TopSpace.AI — Case Study

Neural-network sales performance booster. Two productized SKUs — LeadGen for discovery, LeadTouch for qualification — across five communication channels. Built by topsweteam end to end; the company was sold.

Summary

TopSpace.AI was a productized AI sales suite topsweteam built and went to market with. It packaged two distinct SKUs on the same platform: TopSpace LeadGen searched 24/7 across Telegram, Facebook, Instagram, X, WhatsApp, websites and forums for prospects matching a customer's keyword and ICP definition, dropped them into a smart CRM, and pitched a 60% reduction in lead-gen cost; TopSpace LeadTouch was a chatbot that picked up where LeadGen left off — multi-channel, 24/7, multilingual qualification that turned cold leads into warm ones, populated CRM with name/contacts/case/preferences, and generated calendar plans for sales staff. A team of three fullstack engineers plus one sales lead built it on Node.js. The company was subsequently sold.

• Client: TopSpace.AI (topspace.ai)
• Industry: Sales tech / AI / SaaS
• Engagement: Full product build — internal product, not a client engagement
• Timeline: 2024 – 2025 build, acquired in 2026
• Team: 3 fullstack engineers + 1 sales
• Status: Acquired by a French AI-sales startup (2026)

Challenge

Building a productized AI sales tool in 2023–24 meant entering a crowded category against well-funded incumbents. The differentiator wasn't going to be "we have AI" — that was table stakes. It had to be (1) end-to-end coverage of the funnel, not just one slice; (2) breadth across channels customers actually live in (Telegram, WhatsApp, Instagram), not just Email and LinkedIn that the US market defaults to; and (3) a price point low enough that an SMB could try LeadGen for the cost of a coffee per month per channel. With a four-person company, the architectural decisions had to make all three of those things possible without a 30-engineer team.

Approach

We designed the product as two SKUs on one platform rather than a single big tool, so a customer could buy the cheap entry-level lead discovery (€3.5/month/channel) and upgrade to the expensive qualification chatbot (€1,700/month) only when they were ready. That price-laddering shaped the architecture: the channel-integration layer had to be shared between the two products, the CRM that came with LeadGen had to be the same CRM LeadTouch wrote into, and the system had to be cheap to operate on the LeadGen tier where ARPU was tiny.

• Node.js for the whole stack. Chosen for the team — three fullstack engineers shipping faster than a polyglot split would have allowed.
• Channel integrations as a first-class layer. WhatsApp, Telegram, Facebook, Instagram, Twitter/X — each integration was substantial work, and they had to be reusable across both SKUs.
• One smart CRM under both products. The customer's mental model is "I get leads, I qualify leads, I track leads" — the system had to mirror that, not show seams between the two SKUs.
• Three engineers + one sales — no PM. With a productized tool, the customer is the spec. The sales lead's customer conversations drove the roadmap directly; the engineers shipped what unblocked the next deal.

Solution

Two products on one platform.

TopSpace LeadGen — discovery

24/7 lead discovery across customer-supplied channels (Telegram groups, Facebook/Instagram pages, X, websites, forums). The customer provides keywords and ICP descriptions; the neural network learns from those and surfaces matching prospects. Every lead is filed in the smart CRM with the source link, the timestamp, and the matched content excerpt — so a salesperson can see why the lead was selected and not just that it was. Pricing: €3.5/month/channel on annual, €4.9/month/channel monthly. Reduces lead-gen cost by ~60% vs. manual.

TopSpace LeadTouch — qualification

24/7 multi-channel chatbot that warms cold leads. The customer uploads their product knowledge base and successful conversation examples; the chatbot trains on those and converses across all the channels the customer specifies. Output is a fully populated CRM record (name, contacts, case, psychotype, preferences) plus generated calendar tasks for staff (showings, rentals, consultations, depending on the vertical). Pricing: €1,700/month for 8,000 messages, +€100 per additional 8,000. Reduces qualification cost by ~70%.

Key features shipped

• Neural-network lead discovery across five channels plus websites and forums
• Smart CRM auto-populated with leads, source links, and matched content
• Multi-channel chatbot (Telegram, WhatsApp, Instagram, Facebook, X) for lead qualification
• Knowledge-base ingestion and case-based fine-tuning per customer
• Auto-generated staff calendar plans (showings, rentals, consultations)
• Integration into customers' existing CRMs as an alternative to using the built-in one
• Subscription pricing, annual and monthly, on both products

Outcome

Built across 2024–2025 and acquired in 2026 by a French AI-sales startup. That's the outcome topsweteam ships against — not a metrics dashboard, not a press release, the thing that actually counts when you build your own product.

• Built across two years (2024–2025), acquired in 2026
• Acquirer: a French AI-sales startup (terms not disclosed)
• Two productized SKUs live in market at exit
• Multi-channel coverage across the messaging surfaces SMBs actually use
• Sub-€5/month entry-tier price point on LeadGen
• Four-person company (3 engineers + 1 sales) shipped, sold, exited

Tech stack

Backend / application: Node.js AI: Neural-network discovery model + LLM-driven multi-channel chatbot Channels: WhatsApp, Telegram, Facebook, Instagram, Twitter/X Web: customer self-serve and CRM surfaces

What we learned

• Productized SaaS is a different sport from client services. The customer is the spec, the sales lead is the PM, and the price-point math drives the architecture. Every decision is constrained by ARPU.
• Two SKUs on one platform is the right shape for a tool customers ladder into. A single big tool would have priced out the entry-level customers; two unrelated tools would have lost the cross-sell.
• Three fullstack engineers can build and exit a real product when scope discipline is non-negotiable and the integration surface (channels, CRMs) is built once and reused. Headcount didn't make this work — focus did.
• Local channels matter. Building for Telegram and WhatsApp from day one — not a year after the EN/US launch — was the architectural call that opened up markets the incumbents weren't serving.