Nue.io Demo Shows AI Playbook Build in 2 Minutes at SaaStr AI Day

Nue.io Runs Live AI Demo at SaaStr AI Day

James McArthur, VP of Product Advocacy at Nue.io, conducted a live demo of an AI agent at SaaStr AI Day this week on conference Wi-Fi. The session showed the agent building a guided selling playbook from plain English instructions and validating it against real SKUs and tier limits, according to SaaStr.

The demo covered a typical B2B revenue setup where reps spend days on quotes and deal desk teams rebuild closed deals by hand. One customer had described the prior process as a spreadsheet and a prayer.

Playbook Construction and Validation

McArthur built the playbook in about two minutes by describing rules in plain English, including showing subscriptions up for renewal, checking active subscriptions first, locking the price book, setting discount caps by tier, and generating quote PDFs. The system wrote the instructions into a markdown file, looked up actual SKUs, confirmed sellable quantities, set minimums and maximums, and tied discount caps to the approvals engine.

The agent reported its own limitation unprompted after validation, stating it could check auto-renewal status but could not access usage and consumption data. It distinguished reachable support ticketing from unreachable data sources.

Constraint Handling and Approvals

When asked for 150 units of a tier capped at 75, the agent refused the request. A 35% discount request on a one-year deal triggered an approval requirement inside the flow rather than after the quote was sent. Approvals routed through Approvals Pro, acquired by Nue last year, with options for email or Slack and a full audit trail.

The demo also previewed every invoice across the full contract lifecycle at quote time. McArthur stated there is no world where an invoice should ever need correction, as such issues trace to upstream quoting or ordering errors.

Implementation Timeline and Finance Involvement

The unvarnished implementation number given was about 90 days on average, up to a year, gated on catalog and data quality, according to SaaStr. Finance team involvement was required from initial discovery on every build, including CPQ-only projects.

The agent caught its own error mid-flow by creating an opportunity with the wrong stage, editing it, and continuing. A separate Stripe payment processed but failed to allocate on refresh. The session emphasized keeping a human in the loop for finance processes.