Two founders, one Berlin living room
Needle was founded in late 2024 in the living room of Onur Eken's apartment in Berlin: two founders, a whiteboard, a couch, and a lot of conviction. That is how the company's own About page tells it, and the details of the two men behind it check out. Eken, now CTO, had led engineering teams at Everstores and helped scale Delivery Hero and Celonis — classic fast-growing European scale-ups. Jan Heimes, now CEO, had worked as a data engineer at Climeworks and at Notable Labs, a Y Combinator 2015 company, before leaving his job in Zurich and moving to Berlin in search of something bigger.
The two met in Berlin and the spark was immediate, per the About page. What brought them together was a shared frustration they had each experienced inside their industries: organizations sitting on valuable data that current AI systems cannot easily reach — trapped in legacy systems, fragmented structures, and siloed storage. They wanted to make every organization AI-enabled. On paper, it was a clean thesis. Within months, investors agreed.
$2.2M for the data problem
In March 2025, Needle announced a $2.2 million seed round (€2 million, per EU-Startups) led by 468 Capital and Presight Capital. Coverage across EU-Startups, Pulse2, CityBiz, and Startups Magazine described the same use of funds: expand the engineering team, strengthen the core technology, and accelerate go-to-market. The product at the time was an AI-powered search and agent development platform — a unified search layer that could scan all internal systems, from Gmail to Confluence and GitHub, letting companies connect tools like Slack and Dropbox, index their data, and retrieve answers, automate workflows, and build context-aware AI agents.
"Organisations are sitting on goldmines of data current AI systems simply cannot tap into," Heimes said in the funding announcement. The company's agentic retrieval-augmented generation technology was the technical core. But infrastructure for unlocking internal data, however real the pain, turned out to be one station on the journey rather than the destination. On a GTM Vault podcast episode hosted by Rick Koleta, the founders sat down to unpack "how they raised $2.2M to help companies become AI-enabled in one day" — and, tellingly, the episode's headline topics included "their pivot from RAG APIs to a full-stack execution layer" and the "intern, not overlord" mindset for integrating AI.
Two pivots to the real pain
Needle's own site puts the count plainly: "after two pivots and a seed round." The product's Product Hunt history reads like a pivot diary. In January 2025 it launched as "Needle — enable AI-search across all your data in seconds." By September 2025 it was "automate any process with AI agents." March 2026 brought "Needle 2.0 — vibe-automate workflows." July 2026: "the proactive GTM agent in Slack and Teams." Each version moved the product further from infrastructure and closer to outcomes — from searching data, to automating processes, to running go-to-market itself.
The decisive insight, according to the company's own telling, came from running Needle's own go-to-market. "Doing outbound for Needle, we hit the same wall as everyone else: hours lost on lists full of the wrong people, generic messages nobody wanted, and accounts put at risk by volume," the About page reads. "Working through our go-to-market with agents as sparring partners changed that. They helped us think sharper about who to reach and what to say, not just send more." The GTM Vault conversation reinforces the pattern: grassroots traction meeting top-down sales, building trust, shipping fast, and getting teams live in minutes. The founders stopped selling AI plumbing and started selling the thing every founder needs — pipeline.
The agent-native outbound engine
Today's Needle is what the site calls "the agent-native outbound engine." Instead of opening another sales dashboard, you run your entire outbound motion from the AI agent you already use. The product ships as an MCP server: connect it once inside Claude (web, Cowork, or Claude Code), ChatGPT (web, desktop, or Codex), or Grok, describe your ideal customer in a single prompt, and the agent finds the leads, enriches them, and launches the campaign. Channels are the ones where buyers actually answer — LinkedIn, Instagram, and email today, with WhatsApp and SMS marked as coming soon — and a single campaign can mix channels in one sequence.
Lead discovery works the way a strong sales operator works. Your agent builds lists from LinkedIn people and company search plus post-reaction audiences, from Instagram likes, comments, and followers, and from Places — real businesses as they appear on maps, filtered by category and city. The site's demo shows a single prompt producing 212 matching SaaS founders, with 41 already in the CRM automatically skipped. Messaging is drafted from the lead's profile, the signals that made them match your ICP, and your own past conversations — you approve the angle once per segment, and Needle applies it per lead. A campaign-funnel view tracks sent, accepted, replied, and booked over the last 30 days, with insights like: leads who engaged with your posts accept twice as many invites, and replies sent within 15 minutes book three times more calls.
Safety is the moat
What makes Needle credible in a category infamous for burning accounts is that safety is engineered in, not bolted on. The product enforces per-account daily limits, human-like cadence, and a warm-up schedule for new senders — and, per its FAQ, "will refuse to exceed them." Suppression lists automatically skip customers, open deals, and anyone a colleague already contacted. The company's stated beliefs read like a charter for the category: no spray and pray, no time wasted on the wrong leads, no burned accounts, and outreach people don't mind receiving.
The business model is deliberately simple: pay per sender, leads included. A free 7-day trial (one connected account, no card required) leads to Pro at $69 per connected account per month for individuals and smaller teams, an Agency tier at $34 for every account beyond the first ten, and custom Enterprise pricing. There is no credit meter and no per-lead charge — "a list is only ever as big as you can actually work through." The company, now a tight-knit team building out of Berlin (with co-founders plus chief of staff Maurice Dücker and founding engineer Onur Oruç), reports 400+ teams running their outbound on the product, with testimonials from GTM operators like Shreya Das, who said Needle automated "the hardest part of my prospecting," and Rick Koleta, who called it "a game-changer for GTM teams." From a living room in Berlin to two pivots to an engine for outbound — Needle's story is a reminder that the right product is sometimes the one you were already building for yourself.
The playbook
- Dogfood your own painful problem — Needle's About page is blunt about the origin: "doing outbound for Needle, we hit the same wall as everyone else — hours lost on lists full of the wrong people." Working through their own go-to-market with agents as sparring partners became the product spec.
- Pivot toward the pain that pays — The GTM Vault episode with both founders centers on "their pivot from RAG APIs to a full-stack execution layer." Each pivot moved closer to where the money and pain are: from search infrastructure to an engine that books calls.
- Make safety a feature, not fine print — Per-account daily limits, human-like cadence, warm-up schedules, and automatic suppression of customers and open deals are built in by default. In a category notorious for burning LinkedIn accounts, safety is the selling point.
- Meet operators where they already work — No new dashboard to adopt: Needle ships as an MCP server inside Claude, ChatGPT, or Grok. The product lives in the workflow operators already opened every morning.
- Run hybrid GTM on a 15-minute activation — The founders described a hybrid motion — grassroots traction plus top-down sales — and engineered for speed: connect one account, ask your agent for a list, and "most teams have a campaign running within 15 minutes."