Horizon
Horizon
Horizon
Timeline
2025 - 2026
Team
10+ Engineers, PM, Data
Outcome
Transformed market signals into an AI prospecting companion.
Transformed market signals into an AI prospecting companion.
Context
Context
Context
Surfe is a sales prospecting tool built for how modern revenue teams work. It helps reps find customers, enrich their contact details, and connect with them for that first outreach. But finding prospects is only the beginning. The real work starts after.
Surfe is a sales prospecting tool built for how modern revenue teams work. It helps reps find customers, enrich their contact details, and connect with them for that first outreach. But finding prospects is only the beginning. The real work starts after.
Surfe is a sales prospecting tool built for how modern revenue teams work. It helps reps find customers, enrich their contact details, and connect with them for that first outreach. But finding prospects is only the beginning. The real work starts after.
The problem
The problem
The problem
Every morning, reps started from scratch.
They'd spend the first part of their day explaining who they're looking for: criteria, company size, industry, role. Then they'd scour the data. Then they'd research each prospect individually. They'd jump to LinkedIn to verify freshness. They'd check the company website to confirm technologies.
They'd map the buying committee. Then they'd draft the outreach.
This entire cycle happened every single day. By the time they were ready to reach out, they'd already context-switched five times and lost focus.
Every morning, reps started from scratch.
They’d spend the first part of their day explaining who they’re looking for: criteria, company size, industry, role. Then they’d scour the data. Then they’d research each prospect individually. They’d jump to LinkedIn to verify freshness. They’d check the company website to confirm technologies.
They’d map the buying committee. Then they’d draft the outreach.
This entire cycle happened every single day. By the time they were ready to reach out, they’d already context-switched five times and lost focus.
Every morning, reps started from scratch.
They’d spend the first part of their day explaining who they’re looking for: criteria, company size, industry, role. Then they’d scour the data. Then they’d research each prospect individually. They’d jump to LinkedIn to verify freshness. They’d check the company website to confirm technologies.
They’d map the buying committee. Then they’d draft the outreach.
This entire cycle happened every single day. By the time they were ready to reach out, they’d already context-switched five times and lost focus.
What we learned
What we learned
What we learned
We started with a full discovery phase.
We ran a big round of user interviews to understand the actual workflow of reps when they were searching for leads, prospecting, and qualifying prospects.
We weren't looking for solutions, at this point. We were looking for pain points.
Where did the time actually go? Where did they get stuck?
We started with a full discovery phase.
We ran a big round of user interviews to understand the actual workflow of reps when they were searching for leads, prospecting, and qualifying prospects.
We weren’t looking for solutions, at this point. We were looking for pain points.
Where did the time actually go? Where did they get stuck?
We started with a full discovery phase.
We ran a big round of user interviews to understand the actual workflow of reps when they were searching for leads, prospecting, and qualifying prospects.
We weren’t looking for solutions, at this point. We were looking for pain points.
Where did the time actually go? Where did they get stuck?
01
01
Manual search, manual verification, constant context switching
Manual search, manual verification, constant context switching
Reps found leads in Surfe and other prospecting tools, then immediately jumped to LinkedIn, company websites, and their CRM to verify. Every prospect required multiple tab switches before they could even decide if it was worth pursuing.
Reps found leads in Surfe and other prospecting tools, then immediately jumped to LinkedIn, company websites, and their CRM to verify. Every prospect required multiple tab switches before they could even decide if it was worth pursuing.
02
Curation beats volume
Curation beats volume
Reps had plenty of leads. The problem was noise. They spent hours manually filtering and prioritising just to disqualify prospects that didn't fit.
Reps had plenty of leads. The problem was noise. They spent hours manually filtering and prioritising just to disqualify prospects that didn’t fit.
03
The action layer was invisible
The action layer was invisible
Once qualified, reps had no guidance on next steps. How to approach? What to say? Figuring out the first message was entirely on them.
Once qualified, reps had no guidance on next steps. How to approach? What to say? Figuring out the first message was entirely on them.
Designing the system architecture
Designing the system architecture
Designing the system architecture
The system needed three layers to solve these problems at once.
A data layer that ingests behavioural and buying signals from LinkedIn and the CRM. What reps search for. Who they add. Which deals they're working. This data becomes the foundation for everything else. It answers Learning 01 because it's verifiable. Real data from real sources.
An ML layer that learns patterns from that user history. Over time, it understands which prospects matter to each rep, which buying signals correlate with deals, what a good lead actually looks like for them specifically. This is how we solve Learning 02. Not by showing everything. By learning what matters and prioritising ruthlessly.
An AI layer that sits on top and adds context. Why this account? How to engage? What's the right message? This is the layer that brings it all together and solves Learning 03. Instead of reps jumping between tools to answer these questions, the AI surfaces the answers directly.
The system needed three layers to solve these problems at once.
A data layer that ingests behavioural and buying signals from LinkedIn and the CRM. What reps search for. Who they add. Which deals they’re working. This data becomes the foundation for everything else. It answers Learning 01 because it’s verifiable. Real data from real sources.
An ML layer that learns patterns from that user history. Over time, it understands which prospects matter to each rep, which buying signals correlate with deals, what a good lead actually looks like for them specifically. This is how we solve Learning 02. Not by showing everything. By learning what matters and prioritising ruthlessly.
