gtm systems · outbound · ai agents
systems that find the right technical buyers and turn pull into pipeline.
I’m Dima Durah. I build go-to-market systems to reach ML engineers, AI researchers, and engineering leaders at frontier labs and enterprises, then turn strong product pull into scalable outbound and growth.
I know the buyer: teams building or improving AI agents. I translate technical product value into a sharp commercial message, and I build the machinery myself: targeting, research, outbound, CRM, experiments, automation.
$82m+
Raised by companies and rounds I supported
8+
Products taken from 0 to 1
3
Exits and acquisitions supported
100k+
People brought into communities and programs
15+
Countries reached through launches and partnerships
introduction
strong product pull, turned into a repeatable growth system.
Most AI companies with real traction are not short on demand. They are short on distribution machinery.
Inbound arrives unqualified, outbound sounds like everyone else, and nobody can name the twenty accounts that matter this quarter. I fix that end to end: define the ICP and the buying committee, research the accounts, write messaging technical people respect, build the sequences and CRM structure behind it, and run experiments until the pipeline becomes predictable.
selected work
- 01
Living Assets
Making brand data legible to frontier models
Built product and growth strategy for an AI discovery platform that structures brand and public data so frontier models and search agents can find, understand, and cite it.
- 02
Veil by claai
Consent as an input to model training
Co-created and launched a machine-learning tool that lets creators opt their images out of unauthorised scraping and model training.
- 03
OwnID
Identity as trust infrastructure for AI-native workflows
Shaped the messaging, go-to-market architecture, and sales outreach strategy for a passwordless identity company whose credential signals underpin automated and agentic workflows.
- 04
spectral finance
Productising a machine-learning credit model
Advised the early positioning and product narrative for a protocol training machine-learning models to score creditworthiness from on-chain behaviour, without identity.
- 05
OSS GTM Playbook
An open-source GTM system for ML and developer-tool teams
Created a practical, public playbook for teams building open-source ML, data, and developer products.
- 06
sap startup ecosystem
Ecosystem GTM for enterprise AI and data startups
Designed founder, partner, and ecosystem programmes that connected SAP’s startup work to the people and conversations shaping enterprise AI, data, and emerging technology.
What I do well
I understand the buyer
Teams building or improving AI agents. I know how ML engineers, AI researchers, and engineering leaders at frontier labs and enterprises evaluate tools, who actually signs, and what makes them ignore you.
I translate technical value into a commercial message
I turn evals, latency numbers, and architecture details into a sharp point of view a buyer can repeat internally, and into copy that earns a reply.
I build the GTM system myself
Targeting and ICP research, account and persona lists, outbound sequences, CRM structure, experiment design, and automation. Not a deck: a working pipeline machine.
I’m credible with technical users
I can hold a real conversation about agents, evals, and infrastructure without pretending to be an ML researcher. Technical buyers can tell the difference immediately.
how I work
strategy that looks good in a deck and dies in a handoff is not the job.
I work closely with founders, product leaders, engineers, customers, and commercial teams. I ask enough questions to find the real constraint, make a clear call, build the first version quickly, and use the market to improve it.
Clarity over theatre.
Make the product easy to understand and hard to dismiss.
Momentum over ceremony.
Ship the useful version, then learn in public and in market.
Taste with evidence.
Good judgment matters. So does listening to what users actually do.
Systems over one-offs.
A launch should create the conditions for the next launch to work better.
current
I run Record Scratch, an independent studio where I build products and experiments to stay close to the work.
I am focused on go-to-market for companies selling to teams building AI agents: targeting, technical messaging, outbound systems, and growth experiments.
I also write about technology, culture, AI, and the systems that shape how we live with them, and I keep a public library of what I am reading.
