GrowthStack Advisory / AI Startups
Seed to Series A AI companies
GTM and outbound for AI startups, built from zero.
Built the engine from zero for nuvista.ai, covering markets, ICPs, personas, messaging, tooling and outreach. Advisory work with Hunar.ai and BunkerTech.
AI companies usually have the opposite problem to everyone else. The product is further ahead than the go-to-market, the category is new enough that buyers are not searching for it yet, and the founders are technical enough to have deferred sales entirely. That means outbound has to be built from zero rather than optimised. GrowthStack did exactly that for nuvista.ai, a named client: markets, ICPs, personas, messaging, tooling, account segmentation, event prospecting, email and LinkedIn sequences, meeting generation, post-event follow up, qualification and reporting. It produced roughly $50,000 in revenue plus additional qualified opportunities. Shobhit Gupta has also advised Hunar.ai and BunkerTech on GTM strategy, outbound engine setup and SDR hiring, though no figures are published for those two and none are claimed here. The order matters more than the effort: account list first, then messaging, then infrastructure, then launch. Nothing is sent in the first two weeks.
Client work is labelled separately from prior in-house roles. Named clients published with permission.
Why AI companies need a different starting point
Most outbound advice assumes a category the buyer already recognises. AI companies frequently do not have one, which changes where the work starts.
- Nobody is searching for it yet. If the category is new, inbound will not carry you and outbound is not optional. The account list has to be built from a hypothesis about who has the problem, then corrected against replies.
- The buyer does not know they have the problem. Messaging cannot lead with the product. It has to lead with a symptom the buyer already recognises and measures.
- Adding AI to the stack is not the strategy. An AI tool bought before targeting, messaging and qualification are settled produces better researched messages to the wrong list. That argument is set out in full in AI is a layer, not another tool.
What has actually been done in this vertical
- nuvista.ai, an AI technology company and a named client. There was no outbound motion at the start. GrowthStack developed the markets, ICPs, personas, messaging, tooling, account segmentation, event prospecting, email and LinkedIn sequences, meeting generation, post-event follow up, qualification and reporting. Approximately $50,000 in revenue plus additional qualified opportunities.
- Hunar.ai and BunkerTech, advisory engagements. GTM strategy, outbound engine setup and SDR or AE hiring. No published figures for these two, so none are claimed here.
The nuvista.ai build followed the same order as the ninety day plan: ICP and account list first, then messaging, then infrastructure, then launch.
Events, where AI buyers still gather
For a category buyers are not yet searching for, events do work that cold sequences cannot: they put you in front of people who have self-identified as interested in the problem space, before they know your product exists.
The nuvista.ai engagement included event prospecting as a first-class channel rather than an afterthought: account segmentation before the event, meeting generation during it, and structured post-event follow up, which is the part most teams skip and where the pipeline actually lands. nuvista.ai co-hosted an in-person AWS workshop in Washington DC on modern analytics and AI.
Shobhit Gupta has run the same motion in-house, representing Locus at the Last Mile ASEAN Conference and Expo. Event prospecting is a compressed version of the same ninety day sequence: a named list, a message that names a symptom, and follow up that does not stop after the first attempt.
Who this is for
- Seed to Series A AI companies. Early enough that the GTM function is being built rather than repaired.
- Technical founders without a sales function. The product is credible and nobody has yet built the motion that sells it.
- Teams whose category is not settled. Buyers do not yet search for what you do, which makes outbound and events carry more weight than inbound.
- Teams that need the first engine, not a second one. This was built from zero for nuvista.ai, covering markets, ICPs, personas, messaging, tooling and outreach.
What a ninety day engagement produces
The same six pieces get built every time. What changes by market is the targeting, the proof points, and which channel carries the most weight.
- Market and ICP definition. Which segment to attack first when the category has no established buyer.
- Personas and messaging. Translating a technical capability into a problem the buyer already knows they have.
- Sequences, deliverability and events. Email, LinkedIn and calling, with authentication set up first, alongside the events where AI buyers still gather.
- Toolstack and CRM workflows. Stages, fields and automation that enforce the process rather than rely on memory.
- Live outreach and iteration. Sequences run, replies read across the campaign, and the messaging changed on what came back rather than on opinion.
- Handover. Playbooks, dashboards and a trained team. You own the engine afterwards.
The engagement ends with the handover, not with a renewal conversation. The sequences, workflows, dashboards and playbooks stay with your team, and your team is trained to run them.
Who this is not for
Three or four calls a month end with us saying this is not the right piece of work. Cheaper for everyone if that happens before the call.
- You have nothing in market yet. Outbound tests a hypothesis against real buyers. Before there is something to show them, it measures the pitch rather than the product.
- You want an AI SDR tool rather than a motion. Those automate sending, which was never the constraint. The constraints are list quality, a message a rep can defend, and a written definition of qualified.
Not sure which applies? Run the outbound readiness scorecard, which takes a couple of minutes and gives an answer without speaking to us.
Frequently asked questions
Our category does not exist yet. Can outbound even work?+
It is usually the only thing that can. If buyers are not searching for the category, inbound has nothing to capture. The nuvista.ai engagement started with no outbound motion at all: markets, ICPs and personas were built from a hypothesis about who had the problem, then corrected against what actually replied.
We are technical founders who have never run sales. Where does that start?+
With targeting and a written qualification standard, not with tooling. The most common mistake is buying a sequencer in week one and discovering in week six that nobody can name the two hundred accounts that matter. Nothing should be sent in the first two weeks.
Should we buy an AI SDR tool instead?+
Probably not yet, and that is an unusual thing for a GTM consultancy to say. Tools marketed as AI SDRs automate sending, which was never the constraint. The constraints are list quality, a message a rep can defend and a written definition of qualified. The full argument is in AI is a layer, not another tool.
Related reading
Tell us where the funnel leaks
Thirty minutes is enough to tell you honestly whether it is a targeting, execution or qualification problem, and what it would take to fix.