Confession time. MCP was one of the last things we shipped before this launch, almost a last minute decision. And honestly, I now think it s the most valuable part of k dlo.
Here s why. Ever since we started showing k dlo to people, we kept getting the same question. Why would I use this if I already have Claude Cowork / ChatGPT / my own agent? Fair question. We tried to answer it by explaining the architecture, the shared memory, how all the agents work from the same context. People nodded, but it didn t really land. Nobody gets excited about architecture diagrams
With MCP, they don t have to. You just plug k dlo into the tool you already use and feel the difference right away.
People already sell with their own AI every day. What they re missing is the context. k dlo keeps everything about your leads in one place, so it doesn t matter if you work in Claude, ChatGPT or a coding agent. That context goes with you, and the selling you already do just gets bigger. On top of that, k dlo gives you what a chat window can t, a real place to list your leads, sort them into campaigns and work on docs and everything else a deal needs.
The sales offers and deliverables can change frequently. Does the Knowledge Hub handle changes and reflect the latest version when generating proposals?
@nuseir_yassin1 KH organizes data into projects and works as agentic RAG, so agents have several tools for fetching data from it, including metadata. Agents in kōdlo can see which artifacts are in KH (not just semantically relevant snippets), so they can pick the latest version when asked. I confirmed this with Claude Code too. You can also pin specific projects so only those are visible to the agent, but that toggle is UI-only for now.
On top of this you can create different campaigns for different offers , search prospects for exactly that campaign's ICP and have specific instructions like outreach strategies or smth your agent should know.
keeping all sales tasks and client info in one place seems useful how does kodlo find the right prospects?
@shivam_kushwaha16 kōdlo builds your ICP first, then finds matching prospects using open sources and data from our data providers.
Does kōdlo measure reply rates and conversion rates per each message/offer to help the agent understand what is effective for my ICP?
Btw, Congratulations @andriana_oprysko , @ivan_telnov and team kōdlo 🎊✌️
@aymi_malik Thank you Muhammad! 🙌
kōdlo doesn’t integrate with LinkedIn directly, so it doesn’t track replies there on its own. But if you connect kōdlo through MCP to Claude, for example with the Claude browser extension, Claude can work with all the data kōdlo collects. It can track reply rates and other results for each message and offer, and bring that back into kōdlo. That way kōdlo learns what actually worked for your ICP and what didn’t, and every next round gets better.
How does @kōdlo actually know the best way to warm up a specific prospect, rather than just generating personalized outreach based on their profile?
@phofsahil great question! Personalized outreach only looks at who the person is. A warm-up plan also looks at the timing and the path to them.
kōdlo combines three things. First, fresh research on the prospect, like what they’ve posted recently, company news and where they show up online. Second, what it knows about you from your Knowledge Hub, so your offer, your case studies and how you sell. Third, everything that has already happened with that person, like past messages, notes and replies.
From that it suggests how to approach them step by step. It might be to engage with a recent post first, reference a specific piece of news, or wait for a better moment. You review the plan and decide what to do, and kōdlo learns from what works for you over time.
Quick question, if I mark a prospect as not a good fit, will kōdlo remember that the next time it searches for leads?
@nrupen_mandava yes it will! In fact we recommend doing it, because that’s how kōdlo’s agents learn what works for you. It works even better if you add a quick note on why the prospect isn’t a good fit. kōdlo keeps that in mind and uses your past experience to suggest better prospects next time.
How does kōdlo personalise the outreach message for each prospect while avoiding generating similar-sounding messages?
@ranjan_kumar45 There are several mechanics for this because yes, this the generic sound is a real headache. First of all, our prompts and setting for decoding phase during LLM generation. Second - some prompt engineering. However there is mechanics which might contradict with this set up - its your tone of voice , however it depends what exactly you add there. You can have multiple types of tone of voice and change them depend on audience and you offer.
How often does Kōdlo refresh its data? There may be signals like funding or hiring that can change over time and keeping up with those is important.
@iamanantgupta Kōdlo works a bit differently. Instead of maintaining and updating a static database, it runs a prospect search whenever a user requests one, then saves the findings and contacts to its built-in CRM. Each search draws on the most recent data available online and from our data providers.