An AI layer that sits on top and adds context. Why this account? How to engage? What’s the right message? This is the layer that brings it all together and solves Learning 03. Instead of reps jumping between tools to answer these questions, the AI surfaces the answers directly.
The system needed three layers to solve these problems at once.
A data layer that ingests behavioural and buying signals from LinkedIn and the CRM. What reps search for. Who they add. Which deals they’re working. This data becomes the foundation for everything else. It answers Learning 01 because it’s verifiable. Real data from real sources.
An ML layer that learns patterns from that user history. Over time, it understands which prospects matter to each rep, which buying signals correlate with deals, what a good lead actually looks like for them specifically. This is how we solve Learning 02. Not by showing everything. By learning what matters and prioritising ruthlessly.
An AI layer that sits on top and adds context. Why this account? How to engage? What’s the right message? This is the layer that brings it all together and solves Learning 03. Instead of reps jumping between tools to answer these questions, the AI surfaces the answers directly.
What we built
What we built
What we built
Horizon.
Horizon is an AI engine that learns a rep's ideal customer profile (ICP) through both manual input and behavioural analysis. It watches what they search for, who they add to their CRM, which deals they're working. The machine learning layer learns their pattern without them having to describe it every single time.
But Horizon isn't just a recommendation engine. It's designed to solve the three problems we found in discovery.
Horizon.
Horizon is an AI engine that learns a rep’s ideal customer profile (ICP) through both manual input and behavioural analysis. It watches what they search for, who they add to their CRM, which deals they’re working. The machine learning layer learns their pattern without them having to describe it every single time.
But Horizon isn’t just a recommendation engine. It’s designed to solve the three problems we found in discovery.
Horizon.
Horizon is an AI engine that learns a rep’s ideal customer profile (ICP) through both manual input and behavioural analysis. It watches what they search for, who they add to their CRM, which deals they’re working. The machine learning layer learns their pattern without them having to describe it every single time.
But Horizon isn’t just a recommendation engine. It’s designed to solve the three problems we found in discovery.
01
01
Replace the manual search with daily ready, prioritised leads
Replace the manual search with daily ready, prioritised leads
Instead of reps searching and then manually verifying across tools, Horizon delivers prospects directly to them. Reps start by defining their target customers once in their onboarding. Each prospect is matched against this, ranked by buying signal strength, and comes with all the data needed to qualify it from then onwards.
Instead of reps searching and then manually verifying across tools, Horizon delivers prospects directly to them. Reps start by defining their target customers once in their onboarding. Each prospect is matched against this, ranked by buying signal strength, and comes with all the data needed to qualify it from then onwards.
02
02
Curate ruthlessly
Curate ruthlessly
We don't show reps every possible lead. We surface the prioritized few. Each prospect comes with buying signals explaining why they matter right now. We rank recommendations based on buyer readiness and ICP fit. Reps spend their time on real prospects, not filtering noise.
We don’t show reps every possible lead. We surface the prioritized few. Each prospect comes with buying signals explaining why they matter right now. We rank recommendations based on buyer readiness and ICP fit. Reps spend their time on real prospects, not filtering noise.
03
03
Guide reps from prospect to first message
Guide reps from prospect to first message
We built a message generation feature that pulls from all three layers of the system. It takes the data context (what the prospect just did). It considers the ML layer (how this specific rep likes to sound). It adds the AI layer (the right approach for this person). Reps get guided, informed messaging. No more blank page.
We built a message generation feature that pulls from all three layers of the system. It takes the data context (what the prospect just did). It considers the ML layer (how this specific rep likes to sound). It adds the AI layer (the right approach for this person). Reps get guided, informed messaging. No more blank page.
The outcome
The outcome
The outcome
20% increase in app adoption, and over $1M in partnership deals closed. But the behaviour change mattered more.
Reps started their day in Horizon. Instead of spending two hours hunting for prospects, verifying across tools, and crafting messages from scratch, they spent thirty minutes reviewing curated recommendations and sending informed outreach. They stayed focused. The verification was built in. The prioritisation was done. The action was guided.
20% increase in app adoption, and over $1M in partnership deals closed. But the behaviour change mattered more.
Reps started their day in Horizon. Instead of spending two hours hunting for prospects, verifying across tools, and crafting messages from scratch, they spent thirty minutes reviewing curated recommendations and sending informed outreach. They stayed focused. The verification was built in. The prioritisation was done. The action was guided.
20% increase in app adoption, and over $1M in partnership deals closed. But the behaviour change mattered more.
Reps started their day in Horizon. Instead of spending two hours hunting for prospects, verifying across tools, and crafting messages from scratch, they spent thirty minutes reviewing curated recommendations and sending informed outreach. They stayed focused. The verification was built in. The prioritisation was done. The action was guided.
Up next
Let’s have a conversation.
hello@nasrien.com
© 2026 Designed by Nasrien
•
Made with Figma and Framer.
Let’s have a conversation.
hello@nasrien.com
© 2026 Designed by Nasrien
•
Made with Figma and Framer.
Let’s have a conversation.
hello@nasrien.com
© 2026 Designed by Nasrien
•
Made with Figma and Framer.