost?

Hosted API pricing is roughly an order of magnitude lower per million tokens than frontier dense models — because the mixture-of-experts architecture only activates a fraction of parameters per token. Enterprise deployments are flat-fee per environment with unlimited inference. Talk to sales for a quote; most teams pay back the contract inside a quarter on saved seat costs.

Can I deploy Nico 2.5 in my VPC?

Yes. Enterprise customers run Nico 2.5 inside their own AWS, Azure, or on-prem environment — prospect data never leaves your network. We ship a reference deployment with Bedrock- and Azure OpenAI-compatible endpoints. Hosted API is also available for teams that want the fastest path to production.

Is my data used to train future versions?

No. Customer prompts, completions, and CRM data are never used to train Nico. Training data comes from our own outbound corpus and opt-in, de-identified Salestools customer trajectories. Enterprise contracts include a Data Protection Addendum and zero-retention serving by default.

Is Nico 2.5 the same as Salestools GTM?

Nico 2.5 is the model. Salestools GTM is the agent product that runs on top of it — the CLI, tool calls, workflow orchestration, and CRM integrations. You can use Nico directly via API, embed it inside your own product, or buy Salestools GTM for a batteries-included agentic experience.

How fast can we get started?

API access in under 24 hours. Private VPC deployments typically go live in 2–3 weeks with our solutions team. We run a paid 14-day pilot for enterprise evaluation — bring a real account list and we'll prove the lift on your data, not a demo dataset.

Stop renting a chat model. Buy a GTM brain.

Request API access, book a 30-minute demo on your accounts, or talk to enterprise about a private deployment.

Request API access Book a demo