Nex bets deterministic code-written agents, not chat, win high-volume GTM work

Nex (nex.ai), a Y Combinator S26 startup, sells an AI 'GTM engineer': a user describes a goal and Nex builds and runs high-volume go-to-market workflows across more than 1,000 integrations — lead generation and qualification, enrichment, outbound email and socials, CRM cleanup, and re-engaging lost deals.

The founding bet came from scale experience. Nazz and Francisco spent years at HubSpot running its three largest customer platforms; Francisco was the fourth engineer on the team that built HubSpot's CRM against Salesforce. They watched general agents re-reason every step, exhaust a weekly token budget overnight, and remain unreliable at volume — while almost nobody could hire enough GTM engineers to do the work properly. Both are backed by HubSpot founder Dharmesh Shah.

Nex's mechanism is the wedge: because most GTM work is repetitive and deterministic, the product has AI write code for the predictable parts and run them fast without wasting tokens, saving judgment for genuinely hard cases. The team deliberately avoids the word agent — operators want predictable outcomes and dependable workflows, not conversation with an unpredictable bot. On one onboarding call, Nex audited thousands of HubSpot records, flagged missing data, archived owners and zombie deals, and started fixing them before the call ended.

On Product Hunt, the September 3, 2026 launch ended #1 Product of the Day with 327 upvotes. Commenters praised executable workflows over prompts, and the sharpest question was rollback: what happens if a bad rule corrupts 100,000 CRM rows? The founders answered with audit reports detailing what changed, failed and errored, plus sample-data sandboxing before production access.